专利摘要:
MULTI-LEVEL SIGNIFICANCE MAP SCAN. Encoding and decoding methods for video data are described in which multi-level significance maps are used in the encoding and decoding processes. The significant coefficient indicators that form the significance map are grouped into contiguous groups, and a significant coefficient group indicator means for each group whether that group does not contain non-zero significant coefficient indicators. A multi-level scan order can be used in which significant coefficient indicators are scanned group by group. The group scan order specifies the order in which the groups are processed, and the scan order specifies the order in which individual significant coefficient indicators in the group are processed. The bit stream may interleave the corresponding significant coefficient group indicators, if any.
公开号:BR102012029413B1
申请号:R102012029413-3
申请日:2012-11-19
公开日:2022-02-15
发明作者:Tianying JI;Dake He;Nguyen Nguyen
申请人:Velos Media International Limited;
IPC主号:
专利说明:

Field
[001] The present application relates generally to data compression and, in particular, to methods and devices for encoding and decoding video using multi-level significance maps. background
[002] Data compression occurs in several contexts. It is very commonly used in computer networking and communication to store, transmit and reproduce information efficiently. It finds specific application in encoding images, audio and video. Video presents a significant challenge for data compression due to the large amount of data required for each video frame and the speed at which encoding and decoding often needs to occur. The current state of the art for video encoding is the ITU-T H.264/AVC video encoding standard. Defines several different profiles for different applications, including Main profile, baseline profile and others. A next-generation video coding standard is currently under development through a joint MPEG-ITU initiative: high efficiency video coding (HEVC).
[003] There are several standards for encoding/decoding images and videos, including H.264, which use block-based encoding processes. In these processes, the image or frame is divided into blocks, typically 4x4 or 8x8, and the blocks are spectrally transformed into coefficients, quantized and entropy encoded. In many cases, the data being transformed is not the actual pixel data, but is residual data following a prediction operation. Predictions can be intraframe, ie block to block within the frame/picture, or interframe, ie between frames (also called motion prediction). It is expected that HEVC (may also be called H.265) will also have these characteristics.
[004] When spectrally transforming residual data, many of these standards prescribe the use of a discrete cosine transform (DCT) or some variant thereof. The resulting DCT coefficients are then quantized using a quantizer to produce quantized transform domain coefficients, or indices.
[005] The block or matrix of quantized transform domain coefficients (sometimes referred to as a “transform unit”) is then entropy encoded using a context-specific model. In H.264/AVC and current development work for HEVC, quantized transform coefficients are encoded by (a) encoding a last significant coefficient position indicating the location of the last non-zero coefficient in the block, (b) encoding a map of significance indicating the positions in the block (other than the position of the last significant coefficient) that contain non-zero coefficients, (c) encoding the magnitudes of the non-zero coefficients, and (d) encoding the signs of the non-zero coefficients. This encoding of the quantized transform coefficients often takes up 30-80% of the encoded data in the bitstream.
[006] Transform units are typically NxN. Common sizes include 4x4, 8x8, 16x16, and 32x32, although other sizes are possible. Entropy encoding of symbols in the significance map is based on a context model. In the case of a 4x4 luma or chroma block or transform unit (TU), a separate context is associated with each coefficient position in the TU. That is, the encoder and decoder track a total of 30 (excluding lower-right positions) separate contexts for 4x4 luma and chroma TUs. The 8x8 TUs are divided (conceptually for context association purposes) into 2x2 blocks such that a distinct context is associated with each 2x2 block in the 8x8 TU. Therefore, the encoder and decoder track a total of 16+16 = 32 contexts for the 8x8 luma and chroma TUs. This means that the encoder and decoder keep track of and consult 62 different contexts during the encoding and decoding of the significance map. When 16x16 TUs and 32x32 TUs are taken into account, the total number of distinct contexts involved is 88. Among the 26 additional contexts, 13 are for luma TUs and 13 are for chroma TUs. The assignment of the 13 contexts to the coefficient positions in a 16x16 or 32c32 TU is as follows. Let (r, c) indicate a position in the TU, where 0 <=r, c<=15 if the TU is sized 16x16, and 0<=r, c<=31 if the TU is sized 32x32. Then 3 distinct contexts are assigned to the three positions (0,0), (0,1),(1,0) in the upper left corner including the DC position (0, 0); 5 distinct contexts are assigned to positions in the region {(4, c): 2 <=r+c <5}; and the last 5 distinct contexts are assigned to positions in the region {(4, c) 2 <=r+c <5} and the last 5 distinct contexts are assigned to all remaining positions. Except for the first 3 contexts for (0, 0), (0, 10) and (1,0) the context derivation for a position in the region {(r, c): 2 <= r+c <5} depends on its lower right neighborhood. Let s(r, c) denote the significance indicator of a coefficient at position (r, c), that is, s(r, c) = 1 if the coefficient is not zero and s(r, c) = 1 of another mode. The context for position (r, c) is equal to min (s(r+1, c) +s(r, c+1) +s (r+2, c) +s (r, c+2) + s(r+1, c+1), 4), where min(a, b) returns the smallest value between a and b. the context of a position (r, c) in the remaining region {(r, c): r + c >= 5} is similarly derived.
[007] The contexts for 4x4 and 8x8 significance maps are determined by bit position. The contexts for 16x16 and 32x32 significance maps are mostly determined by the values of neighboring bits. Determining the context for the 16x16 and 32x32 significance maps is relatively computationally intensive, because in most cases the processor determines context by looking at the values of neighboring significant indicators, which involves expensive memory access operations. summary
[008] This application describes methods and encoders/decoders for encoding and decoding significance maps with context-adaptive encoding or decoding. The encoder and decoder use multi-level significance maps. In at least one case, multilevel maps are used with larger transform units, such as 16x16 and 32x32 TUs.
[009] In one aspect, the present application describes a method of reconstructing significant coefficient indicators for a transformed unit of an encoded data bit stream, the bit stream including significant coefficient group indicators, and wherein each indicator of significant coefficient group corresponds to a respective group of significant coefficient indicators. The method includes, for each of the respective groups of significant coefficient indicators in a group scan order, reconstructing the corresponding significant coefficient group indicator from that group and reconstructing the significant coefficient indicators in that group by decoding the significant coefficient indicators to from the bit stream in a group scan order if the corresponding significant coefficient group indicator is non-zero, and set the significant coefficient indicators to zero if the corresponding significant coefficient group indicator is zero.
[0010] In another aspect, the present application describes a method for encoding significant coefficient indicators for a transformed unit. The method includes, for each respective group of significant coefficient indicators in a group scan order, coding the corresponding significant coefficient group indicator of that group, unless a special case applies, and in which the coefficient group indicator significant is adjusted in error to indicate that that corresponding group contains non-zero significant coefficient indicators, and to encode the significant coefficient indicators in that group in a group scan order if the corresponding significant coefficient group indicator is non-zero.
[0011] In one aspect, the reconstruction of significant coefficient indicators is performed in a given order, such as a scan order. In another aspect, the reconstruction of the significant coefficient group indicators is performed in a given order, which may be the same determined order as used with the reconstruction of the significant coefficient indicators, such as the scan order.
[0012] In one aspect, a significant coefficient group indicator is set to zero if all significant coefficient indicators in the corresponding group are zero. In another aspect, a significant coefficient group indicator is non-zero if at least one significant coefficient indicator in the corresponding group is non-zero. In another aspect, a special case may result in a significant coefficient group indicator being non-zero even if all significant coefficient indicators in the corresponding group are zero, in which case the decoder will decode all significant coefficient indicators from zero to that corresponding group from the bit stream.
[0013] In a further aspect, this application describes encoders and decoders configured to implement such encoding and decoding methods.
[0014] In yet a further aspect, the present application describes non-transient computer-readable media that stores computer-executable program instructions that, when executed, configured a processor to perform the described methods of encoding and/or decoding.
[0015] Other aspects and features of the present application will be understood by those of ordinary skill in the art from an examination of the following description of examples in combination with the accompanying figures. Brief description of drawings
[0016] Reference will now be made, by way of example, to the accompanying drawings showing exemplary embodiments of the present application, and in which: Figure 1 shows, in block diagram form, an encoder for encoding video; Figure 2 shows , in block diagram form, a decoder for decoding video; Figure 3 shows an example transform unit containing quantized transform domain coefficients; Figure 4 shows an example significance map containing significant coefficient indicators for the unit Figure 3; Figure 5 shows an example L1 significance map containing significant coefficient group indicators for the significance map of Figure 4; Figure 6 shows, in flowchart form, an example method of encoding indicators of significant coefficient; Figure 7 shows, in flowchart form, an example method for reconstructing a significance map from encoded data; 8 shows, in flowchart form, an example process for decoding and building significant coefficient group indicators; Figure 9 shows, in flowchart form, an example rate distortion optimized quantization process for significance map encoding figure 10 shows the example LO significance map of figure 4 after application of L1 RDOQ; Figure 11 shows the L1 significance map corresponding to the L0 significance map of figure 10 after L1 RDOQ; 12 shows a simplified block diagram of an example embodiment of an encoder; Fig. 13 shows a simplified block diagram of an exemplary embodiment of a decoder; Fig. 14 shows a 16x16 transform unit with 4x4 coefficient groups and a diagonal forward scan order; Figure 15 shows the 16x16 transform unit with 4x4 groups of coefficients and a base multi-level diagonal forward scan order ad in a group; Figure 16 shows, in flowchart form, an example method for decoding a significance map; and Figure 17 shows, in flowchart form, an example process for RDOQ encoding a multilevel meaning map with a multilevel order of digitization.
[0017] Similar reference numerals may have been used in different figures to indicate similar components. Description of example modalities
[0018] In the description that follows, some example modalities are described with reference to the H.264 standard for video encoding and/or the development of HEVC standard. Those of ordinary skill in the art will understand that the present application is not limited to H.264/AVC or HEVC, but may be applicable to other video encoding/decoding standards, including possible future standards, multi-view encoding standards, scalable video encoding standards, and reconfigurable video encoding standards.
[0019] In the description that follows, when referring to video or images the terms frame, image, slice, tile and rectangular slice group can be used somewhat interchangeably. Those of skill in the art will recognize that, in the case of the H.264 standard, a frame may contain one or more slices. It will also be recognized that certain encoding/decoding operations are performed on a frame-by-frame basis, some are performed on a slice-by-slice basis, some image-by-image, some tile-by-tile, and some by rectangular slice group, depending on specific requirements or applicable image terminology or video coding standard. In any specific embodiment, the applicable video or image encoding standard may determine whether the operations described below are performed with respect to frames and/or slices and/or images and/or tiles and/or rectangular slice groups, as appropriate. . Accordingly, those of ordinary skill in the art will understand, in light of the present disclosure, that specific operations or processes described herein and specific references to frames, slices, images, tiles, rectangular slice groups are applicable to frames, slices, images, tiles. , rectangular slice groups, or some or all of those for a given modality. This also applies to transform units, encoding units, groups of encoding units, etc., as will become evident in light of the description below.
[0020] The present application describes example processes and devices for encoding and decoding significance maps. A significance map is a block, matrix, or group of indicators that map to, or correspond to, a defined unit of transform or unit of coefficients (e.g., multiple transform units, a portion of a transform unit, or a encoding unit). Each indicator indicates whether or not the corresponding position in the transform unit or the specified unit contains a non-zero coefficient. In existing standards, these indicators can be referred to as significant coefficient indicators. In existing standards, there is one indicator per coefficient and the indicator is a bit which is zero if the corresponding coefficient is zero and is set to one if the corresponding coefficient is non-zero. The term "significance map" as used herein is intended to refer to an array or ordered set of indicators of significant coefficient for a transform unit, as will be understood from the description below, or a defined unit of coefficients, which will be evident at from the context of the applications.
[0021] Although the examples described here refer to significance maps, the multi-level encoding and decoding processes can be applied to other syntax elements in video encoding, for example, coefficient levels, filter coefficients, and motion vectors (after binarization), which may have group structures. For example, a local group of coefficient levels could be all one with high probability. Similarly, a local group of motion vectors could be all zero in one direction (zero horizontal motion, for example), or a set of filter coefficients could be all zero in neighboring frequency bands.
[0022] It will also be understood, in light of the following description, that the multi-level coding and decoding structure could be applied in certain situations, and those situations can be determined from secondary information such as video content type (video natural or graphic as identified in sequence, image, or slice headers). For example, two levels can be used for natural video, and three levels can be used for graphics (which is typically much more sparse). Yet another possibility is to provide an indicator in one of the sequence, image, or slice headers to indicate whether the structure has one, two, or three levels, thereby allowing the encoder the flexibility to choose the most appropriate structure for the present content.
[0023] Reference is now made to Figure 1, which shows, in block diagram form, an encoder 10 for encoding video. Reference is also made to Fig. 2, which shows a block diagram of a decoder 50 for decoding video. It will be appreciated that the encoder 10 and decoder 50 described herein may be individually implemented in a general-purpose or application-specific computing device, containing one or more processing and memory elements. The operations performed by the encoder 10 or decoder 50, as the case may be, may be implemented by means of an application-specific integrated circuit, for example, or by means of stored program instructions executable by a general-purpose processor. The device may include additional software, including, for example, an operating system to control basic device functions. The range of devices and platforms on which the encoder 10 or decoder 50 can be implemented will be recognized by those of ordinary skill in the art in view of the following description.
[0024] Encoder 10 receives a video source 12 and produces an encoded bit stream 14. Decoder 50 receives encoded bit stream 14 and transmits a decoded video frame 16. Encoder 10 and decoder 50 can be configured to operate in compliance with various video compression standards. For example, encoder 10 and decoder 50 may be H.264/AVC compliant. In other embodiments, encoder 10 and decoder 50 may conform to other video compression standards, including evolutions of the H.264/AVC standard such as HEVC.
[0025] Encoder 10 includes a spatial predictor 21, an encoding mode selector 20, transform processor 22, quantizer 24 and entropy encoder 26. As will be recognized by those of ordinary skill in the art, the encoding mode selector 20 determines the appropriate encoding mode for the video source, e.g., whether the present frame/slice is type I, P, or B, and whether specific encoding units (e.g., macroblocks, encoding units, etc.) in the frame/slice are inter- or intra-coded. The transform processor 22 performs a transform on the spatial domain data. In particular, transform processor 22 applies a block-based transform to convert spatial domain data into spectral components. For example, in many embodiments a discrete cosine transform (DCT) is used. Other transforms such as a discrete sine transform or others may be used in some instances. The block-based transform is performed on a coding unit, macroblock or sub-block basis, depending on the size of the macroblocks or coding units. In the H.264 standard, for example, a typical 16x16 macroblock contains sixteen 4x4 transform blocks and the DCT process is performed in 4x4 blocks. In some cases, the transform blocks may be 8x8, meaning there are four transform blocks per macroblock. In still other cases, the transform blocks can be of other sizes. In some cases, a 16x16 macroblock may include a combination of non-overlapping 4x4 and 8x8 transform blocks.
[0026] Applying the block-based transform to a block of pixel data results in a set of transform domain coefficients. A “set” in this context is an ordered set in which the coefficients have coefficient positions. In some instances the set of transform domain coefficients can be considered as a “block” or matrix of coefficients. In describing the present invention the phrases "a set of transform domain coefficients" or "a block of transform domain coefficients" are used interchangeably and are intended to indicate an ordered set of transform domain coefficients.
[0027] The set of transform domain coefficients is quantized by quantizer 24. The quantized coefficients and associated information are then encoded by entropy encoder 26.
[0028] The block or matrix of quantized transform domain coefficients may be referred to here as a “transform unit”.
[0029] Intra-encoded (ie Type I) frames/slices are encoded without reference to other frames/slices. In other words, they do not employ temporal prediction. However, intra-coded frames are based on spatial prediction in the frame/slice, as illustrated in figure 1 by the spatial predictor 21. That is, when encoding a specific block the data in the block can be compared with the data of nearby pixels in the already encoded blocks. for that frame/slice. Using a prediction algorithm, block source data can be converted to residual data. The transform processor 22 then encodes the residual data. H264, for example, prescribes nine spatial prediction modes for 4x4 transform blocks. In some embodiments, each of the nine modes can be used to independently process a block, and then rate distortion optimization is used to select the best mode.
[0030] The H.264 standard also prescribes the use of motion prediction/compensation to take advantage of temporal prediction. Accordingly, encoder 10 has a feedback loop that includes a dequantizer 28, inverse transform processor 30, and unlocking processor 32. The unlocking processor 32 may include an unlocking processor and a filtering processor. These elements mirror the decoding process implemented by the decoder 50 to reproduce the frame/slice. A frame store 34 is used to store the reproduced frames. In this way, the motion prediction is based on what the reconstructed frames will be at the decoder 50 and not the original frames, which may differ from the reconstructed frames due to the lossy compression involved in encoding/decoding. A motion predictor 36 uses the frames/slices stored in the frame store 34 as source frames/slices for comparison with a current frame for purposes of identifying similar blocks. Therefore, for macroblocks or encoding units to which motion prediction is applied, the "source data" that transform processor 22 encodes is the residual data that exits the motion prediction process. For example, it may include information regarding the reference frame, a spatial displacement or “motion vector”, and residual pixel data representing the differences (if any) between the reference block and the current block. Information regarding frame of reference and/or motion vector may not be processed by transform processor 22 and/or quantizer 24, but instead may be provided to entropy encoder 26 to encode as part of the bit stream along with the quantized coefficients.
[0031] Those of ordinary skill in the art will recognize the details and possible variations for implementing video encoders.
[0032] The decoder 50 includes an entropy decoder 52, dequantizer 54, inverse transform processor 56, spatial compensator 57 and unlocking processor 60. The unlocking processor 60 may include filtering and unlocking processors. A frame buffer 58 provides reconstructed frames for use by a motion compensator 62 in applying motion compensation. Spatial buffer 57 represents the operation of recovering video data for a specific intra-encoded block from a previously decoded block.
[0033] Bitstream 14 is received and decoded by entropy decoder 52 to retrieve the quantized coefficients. Secondary information can also be retrieved during the entropy decoding process, some of which can be fed to the motion compensation loop for use in motion compensation, if applicable. For example, entropy decoder 52 can retrieve motion vectors and/or reference frame information for intercoded macroblocks.
[0034] The quantized coefficients are then dequantized by the dequantizer 54 to produce the transform domain coefficients, which are then subjected to an inverse transform by the inverse transform processor 56 to recreate the "video data". It will be recognized that in some cases, as with an intra-encoded macroblock or encoding unit, the recreated "video data" is the residual data for use in spatial compensation with respect to a previously decoded block in the frame. Spatial compensator 57 generates video data from residual data and pixel data from a previously decoded block. In other cases, such as inter-coded macroblocks or encoding units, the "video data" recreated from the inverse transform processor 56 is the residual data for use in motion compensation with respect to a reference block from a different frame. Both spatial and motion compensation may be referred to here as “prediction operations.”
[0035] Motion compensator 62 locates a reference block in frame buffer 58 specified for an inter-coded macroblock or specific encoding unit. It does this based on the reference frame and motion vector information specified for the inter-coded macroblock or encoding unit. It then provides the reference block pixel data for combination with the residual data to arrive at the reconstructed video data for that encoding unit/macroblock.
[0036] An unlocking/filtering process can then be applied to a reconstructed frame/slice, as indicated by unlocking process 60. After unlocking/filtering, the frame/slice is transmitted as the decoded video frame 16, e.g. for display on a display device. It will be understood that a video playback machine such as a computer, frequency converter, DVD or Blu-ray player, and/or portable mobile device, may store decoded frames in memory prior to display on an output device.
[0037] HEVC-compliant encoders and encoders are expected to have many of these same or similar characteristics. Significance mapping coding
[0038] As noted above, entropy encoding of a block or set of quantized transform domain coefficients includes encoding the significance map (eg, a set of significant coefficient indicators) for that block or set of domain coefficients quantized transform. The significance map is a binary mapping of the block indicating in which positions (other than the last position) non-zero coefficients appear. The block may have certain characteristics with which it is associated. For example, it can be an intra-coded slice or an inter-coded slice. It can be a luma block or a chroma block. The QP value for the slice may vary from slice to slice. All of these factors can have an impact on the best way in which to entropy encode the significance map.
[0039] The significance map is converted into a vector according to the scan order (which can be vertical, horizontal, diagonal, zigzag, or any other scan order determined by the applicable decoding standard). The scan is typically done in “reverse” order, that is, starting with the last significant coefficient and working back through the significant map in reverse direction until the indicator at [0,0] is reached. In the present description, the term "scan order" is intended to mean the order in which indicators, coefficients or groups, as the case may be, are processed and may include orders which are referred to colloquially as "reverse scan order".
[0040] Each significant coefficient indicator is then entropy encoded using the applicable context adaptive encoding scheme. For example, in many applications a context-adaptive binary arithmetic (CABAC) encoding scheme can be used. Other implementations may use other context-adaptive codecs with binarization. Examples include binary arithmetic encoding (BAC), variable-to-variable encoding (V2V), and variable-to-fixed-length encoding (V2F). with 4x4 and 8x8 maps, a context is assigned to each bit position. By encoding the bit (significant coefficient indicator) at that bit position, the assigned context and context history for that point determine the estimated probability of a least likely symbol (LPS) (or in some implementations a most likely symbol (MPS)). ).
[0041] In existing video encoders, context assignment is predetermined for the encoder and decoder. For example, with a 4x4 luma block, the current draft HEVC standard prescribes that each bit position in the 4x4 significance map has a unique context. Excluding the last position meaning 15 contexts are tracked for coding 4x4 luma significance maps. For each bit position, the context assigned to that position determines the estimated probability associated with an LPS at that position. The effective bit value is then encoded using that estimated probability. Finally, the context assigned to that position is updated based on the effective bit value. In the decoder, the encoded data is decoded using the same context model. A context for each bit position is tracked and used to determine the estimated probability to decode data to retrieve bits for that position.
[0042] With 16x16 and 32x32 significance maps, the context for a signifier is (mostly) based on neighboring significant coefficient indicator values. Among the 13 contexts used for 16x16 and 32x32 significance maps, there are certain contexts dedicated to the bit position at [0,0] and neighboring bit positions, but most significant coefficient indicators take one of five contexts depending on the cumulative values of significant coefficient indicators. In these instances, determining the correct context for a significant coefficient indicator depends on determining and summing the values of significant coefficient indicators at neighboring locations (typically five locations, but could be a higher or lower number in some cases). This involves multiple memory accesses, which can be expensive in memory bandwidth requirements. Also, in many instances the 16x16 and 32x32 significance maps contain a large number of zeros. Therefore, there is substantial cost involved in encoding and transmitting large maps that have few coefficient values.
[0043] In accordance with one aspect of the present application, the encoder and decoder use multi-level significance maps for certain transform units. In the examples described below, multi-level significance maps are used for transform units of size 16x16 and 32x32: however, it will be understood that they may be used for 8x8 or 64x64 or transform units of another size in some embodiments.
[0044] Significant coefficient indicators are grouped. Each significant coefficient indicator is included in one of the groups. For simplicity in many embodiments the groups are formed by (conceptually) dividing the transform unit structure into blocks. For example, a 16x16 map can be divided into 4x4 blocks each containing sixteen of the coefficient positions. A 32x32 map can be divided into 8x8 blocks each containing sixty-four of the coefficient positions. Significant coefficient indicators are thus grouped based on what they are comprised of in these defined blocks in the matrix structure. In another example, both 16x16 and 32x32 maps can be divided into 4x4 blocks each containing sixteen of the coefficient positions.
[0045] Figure 3 shows an example 16x16 transform unit, 100 (the quantized transform domain coefficient matrix). For indexing purposes, the bit position in the transform unit can be specified by [xC, yC] where xC = 0, 1, 2, ... 15 and yC = 0, 1, 2, ... 15. Using (for example, a diagonal scan order, it will be noted that the last significant coefficient in this example is at [12, 10] as indicated by the reference numeral 112.
[0046] Reference is now also made to figure 4, which shows an example significance map 102. The significance map 102 contains the significant coefficient indicators that are generated from the example transform unit are shown in figure 4. It will be observed that a significant coefficient indicator appears at every bit position in scan order from [0,0] upwards to but excluding the last significant coefficient in [12, 10]. The significant coefficient indicator at each bit position for which there is a non-zero coefficient in transform unit 100 is set to 1, while each significant coefficient indicator at each bit position for which there is a zero coefficient is set to zero .
[0047] The significance map 102, ie the set of significant coefficient indicators, can be grouped based on a uniform division of the transform unit structure into contiguous blocks in one embodiment. The size of the transform unit can determine the size of blocks. In the case of a 16x16 transform unit, the blocks may be 4x4 in some embodiments. The groupings are illustrated in figure 4 by the lines demarcating the 4x4 blocks. A larger transform unit, such as a 32x32 transform unit, can have its significant coefficient indicators grouped into 4x4 blocks, 8x8 blocks, or contiguous blocks of other sizes. In one embodiment, 4x4 coefficient groups are used for transform units of sizes 16x16, 4x16, 16x4, 8x32, 32x8, and 32x32.
[0048] Although the examples given here use groups defined as contiguous square blocks for simplicity, the present application is not limited to square groups. Groups can be formed as rectangular blocks in some embodiments. In still other embodiments, other formates may be used. For example, with a diagonal sweep order, it may be advantageous to use groups formed from diagonal slices of the transform unit, in which case some of the groups may be somewhat trapezoidal in shape. For example, rectangular groups can be used as horizontal or vertical scan orders. In one example, with an 8x8 transform unit, 2x8 coefficient groups can be used if a horizontal sweep is employed, and 8x2 coefficient groups can be used if a vertical sweep is employed. Other variations will be understood by those skilled in the art.
[0049] A higher level significance map corresponding to the cluster matrix can then be generated. The highest level significance map is an ordered set of significant coefficient group indicators. There is a significant coefficient group indicator for each group containing at least one significant coefficient indicator. The group containing the last significant coefficient does not need to be included in the higher-level significance map because it will already be known to contain at least one non-zero coefficient, that is, the last significant coefficient. The significance map can be referred to as the level 0 map, or L0. The highest level significance map (ie, containing the significant coefficient group indicators) may be referred to as the level 1, or L1, map.
[0050] Figure 5 illustrates the highest level significance map L1 104 corresponding to the example significance map 102 shown in figure 4. It will be noted that the L1 map 104 contains a significant coefficient group indicator for each group that contains at least one significant coefficient indicator. If any of the significant coefficient indicators in the group are non-zero, then the significant coefficient group indicator is set to one. Otherwise, it is set to zero.
[0051] The indexing of groups can be specified by [xGC, yGC] where in this example xCG = 0, 1, 2, 3 and yCG = 0, 1, 2, 3. The group containing the last significant coefficient is in [ 3, 2]. The group in [3, 3] does not contain any significant coefficient indicators, so it is not included in the L1 map.
[0052] Significant coefficient group indicators can be converted to vector form in a scan order in some embodiments. The sweep order can be equal to the specified sweep order for use with the transform unit generically. In one embodiment, the significant coefficient group indicator may use a predefined sweep order which may be different from the selected sweep order for the transform unit. In some cases, the L1 map may exclude certain groups such as the [0, 0] group or the last significant coefficient group, which will have an assumed indicator value, as will be further described below.
[0053] It will be recognized that the L1 map need not be derived directly from the L0 map, but instead can be derived by scanning the coefficients in the transform unit in the scan order.
[0054] It will also be recognized that additional higher level maps may be used in some modalities. For example, if the transform unit is a 64x64 transform unit, the L1 map can be based on dividing the transform unit into 256 4x4 groups. Thus, the L1 map would be a 16x16 map containing L1 group indicators. An additional L2 map can be generated by grouping the L1 indicators into an additional set of 4x4 blocks (each of which would correspond to a group of 16x16 transform unit coefficients). Additional levels of abstraction and/or granularity may be employed in other modalities.
[0055] Reference is now made to Figure 6, which shows, in flowchart form, an example process 200 for encoding significant coefficient indicators. Process 200 begins in operation 202 with the encoder determining significant coefficient indicators and significant coefficient group indicators. In one embodiment, the encoder scans the transform block in scan order to determine the last significant coefficient and set of significant coefficient indicators. Significant coefficient group indicators can be determined during the same scan (although some amount of value storage can be used in practical implementations as the scan order may involve crossing through multiple blocks; in some cases, the determination of the indicator of significant coefficient group is made when the encoder determines to have swept the last coefficient for that group, e.g. the output coefficient). In some implementations, the encoder may perform a second scan of the L0 significance map or the transform unit to determine significant coefficient group indicators.
[0056] In operation 204, for each significant coefficient group indicator, the encoder determines the context to use and then entropy encodes that significant coefficient group indicator based on the given context. Significant coefficient group indicators can be processed in a prescribed order. In some embodiments, the prescribed order is equal to the scan order for the transform unit. The number of contexts and their determination can be structured in any appropriate way. An example set of contexts and method for determining contexts for significant coefficient group indicators is described further below.
[0057] Having encoded the set of significant coefficient group indicators, the encoder then encodes the significant coefficient indicators. In operation 206, the encoder (working in scan order) determines the context of and encodes each significant coefficient indicator if that significant coefficient indicator falls in a group for which the significant coefficient group indicator is set to 1. the corresponding significant coefficient group indicator is set to zero, then any of the significant coefficient indicators in that group are not encoded, that is, they are skipped during the entropy encoding process.
[0058] Therefore, after process 200 the encoder has produced an encoded data bit stream that contains the encoded significant coefficient group indicators and the encoded significant coefficient indicators that are comprised in a group that has at least one indicator of significant non-zero coefficient. The bit stream does not include any significant coefficient indicators from any group that does not have at least one non-zero significant coefficient indicator.
[0059] In the decoder. Significant coefficient indicators need to be reconstructed from the encoded data of the bit stream. Reference is now made to Fig. 7, which shows, in flowchart form, an example process 300 for reconstructing significant coefficient indicators from an encoded data bit stream. The bit stream may be received over a network connection, i.e. streamed, or read from a computer readable medium such as memory (e.g. flash memory, etc.) or storage disk (e.g. DVD, BluRay™, CD-ROM, etc.) process 302 is applied in the process of rebuilding a drive transformed into a decoder. Not shown is the decoding of header information, both for the sequence and for each slice or image (depending on the syntax of the video encoding standard in use).
[0060] In operation 302, the position of the last significant coefficient is decoded from the bit stream. This information can be represented in any applicable syntax. Some standards provide that the last significant coefficient must be specified using matrix notation, eg local based on x and y in the transform unit; some standards predict that the last significant coefficient should be signaled using a vector of 0's with a 1 at the last significant coefficient position, where the vector is mapped to the transform unit by scan order. Any syntax suitable for specifying the last significant coefficient can be used in operation 302.
[0061] In operation 304, significant coefficient group indicators are decoded from the bit stream. Significant coefficient group indicators may have been entropy encoded using any applicable binarization scheme specified by the standard or specified in the header information. For example, context-adaptive binary arithmetic encoding may be used in some instances. Significant coefficient group indicators are decoded by determining the context for each indicator position (bit position in the highest level significance map - for example, the L1 significance map) and then decoding the indicator value from the bitstream and update the context based on the indicator value. The size of the significant coefficient group indicator set is known because the scan order is known and the last significant coefficient was identified in operation 302; thus, the size of the L1 significance map is determined. In the case of groups not evenly split, an appropriate flag of group sizes and positions can be provided in the syntax.
[0062] As noted above, each significant coefficient group indicator corresponds to a respective of the contiguous groups defined for the transform unit. One or more of the significant coefficient indicators is comprised in each of these groups having a significant coefficient group indicator. Therefore, each significant coefficient group indicator corresponds to a respective group of significant coefficient indicators.
[0063] After decoding the set of significant coefficient group indicators, then the remaining operations to reconstruct the significance map, ie the set of significant coefficient indicators, are performed in the prescribed scan order. Processing starts from the last significant coefficient (but excluding the last significant coefficient position, as it is already known to contain a non-zero coefficient). In operation 304, for each significant coefficient indicator the decoder determines whether its corresponding significant coefficient group indicator is zero. If the corresponding significant coefficient group indicator is non-zero, then a significant coefficient indicator is decoded from the bit stream as indicated by operation 306. That is, if the associated or corresponding significant coefficient group indicator indicates that the group can contain at least one non-zero coefficient, then the decoder decodes a significant coefficient indicator from the bit stream to the current position.
[0064] If the associated or corresponding significant coefficient group indicator is zero, that is, indicates that there are no non-zero coefficients in the group, then the decoder sets or reconstructs the current significant coefficient indicator to a zero, as indicated by operation 308. Does not decode the same from the bitstream.
[0065] In operation 310, the decoder determines whether it has reached the end of the scan order, ie the coefficient in the upper left corner of the transform unit, eg [0, 0]. If so, then process 300 ends; if not, then the decoder moves to the next position in the scan order in operation 312 and repeats operations 306 and 308 to reconstruct the significant coefficient indicator for the next position.
[0066] It will be recognized that in this mode, the scan order does not result in reconstructing all significant coefficient indicators of a group before moving on to the next group. Instead, the scan order (depending on scan order and group geometry) sweeps across the boundaries of the group in such a way that the decoder reconstructs some indicators from one group, some from an adjacent group, etc. way back to position [0, 0] in the scan order. A scanning process will be further described below that avoids this problem.
[0067] There may be special cases that are considered in encoding and decoding processes to save bits. For example, as noted above, the group containing the last significant coefficient will always have a significant coefficient group indicator that indicates a non-zero coefficient, so the significant coefficient group indicator need not be encoded and transmitted to the decoder. The encoder always encodes the significant coefficient indicators for that group, and the decoder is configured to always encode the significant coefficient indicators for that group.
[0068] Another special case that can be included in some modalities is to always encode and decode the first group. This group contains a DC coefficient at [0, 0]in the transform unit. The probability that this group does not contain non-zero coefficients is extremely low. Therefore, instead of transmitting a significant coefficient group indicator for the group [0, 0], the encoder can be configured to always encode the significant coefficient indicators of that group and the decoder can be configured to always decode the significant coefficient indicators of that group. significant coefficient of that group.
[0069] Yet another special case that can be implemented in some modalities is also probability-based. It has been observed that when the group to the right and the group below a specific group both contain non-zero coefficients, then the probability that the specific group contains a non-zero coefficient is very high. Therefore, in some embodiments, the encoder and decoder can assume that any group that has a right neighbor group and a lower neighbor group that both contain non-zero coefficients, then that group has a non-zero coefficient. Thus, with respect to a certain group, if the significant coefficient group indicator for the group on the right is set to 1, and if the significant coefficient group indicator for the group below is set to 1, then the encoder does not encode a significant coefficient group indicator for a certain group and always the encoder for the significant coefficient indicators for a certain group. The decoder will recognize that the right and lower neighbors have significant coefficient group indicators indicating non-zero coefficients, so it will automatically assume that a certain group has non-zero coefficients and decode the significant coefficient indicators.
[0070] Reference is now made to figure 8, which shows operation 304 of figure 7 with additional detail to reflect an exemplary embodiment of handling the special cases described above. Operation 304 includes an operation 304-1 in which the significant coefficient group indicator for the group containing the last significant coefficient is set to 1. The location of the last significant coefficient is decoded from the bit stream in a previous operation ( not shown).
[0071] The decoder then moves through the groups in scan order. As noted in operation 304-2, the decoder moves from/to the next group in the scan wheel from the group containing the last significant coefficient. For this group, the decoder evaluates whether the significant coefficient group indicator for the group on the right and the significant coefficient group indicator for the group below the current group are equal to 1. Initially the decoder will have no indicators on the right and below because it has just started, but later in the scan order (either horizontally, vertically or diagonally) the decoder may sometimes have reconstructed significant coefficient group indicators at those positions relative to the current group (for groups located at the lower edge of the transform unit , the deedifier may not have an indicator for a group below). If these two adjacent groups are set to 1, then the probability of the current group also being set to 1 is high enough that both the encoder and decoder assume it is set to one. Therefore, in operation 304-6, the decoder sets the significant coefficient group indicator to 1 if the special case condition is met. Otherwise, the decoder moves to operation 304-4. In another embodiment, this special case can be modified to be based on significant coefficient group indicators from other adjacent groups, or other groups.
[0072] In operation 340-4, the decoder decodes the significant coefficient group indicator for the current group from the bit stream. Decoding includes determining the context and then decoding according to the given context. Decoding can be based on binary arithmetic coding (BAC), or other binary coding/decoding processes.
[0073] In operation 304-5, the decoder determines if this is the next to last group in the scan order. If not, then decoding has not finished yet, so the decoder cycles back to operation 340-2 to advance to the next group in scan order. If it is the next to last group in the scan order, then the decoder moves to operation 304-7 where the decoder sets the significant coefficient group indicator to the last group, i.e. group [0, 0] to 1. This is based on the special case in which that specific group is always assumed by the encoder and decoder to have at least one non-zero coefficient, so the significant coefficient group indicator is always preset to 1 so that the significant coefficient indicators for that group are always encoded and decoded. After this operation, the decoder goes to operation 306 or 308 (figure 7).
[0074] It will be recognized that the example process illustrated above in Figures 7 and 8 decoding the L1 significance map (the significant coefficient group indicators) and decoding the L0 significance map (the significant coefficient indicators) as a two-stage process in which the L1 significance map is fully decoded and the L0 significance map is then decoded. In some embodiments this may be the case; however, in some other modalities, the decoding processes may be partially intertwined. That is, decoding of the L0 map can start before the L1 map is fully decoded. It will be recognized that in some embodiments the decoding of the L0 significance map may begin as soon as the first significant coefficient group indicator has been reconstructed.
[0075] In some embodiments, multi-level significance map encoding can be turned on and off depending on, for example, the image type. For example, multilevel significance map coding can be enabled for I and P images, but disabled for B images. Multi-level scan order
[0076] As outlined above, when coefficient groups are formed as illustrated in the examples above, i.e. in contiguous blocks, the scan order (vertical, horizontal or diagonal) will result in crossing the boundaries of the groups when scanning the coefficient indicators significant. This could create difficulties in encoding and decoding from a hardware implementation perspective because significant data storage may be required to keep track of significant coefficient indicators from partially decoded groups to implement a one-pass scan. Otherwise, it may be necessary to scan twice (or more): one scan order pass for L1 map indicators and one scan order pass for the L0 map. One option to implement a one-pass scanning process and avoid some of the computational complexity and memory issues is to use a multi-level or group-based scan order.
[0077] Referring now to Fig. 14, a 16x16 unit transform of example 600 is illustrated. In the context of significant map encoding, the example 16x16 transform unit 600 includes 16 coefficient groups, where each coefficient group is a 4x4 block of significant coefficient indicators. A diagonal scan order is illustrated in figure 14. The scan order starts with the significant coefficient indicator at [15, 15] and the diagonal scan is from upper right to lower left while traversing transform unit 600 from the right bottom [15, 15] to top left [0, 0]. As discussed above, the significance map encoding process uses a scan order starting from the last significant coefficient.
[0078] It will be observed that the scan order cuts across the coefficient group limits. For example, consider the significant coefficient indicators in [10, 12], [11, 11] and [12, 10] as indicated by the reference numerals 610, 612, and 614, respectively. In scan order, the encoder and decoder first find the significant coefficient indicator 614 at position [12, 10]. The scan order then finds the significant coefficient indicator 612 at position [11, 11] which is the first significant coefficient indicator in that coefficient group. Then, the scan order then crosses into another coefficient group when it reaches significant coefficient indicator 610 at position [10, 12].
[0079] On the encoder side, it will be recognized that this frequent crossing of coefficient group boundaries can result in significant storage during the scanning process to track the values of significant coefficient indicators across multiple coefficient groups, and that a determination with relative to the significant coefficient group indicator for that coefficient group may need to wait until the last significant coefficient indicator (top left) in the group has reached scan order. To process the entire transform unit 600 in a one-pass scan, the encoder may require some complex storage to avoid excessive memory access operations. Frequent crossing of coefficient group boundaries presents a specific complication for rate distortion optimized quantization (RDOQ), ie soft decision quantization (SDQ). Tracking the rate costs associated with encoding when making RDOQ/SDQ determinations becomes significantly more complex. On the decoder side, the decoder may also require storage to track the previously decoded significant coefficient group indicators for the various groups as the decoder traverses them in scan order reconstructing the significant coefficient indicators. This results in higher memory/storage requirements on the decoder.
[0080] Therefore, in one embodiment the encoding and decoding processes may employ a multi-level scan order. Reference is now made to Fig. 15, which shows the transform unit 600 of Fig. 14 with an illustrated multi-level diagonal scan order. In each coefficient group, a diagonal sweep order is applied at the group level rather than across the entire transform unit 600. The coefficient groups themselves are processed in a sweep order, which in this example implementation is also a scan order. diagonal sweep.
[0081] It will be recognized that diagonal is an option, and in other modes horizontal, vertical, zigzag or other scan orders can be applied, at the coefficient groups and/or at the group level to order the processing of the coefficient groups.
[0082] Using multi-level or group-based scan order, each group of significant coefficient indicators is encoded and decoded in order. That is, encoding/decoding of the next group of significant coefficient indicators only begins after encoding/decoding of the present group has been completed. For example, using a diagonal group scan order, the entire group of significant coefficient indicators containing significant coefficient indicator 614 at position [12, 10] is decoded before the decoder starts decoding the significant coefficient indicator group. which contains the significant coefficient indicator 610 at position [10, 12]. Similarly, these two groups are fully decoded before the decoder starts decoding the group containing the significant coefficient indicator 612 at position [11, 11]. This allows the encoder/decoder to more easily process multi-level significance maps in one pass since all significant coefficient indicators of a coefficient group are processed sequentially in scan order in the group.
[0083] Advantageously, the group-based or multi-level scan order further facilitates the interleaving of significant coefficient group indicators in the bit stream. As each coefficient group is processed in scan order, the encoder can write the significant coefficient group indicator to the bit stream and can then input the significant coefficient group indicators for that coefficient group if the significant coefficient group indicator it's not zero. At the decoder, the decoder decodes the significant coefficient group indicator from the bit stream and if not zero then decodes the significant coefficient indicators from the bit stream and if not zero then decodes the significant coefficient indicators for that group in group scan order. If the significant coefficient group indicator is zero, then it sets all significant coefficient indicators for that group to zero and reads decode the next significant coefficient indicator from the bitstream. Thus, it reconstructs the significance map group by group in group scan order.
[0084] It will be understood that the scan order used in the group does not necessarily need to match the group scan order used to progress from group to group. For example, in groups a diagonal scan order can be used while processing the groups is done in a horizontal group scan order.
[0085] It will also be understood that the multi-level scan order described here matches the coefficient cluster used for the multi-level significance map; however, in the most general case the group-based or multi-level scan order may be organized into blocks that do not necessarily correspond to groups of significant coefficient indicators. For example, with reference to figures 14 and 15, instead of applying a diagonal scan order to each 4x4 coefficient group, in another example implementation the cluster or block for scan order purpose can use 8x8 blocks. In other words, in this example, the diagonal scan order would be applied to four 8x8 blocks. Note that if there is a mismatch between the cluster used for the multi-level scan order and the coefficient groups used in the multi-level significance map, then the encoder/decoder will not necessarily process each cluster fully before moving to the cluster. Following.
[0086] Reference is now made to Fig. 16, which shows an example method 700 for decoding a multi-level significance map using a multi-level scan order. Method 700 is an example process for reconstructing significant coefficient indicators from an encoded data bit stream. The encoded data bit stream includes encoded significant coefficient group indicators. Each significant coefficient group indicator corresponds to a respective group of significant coefficient indicators. Each non-zero significant coefficient group indicator in the bit stream is followed by the significant coefficient indicators of its respective group in a scan order.
[0087] Method 700 begins with decoding the last significant coefficient position from the bit stream in operation 702. As mentioned earlier, the last significant coefficient position can be signaled in any of several ways. After the position of the last significant coefficient is known, then in operation 704 the decoder decodes the significant coefficient indicators from the bit stream to the significant coefficients in the coefficient group containing the last significant coefficient. Decoding in operation 704 is performed in a scan order (which can be diagonal, vertical, horizontal, etc.) on a coefficient group, starting with the significant coefficient position after the position of the last significant position and working back in towards the upper left coefficient in the scan order.
[0088] In operation 706, with reference to a group scan order, the current coefficient group (indexed as CG) is set to be the group after the group containing the last significant coefficient. The group scan order is the order in which the coefficient groups are decoded, starting with the group after the group containing the last significant coefficient and working back towards the upper left group in the transform unit (the group containing the coefficient DC at position [0, 0]) in group scan order. The group scan order can be vertical, horizontal, diagonal, etc.
[0089] In operation 708, the decoder reconstructs the significant coefficient group indicator to the current coefficient group (indexed as CG). This reconstruction includes decoding the significant coefficient group indicator from the bitstream, unless a special case applies. For example, an exemplary special case is when the significant coefficient group indicators for the group to the right and the group below the current coefficient group are both non-zero. In this situation, the significant coefficient group indicator can be assumed to be non-zero by default.
[0090] As indicated by operation 710, if the reconstructed significant coefficient group indicator is zero, then in operation 712 the significant coefficient indicators of the corresponding coefficient group are all set to zero. If the reconstructed significant coefficient group indicator is non-zero, then in operation 714 the significant coefficient indicators of the corresponding coefficient group are decoded from the bit stream in scan order in that group.
[0091] After all significant coefficient indicators of the current coefficient group are reconstructed in either operation 712 or 714, then in operation 716, the CG index is decreased to move to the next coefficient group in the group scan order. If in operation 718, the next coefficient group in the group scan order is CG > 0, that is, it is not the upper left group containing the DC coefficient at [0, 0], then the process returns to operation 708 for fully reconstruct the significant coefficient of the next coefficient group. However, if CG = 0, then method 700 proceeds to operation 720, where the significant coefficient indicators of the upper left coefficient group are decoded from the bit stream. These significant coefficient indicators are always encoded in the bitstream in this example embodiment, so the decoder assumes that the significant coefficient group indicator for that coefficient group is effectively always non-zero. Context Modeling
[0092] To improve encoding efficiency, the BAC engine (or other entropy decoding/encoding engine) uses contexts. The present application proposes to use four new contexts to code significant coefficient group indicators. Two contexts are for luma encoding/decoding and two are for chroma encoding/decoding.
[0093] Determining which of the two contexts applies to a given significant coefficient group indicator can proceed as follows. If the significant coefficient group indicator for the group adjacent to the right of the current group is zero, and the significant coefficient group indicator for the adjacent group below the current group is zero, then the context for encoding the coefficient group indicator significant of the current group is 0. Otherwise, the context is 1. If the indicators for those two adjacent groups are not available then the unavailable indicators are assumed = 0 for context determination purposes.
[0094] Note that if a different scan order direction (such as from upper left to lower right) is used, the context model can be changed to use the significant coefficient group indicator for the adjacent group to the left of the current group and the significant coefficient group indicator for the adjacent group above the current group to determine context.
[0095] The context determination process may also include special cases. For example, the top left group can always be assigned context 1.
[0096] There are other possible context models and methods to determine which context can be used. Some examples are given below.
[0097] To define notation, let L[i] indicate the significance indicator of coefficient group i at level L and let N indicate the number of coefficient groups at level L. in general, for a given L and coefficient group i, we use a function c(*) of i and every available L[j] to determine a context C_i for L[i]. the context is thus given by: C_i = c(i, L[0], L[1], ... L[N - 1]), where j! = i. note that to use L[j] to determine a context for L[i], L[j] itself must be available. Therefore, the selected scan order must ensure that any L[j] used in c(*) has been determined previously.
[0098] In a modality similar to the context determination mode described above, the context can be determined by: C_i = c(i, L0[0], L0[1], ... , L0[15])= sum {bj * L0[j]}where j= 0, 1, ..., N and j! = i, bj = 1 if coefficient group j is the right or lower neighbor of coefficient group i, and bj = 0 otherwise. This particular modality has 3 contexts (6 if 3 different contexts are used for chroma).
[0099] Another modality of c(*) is given by: C_i = c(i, L[0], L[1], ... , L[N - 1])= sum {bj * L[j] }where j = 0, 1, ..., N and j! = i, bj is non-zero if the coefficient group j is any neighboring coefficient group of i that has already been determined and bj = 0 otherwise. In this embodiment, the weighting coefficients bj may not necessarily be constant.
[00100] Another modality of c(*) ignores the significant coefficient group indicator of other coefficient groups in L and determines the context based solely on position i of the current coefficient group. This can be expressed as: C_i = c(i, L[0], L[1], ... , L[N - 1])= i
[00101] Other context models and processes for determining context can be used with multi-level significance maps.
[00102] Below is an example syntax for the two-context modality discussed above. In this example, consider inputs to be the current coefficient group scan position (xCG, yCG) and the previously decoded buckets of the significant_coeffgroup_flag syntax element. The output of this process is ctxIdxInc. In this example, the transform unit is assumed to have been divided into sixteen contiguous blocks to form the coefficient groups. For example, a 16x16 TU is divided into 4x4 blocks and a 32x32 TU is divided into 8x8 blocks. In another example, the transform units can be divided into 4x4 blocks or blocks of other sizes.
[00103] The ctxIdxInc variable depends on the current position (xCB, yCB), and previously decoded deposits of the significant_coeffgroup_flag syntax element. For the ctxdxInc derivation, the following applies.
[00104] If xCG is equal to 3, and yCG is equal to 3, ctxIdxInc is set equal to a default context value. In this example embodiment, the default value is ctxIdxInc = 44. In other embodiments, it can be another value, including 0.
[00105] If xCG is less than 3, and yCG is equal to 3, ctxIdxInc is set to ctxIdxInc = 44 + significant_subblock_flag [ xCG+ 1][ yCG]
[00106] If xCG is equal to 3, and yCG is less than 3, ctxIdxInc is set to ctxIdxInc = 44 + significant_subblock_flag [ xCG][ yCG +1]
[00107] If xSG is less than 3, and ySB is equal to 3, ctxIdxInc is set to ctxIdxInc = 44 + Max{significant_subblock_flag [ xCG+ 1][ yCG], significant_subblock_flag [xCG][yCG +1]}
[00108] The value 44 in the expressions above is an example of a default index value. Other values can be used in other modalities, including 0.Rate distortion-optimized quantization
[00109] Some encoding processes employ rate distortion-optimized quantization (RDOQ), or what is sometimes referred to as “soft quantization”. RDOQ is a process of determining optimal quantized transform domain coefficients based on a rate distortion optimization expression. Therefore, the quantized transform domain coefficients that result from RDOQ may or may not be the same as the quantized transform domain coefficients arrived at through the normal transform and quantization process. In some cases, the coefficient values may have been modified by the RDOQ process because the resulting distortion was determined to have been less expensive than the consequent savings in transmission cost.
[00110] The RDOQ process normally assesses four costs in determining the rate component. The four rate costs include the last position rate, the significance rate (L0 rate), the coefficient rate, and the coded block parameter rate (CBP). To implement multi-level significance maps, it may be advantageous to modify RDOQ to also include the highest level significance rate (eg L1 rate) in the RDOQ calculation.
[00111] In one embodiment the RDOQ process can be modified to perform a two-stage RDOQ with respect to significance maps. First, the RDOQ process is applied to determine the best coefficient and last position values, and thus, the indicators of significant coefficient L1. In a second stage, with the last position fixed, the RDOQ process can then be applied again against the L1 rate to determine if there is a rate distortion (RD) cost justification for zeroing any coefficients.
[00112] Figure 9 shows, in flowchart form, an example RDOQ process 400 for encoding multi-level significance maps. Process 400 uses RDOQ to obtain the optimal quantized transform coefficients and determine the position of the last significant coefficient, ie an L0 RDOQ. Process 400 then fixes the last position and adjusts the current RD cost to account for the effect of additional levels of significance maps. Then, it takes an avid approach to further optimize the transform coefficients.
[00113] Operation 402 reflects the use of RDOQ to obtain optimal quantized transform domain coefficients, which provides a position of the last significant coefficient. Operation 402 results in a certain RD cost based on the rates for transmitting the last position, significance map corresponding to the optimal coefficients, the coefficient values and CBP.
[00114] In operation 404, the last significant position is fixed. That is, the last group will contain a non-zero coefficient, that is, the last significant coefficient group indicator is fixed at 1. The encoder then eagerly determines whether cost savings are considered by zeroing coefficients in other groups. Process 400 may run in scan order in some embodiments, although it may be run in another order.
[00115] In operation 406, starting with the next to last group as the current group, the encoder determines whether the current group has a group indicator of significant coefficient = 1. If not, then the group contains only zeros already and the encoder jumps to the next group. If the significant coefficient group indicator = 1, then the encoder calculates an RD cost that would result if all coefficients in the current group were zero. In operation 408, the encoder evaluates whether the newly calculated RD cost is better than (eg, less than) the current RD cost. If so, then in operation 410 all coefficients in the current group are cleared, and the current RD cost is updated to reflect the change. In operation 412, the encoder evaluates whether it is done with L1 RDOQ, for example, if it has reached the group just before the [0, 0] group (the [0, 0] group is not zeroed if the encoder and decoder are set to assume that there is at least one non-zero coefficient in that group, as described in the special cases outlined above). If there are additional groups to evaluate, then process 400 continues at operation 414, where the encoder moves to the next group (using scan order in some embodiments).
[00116] The RDOQ process will now be illustrated as an example. References will again be made to the example given above with respect to figures 3, 4 and 5. Before L1 RDOQ, but after L0 RDOQ, the optimal quantized transform domain coefficients are shown in figure 3. The corresponding L0 significance map is shown in figure 4, and the L1 significance map is shown in figure 5.
[00117] The L1 RDOQ process can result, for example, in an optimized L0 significance map 500 shown in figure 10, and the associated or corresponding L1 significance map 502 shown in figure 11.
[00118] It will be observed that the significant coefficient indicators in the last significant group, that is, the group [3, 2] are unchanged. However, the group [3, 0], the group [0, 3], and the group [1, 3] were all zeroed. As a result, the corresponding significant coefficient group indicators for these three groups were changed to zeros as well, as indicated in figure 11. The result is that the encoder will not need to code these three groups. The distortion that results from zeroing out the few coefficients that were found in these groups is outweighed by the cost savings in reducing the number of encoded bits, as determined by evaluating RDOQ.
[00119] In one possible embodiment, the RDOQ process can be extended to determine the optimal coefficient group size for the current TU. In this embodiment, process 400 is repeated for multiple rounds, with each round assuming a different coefficient group size and with operation 410 modified so that the transform coefficients are not actually set to 0. Essentially, and each round, this modified RDOQ process calculates the RD cost for a specific coefficient group size. After all rounds have completed, RDOQ selects the coefficient group size that provides the minimum RD cost and finally sets any transform coefficients to 0 as required. The encoder encodes the optimal coefficient group size value into the bit stream so that it can be obtained and used by the decoder.
[00120] The coefficient group sizes that are tested can be based on the transform unit size. For example, a 32x32 transform unit can test 8x8, 4x4, and 2x2 group sizes. The groups to be tested can be selectable, and the encoder can indicate (eg in the sequence header) which group sizes will be tested for each transform unit size. Suppose, for example, that the encoder and decoder have agreed that for 16x16 TUs, the modified RDOQ will test for two different coefficient group sizes: 2x2 and 4x4 denoted by 1 and 0, respectively. If the modified RDOQ determines that 2x2 is optimal, the encoder encodes a Bin 1 in the bitstream before the significant coefficient group indicators. The decoder decodes this Bin before the significant coefficient group indicators and knows that the coefficient group size for the current TU is 2x2.
[00121] In another embodiment the RDOQ process takes advantage of multi-level scanning: L1/L0 RDOQ is executed in a group-by-group mode followed by determining the last position. Specifically, an exemplary RDOQ process for a 16x16 TU, 4x4 coefficient groups, and a multi-level sweep over 4x3 groups, is as follows.
[00122] Step 1: Set nCG = 15 (start from the last coefficient group).
[00123] Step 2: Make L0 RDOQ for each coefficient in the coefficient group at the nCG group position following the scan order in the group as specified by the multi-level scan.
[00124] Step 3: If the resulting coefficient group after step 2 has non-zero coefficients, nCG is greater than 0, and the right neighbor or the neighbor below has its significant coefficient group indicator set to zero, do L1 RDOQ for the coefficient group: 1) calculate the cost of RD by setting the L1 indicator to zero for the present coefficient group; 2) if the cost is less than the RD cost resulting from step 2, set all coefficients in the coefficient group to zero, and the group's significant coefficient group indicator present to 0.
[00125] Step 4: Decrease nCG by 1.
[00126] Step 5: Repeat steps 1-4 until nCG is equal to 0.
[00127] Step 6: Determine the last position minimizing the RD cost.
[00128] Figure 17 illustrates this example in flowchart form showing the example process 450 for RDOQ encoding a multi-level significance map with a multi-level scan order. Process 450 includes a first operation 452 of adjusting the number of nCG coefficient groups based on the group containing the last significant coefficient. In operation 454, a LO RDOQ process is performed on the coefficients in the current coefficient group; that is, rate distortion optimization is used to determine optimal coefficient values for each coefficient in the current group. In operation 456, if there are no non-zero coefficients, then process 450 skips to operation 464 to move to the next coefficient group in scan order and cycles back to operation 454 to perform LO RDOQ on that next coefficient group. Note that process 450 also jumps to operation 464 if the lower and right neighbor groups have significant coefficient indicators that are both non-zero.
[00129] In operation 456, if there are non-zero coefficients in the current group, then process 450 moves to operation 458 in which L1 RDOQ is performed with respect to the current group. That is, the RD cost is calculated if the L1 indicator (significant coefficient group indicator) was set to zero, thereby resulting in all coefficients being zero in the decoder. If rate savings versus distortion results in a lower RD cost, as evaluated in operation 460, then in operation 462 the coefficients are zeroed and the significant coefficient group indicator is set to zero.Example syntax
[00130] An example syntax for implementing multi-level significance maps is provided below. This example syntax is just one possible implementation.
[00131] Significant coefficient group flags can be indicated and set as: significant_coeffgroup_flag[ xCG ][ yCG ]This flag specifies, for the coefficient group (xCG, yCG) position in the current 16x16 or 32x32 transform block, if the corresponding in-place coefficient group (xCG, yCG) has non-zero coefficients as follows: if significant_coeffgroup_flag [xCG][yCG] is equal to 0, the number of non-zero coefficients in the in-place coefficient group (xCG, yCG) is set equal to 0; otherwise (significant_coeffgroup_flag [xCG][yCG] is equal to 1), the number of non-zero coefficients in the in-place coefficient group (xCG, yCG) is non-zero except for the special cases defined below .
[00132] Special cases are defined as follows:1. The significant_coeffgroup_flag[0][0] at the first coefficient group position (0, 0) in scan order is inferred to be equal to 1.2. The significant_coeffgroup_flag [xCG][yCG] at the position of coefficient group (xCG, yCG) in the scan order is inferred to be equal to 1 if Significant_coeffgroup_flag[xCG][uCG+1] = 1 and Significant_coeffgroup)flag[xCG+1 ][yCG] = 1. When significant_coeffgroup_flag [xCG][yCG] is not present, it is inferred to be equal to 0.
[00133] In some embodiments, thesignificant_coeffgroup_flag[xCG][yCG] does not apply in 4x4 and 8x8 transform blocks.
[00134] The following pseudo-code illustrates an example implementation of multi-level significance maps in the decoding process for reconstruction (residual) quantized transform domain coefficients.
[00135] It will be noted that the first portion of the pseudo-code includes decoding the position of the last significant coefficient. The number of coefficient groups is then determined, whether the transform unit is 16x16 or greater (as indicated by if (log2TrafoSize >)), and the number of coefficients in each coefficient group. The second if-or statement reflects the decoding of significant coefficient indicators in the coefficient group containing the last significant coefficient.


[00136] The second to loop in the if-or-main statement reflects group-by-group processing in group scan order. In the first part of that loop, two special cases are handled in an if-or statement: the significant coefficient group indicator is set to 1 if the coefficient group is the upper left group (that is, nCG = 0) or if the groups to the right and bottom of the current group have significant coefficient group indicators that are both non-zero. If these two cases do not apply (the “or” clause), then the significant coefficient group indicator is decoded from the bit stream (significant_coeffgroup_flag [xCG][yCG].
[00137] Following this if-or statement, an additional if-or statement provides that if the significant coefficient group indicator is non-zero, then the significant coefficient indicators for that group are decoded from the bitstream in order of sweep. All but the last (upper left) significant coefficient indicator in the group are decoded from the bit stream. The decoder then evaluates whether any of the significant coefficient indicators decoded for that group are non-zero. If so, then decodes the last significant coefficient indicator (upper left) for that group; and otherwise, sets the same to 1 because it knows it cannot be zero.
[00138] The following “or” statement applies to the situation where the significant coefficient group indicator is zero. In this case, all significant coefficient indicators for this group are set to zero.
[00139] The above pseudo code shows an example implementation of example method 700 described above with respect to figure 16. Significant coefficient group indicators and their corresponding significant coefficient indicators (if any) are interleaved in this mode. A distinction between example method 700 in Figure 16 and example pseudocode is that example method 700 handles the upper left group special case in operation 720, whereas the pseudocode handles that special case in operation 708 by adjusting the significant coefficient group indicator for that group to be non-zero and thereby decode the significant coefficient indicators of that group from the bit stream in operation 714.
[00140] In another embodiment the coefficient group size can be fixed. The example syntax using fixed 4x4 coefficient groups is exposed in the following pseudocode:



[00141] In yet another embodiment, significant coefficient group indicators cannot be interleaved with significant coefficient indicators in the bit stream. That is, the significant coefficient group indicators are encoded in the bit stream together and the significant coefficient indicators, group by group in scan order, follow the same in the bit stream. The example syntax corresponding to a non-interleaved modality is exposed in the following pseudocode:



[00142] Reference is now made to Fig. 12, which shows a simplified block diagram of an example embodiment of an encoder 900. Encoder 900 includes a processor 902, memory 904, and a coding application 906. encoding 906 may include a computer program or application stored in memory 904 and containing instructions for configuring processor 902 to perform operations such as those described herein. For example, encoding application 906 may encode and transmit encoded bit streams in accordance with the multi-level significance map processes described herein. It will be understood that the encoding application 906 may be stored on a computer readable medium such as a compact disc, flash memory device, random access memory, hard drive, etc.
[00143] Reference is now also made to Fig. 13, which shows a simplified block diagram of an example embodiment of a decoder 1000. The decoder 1000 includes a processor 1002, a memory 1004, and a decoding application 1006. decoding application 1006 may include a computer program or application stored in memory 1004 and containing instructions for configuring processor 1002 to perform operations such as those described herein. Decoding application 1006 may include an entropy decoder configured to reconstruct residuals based on multi-level significance maps, as described herein. It will be understood that the decoding application 1006 may be stored on a computer readable medium such as a compact disc, flash memory device, random access memory, hard drive, etc.
[00144] It will be recognized that the decoder and/or encoder in accordance with the present application may be implemented in a variety of computing devices, including, without limitation, servers, appropriately programmed general purpose computers, playback devices, and audio/video encoding , television frequency converters, television broadcast equipment, and mobile devices. The decoder or encoder can be implemented through instructions containing software to configure a processor to perform the functions described here. Software instructions may be stored in any suitable non-transient computer-readable memory, including CDs, RAM, ROM, flash memory, etc.
[00145] It will be understood that the encoder described herein and the module, routine, process, thread or other software component implementing the method/process described for configuring the encoder may be realized using computer programming techniques and standard languages. The present application is not limited to specific processors, computer languages, computer programming conventions, data structures, other such implementation details. Those skilled in the art will recognize that the processes described can be implemented as a piece of computer executable code stored in volatile or non-volatile memory, as part of an application-specific integrated chip (ASIC), etc.
[00146] Certain adaptations and modifications of the described modalities can be made. Therefore, the embodiments discussed above are considered to be illustrative and not restrictive.
权利要求:
Claims (16)
[0001]
1. A method, in a decoder, of reconstructing significant coefficient indicators for a transform unit, wherein the transformed unit comprises non-overlapping blocks of transform coefficients, each of the non-overlapping blocks corresponding to a respective group of indicators of transformation. significant coefficient, each significant coefficient indicator corresponds to a respective transform coefficient and each respective group corresponds to a respective significant coefficient indicator group, groups of significant coefficient indicators are reconstructed in a group scan order, and coefficient indicators significant are reconstructed in a scan order within their groups, the method characterized by the fact that it comprises: decoding a position to a last significant coefficient of the transform unit from a stream of encoded data bits; in a corresponding first group to the position of the u last significant coefficient, decode from the bitstream each significant coefficient indicator in the scan order from a position following the position of the last significant coefficient in the scan order through the position (0,0) of the first group; for each group in group scan order from a second group, which follows the first group in group scan order, to a group that precedes a DC group in group scan order, where the DC group corresponds to a block containing a transform coefficient at position (0,0) of the transform unit, decode a significant coefficient group indicator from the bitstream for that group, and reconstruct the significant coefficient indicators in that group in a scan order on that group group, by, for each significant coefficient indicator in that group, if that significant coefficient group indicator is in position (0,0) in that group, its in corresponding significant coefficient group indicator is decoded to be non-zero, and all significant coefficient indicators that precede that significant coefficient indicator in the scan order in that group are zero, and then infers, without decoding from the bitstream, that this significant coefficient indicator is 1, and otherwise decoding that significant coefficient indicator from the bitstream if that significant coefficient indicator is in a group that has a corresponding significant coefficient group indicator that is non-zero, and inferring, without decoding from the stream of bits, that significant coefficient indicator to zero, if that significant coefficient indicator is in a group that has a corresponding significant coefficient group indicator that is zero; and in the DC group, decode each corresponding significant coefficient indicator in scan order.
[0002]
2. Method according to claim 1, characterized in that the group scan order is one of horizontal, vertical or diagonal.
[0003]
3. Method according to claim 1, characterized in that the scanning order is horizontal, vertical or diagonal.
[0004]
4. Method according to claim 1, characterized in that the significant coefficient indicators of a following of the respective groups are reconstructed only after all significant coefficient indicators of all previous groups in the group scan order are reconstructed .
[0005]
5. Method according to claim 1, characterized in that it further comprises reconstructing a significant coefficient group indicator for the first group by inferring, without decoding from the bit stream, that the significant coefficient group indicator for the first group is 1.
[0006]
6. Method according to claim 1, characterized in that it further comprises reconstructing a significant coefficient group indicator for the DC group by inferring, without decoding from the bit stream, that the significant coefficient group indicator for the DC group is 1.
[0007]
7. Method according to claim 1, characterized in that the blocks are contiguous square blocks.
[0008]
8. Method according to claim 1, characterized in that decoding at least one of the significant coefficient group indicators includes determining a context for that significant coefficient group indicator based on the significant coefficient group indicators decoded for two neighboring groups.
[0009]
9. A decoder that reconstructs significant coefficient indicators for a transform unit, wherein the transform unit comprises non-overlapping blocks of transform coefficients, each of the non-overlapping blocks corresponds to a respective group of significant coefficient indicators, each significant coefficient indicator corresponds to a respective transform coefficient and each respective group corresponds to a respective group of significant coefficient indicators, groups of significant coefficient indicators are reconstructed in a group scan order and significant coefficient indicators are reconstructed in an order scanning within its group, the decoder characterized by the fact that it comprises: a processor; a memory; and a decoding application stored in memory and containing instructions for configuring the processor to: decode a position for a last significant coefficient of the transform unit from a stream of encoded data bits; into a first group corresponding to the position of the last significant coefficient , decode from the bitstream each significant coefficient indicator in the scan order from a position following the position of the last significant coefficient in the scan order through the position (0,0) of the first group; for each group in the order group scan from a second group, which follows the first group in group scan order, to a group that precedes a DC group in group scan order, where the DC group corresponds to a block containing a transform coefficient at position (0,0) of the transform unit, decode a significant coefficient group indicator from the stream d and bits for that group, and reconstruct the significant coefficient indicators in that group in a scan order in that group, by, for each significant coefficient indicator in that group, if that significant coefficient group indicator is at position (0,0) in that group group, its corresponding significant coefficient group indicator is decoded to be non-zero, and all significant coefficient indicators that precede that significant coefficient indicator in scan order in that group are zero, and infer, without decoding from the bitstream , that the significant coefficient indicator is 1, and otherwise decode that significant coefficient indicator from the bitstream if that significant coefficient indicator is in a group that has a corresponding significant coefficient group indicator that is non-zero, and infer, without decoding from the bitstream, this significant coefficient indicator to zero if that significant coefficient indicator is in a group that has a corresponding significant coefficient group indicator that is zero; and in the DC group, decode each corresponding significant coefficient indicator in scan order.
[0010]
10. Decoder, according to claim 9, characterized in that the group scan order is horizontal, vertical or diagonal.
[0011]
11. Decoder, according to claim 9, characterized in that the scanning order is horizontal, vertical or diagonal.
[0012]
12. Decoder, according to claim 9, characterized in that the processor is configured to reconstruct significant coefficient indicators from a next one of the respective groups only after all significant coefficient indicators from all previous groups in the scan order of group be reconstructed.
[0013]
13. Decoder, according to claim 9, characterized in that the processor is further configured to reconstruct a significant coefficient group indicator for the first group by inferring, without decoding from the bit stream, that the group indicator significant coefficient for the first group is 1.
[0014]
14. Decoder, according to claim 9, characterized in that the processor is further configured to reconstruct a significant coefficient group indicator for the DC group by inferring, without decoding from the bit stream, that the group indicator of significant coefficient for the DC group is 1.
[0015]
15. Decoder, according to claim 9, characterized in that the blocks are contiguous square blocks.
[0016]
16. Decoder, according to claim 9, characterized in that the processor is configured to decode at least one of the significant coefficient group indicators by determining a context for that significant coefficient group indicator based on the group indicators of significant coefficient decoded for two neighboring groups.
类似技术:
公开号 | 公开日 | 专利标题
BR102012029413B1|2022-02-15|Method and decoder for reconstructing significant coefficient indicators for a transform unit
US10911758B2|2021-02-02|Multi-level significance maps for encoding and decoding
BR102013001124A2|2018-07-31|HIDING MULTIPLE SIGNAL BITS INSIDE A TRANSFORMED UNIT
BR112020019715A2|2021-02-09|combination of extended position-dependent intraprediction with angular modes
BR112019021593A2|2020-05-12|METHOD AND DEVICE FOR ENTHROPIC VIDEO SIGNALS ENCODING AND DECODING
TWI510031B|2015-11-21|Multi-level significance maps for encoding and decoding
BR102013001517B1|2020-12-15|methods and devices for selecting a set of contexts
US20130188724A1|2013-07-25|Methods and systems for pipelining within binary arithmetic coding and decoding
BR112020000166A2|2020-07-14|apparatus for dividing a portion of the image improving the forced border partition, encoder and decoder comprising the same and corresponding methods
BR102012027968B1|2021-12-21|METHOD OF RECONSTRUCTION OF SIGNIFICANT COEFFICIENT FLAGS AND DECODER FOR DECODING A BITS STREAM OF ENCODED DATA
CA2800119C|2016-05-17|Methods and systems for pipelining within binary arithmetic coding and decoding
同族专利:
公开号 | 公开日
US10681362B2|2020-06-09|
BR102012029413A2|2016-02-16|
SG190516A1|2013-06-28|
HK1179451A1|2013-09-27|
US9743098B2|2017-08-22|
US10116950B2|2018-10-30|
JP2013110740A|2013-06-06|
KR20130056183A|2013-05-29|
TWI483618B|2015-05-01|
JP2014195269A|2014-10-09|
JP5925830B2|2016-05-25|
CA2773990A1|2013-05-19|
AU2012245178B2|2013-11-28|
TW201334550A|2013-08-16|
US20130128985A1|2013-05-23|
BR102012029413A8|2021-08-24|
US20190052890A1|2019-02-14|
KR101538832B1|2015-07-22|
CN103124349B|2016-09-14|
US20180007376A1|2018-01-04|
EP2595380A1|2013-05-22|
JP5542188B2|2014-07-09|
EP2595380B1|2015-10-21|
AU2012245178A1|2013-06-06|
CN103124349A|2013-05-29|
CA2773990C|2015-06-30|
引用文献:
公开号 | 申请日 | 公开日 | 申请人 | 专利标题

US7190840B2|2002-01-07|2007-03-13|Hewlett-Packard Development Company, L.P.|Transform coefficient compression using multiple scans|
KR100667808B1|2005-08-20|2007-01-11|삼성전자주식회사|Method and apparatus for intra prediction encoding and decoding for image|
US20070071331A1|2005-09-24|2007-03-29|Xiteng Liu|Image compression by economical quaternary reaching method|
KR100873636B1|2005-11-14|2008-12-12|삼성전자주식회사|Method and apparatus for encoding/decoding image using single coding mode|
KR101356733B1|2007-03-07|2014-02-05|삼성전자주식회사|Method and apparatus for Context Adaptive Binary Arithmetic Coding and decoding|
KR101375668B1|2008-03-17|2014-03-18|삼성전자주식회사|Method and apparatus for encoding transformed coefficients and method and apparatus for decoding transformed coefficients|
EP2154894A1|2008-08-15|2010-02-17|Thomson Licensing|Video coding with coding of the locations of significant coefficients in a block of coefficients|
TWI482499B|2008-10-03|2015-04-21|Ind Tech Res Inst|Video signal processing apparatus and method|
KR101457894B1|2009-10-28|2014-11-05|삼성전자주식회사|Method and apparatus for encoding image, and method and apparatus for decoding image|
US8879632B2|2010-02-18|2014-11-04|Qualcomm Incorporated|Fixed point implementation for geometric motion partitioning|
US8942282B2|2010-04-12|2015-01-27|Qualcomm Incorporated|Variable length coding of coded block pattern in video compression|
CN103119849B|2010-04-13|2017-06-16|弗劳恩霍夫应用研究促进协会|Probability interval partition encoding device and decoder|
KR101605163B1|2010-04-13|2016-03-22|지이 비디오 컴프레션, 엘엘씨|Coding of significance maps and transform coefficient blocks|
US9154801B2|2010-09-30|2015-10-06|Texas Instruments Incorporated|Method and apparatus for diagonal scan and simplified coding of transform coefficients|
US8767824B2|2011-07-11|2014-07-01|Sharp Kabushiki Kaisha|Video decoder parallelization for tiles|JPH10176153A|1996-12-17|1998-06-30|Bunchiyou:Kk|Solid adhesive|
CN108600761B|2012-01-03|2020-05-08|寰发股份有限公司|Method and apparatus for coding importance group flags|
US9621894B2|2012-01-13|2017-04-11|Qualcomm Incorporated|Determining contexts for coding transform coefficient data in video coding|
AU2012200319B2|2012-01-19|2015-11-26|Canon Kabushiki Kaisha|Method, apparatus and system for encoding and decoding the significance map for residual coefficients of a transform unit|
US9106918B2|2012-01-20|2015-08-11|Sony Corporation|Coefficient coding harmonization in HEVC|
US9350998B2|2012-06-29|2016-05-24|Qualcomm Incorporated|Coding of significance flags|
US9386306B2|2012-08-15|2016-07-05|Qualcomm Incorporated|Enhancement layer scan order derivation for scalable video coding|
KR102134367B1|2012-09-10|2020-07-15|선 페이턴트 트러스트|Image coding method, image decoding method, image coding device, image decoding device, and image coding/decoding device|
US9538175B2|2012-09-26|2017-01-03|Qualcomm Incorporated|Context derivation for context-adaptive, multi-level significance coding|
KR101502144B1|2013-08-22|2015-03-12|주식회사 에스원|Method and apparatus for transforming coefficient information|
US9445132B2|2013-09-09|2016-09-13|Qualcomm Incorporated|Two level last significant coefficientposition coding|
KR102218196B1|2013-10-28|2021-02-23|삼성전자주식회사|Encoder, method thereof, and apparatuses including the encoder|
US9781424B2|2015-01-19|2017-10-03|Google Inc.|Efficient context handling in arithmetic coding|
JP6476900B2|2015-01-21|2019-03-06|富士通株式会社|Moving picture coding apparatus, moving picture coding method, and computer program for moving picture coding|
US10574993B2|2015-05-29|2020-02-25|Qualcomm Incorporated|Coding data using an enhanced context-adaptive binary arithmetic codingdesign|
CN106303535A|2015-06-08|2017-01-04|上海天荷电子信息有限公司|Reference pixel takes from method for compressing image and the device of reconstructed pixel in various degree|
US10631013B2|2015-06-11|2020-04-21|Lg Electronics Inc.|Method for re-positioning significant coefficient in image encoding/decoding process and device therefor|
US10674146B2|2015-09-30|2020-06-02|Lg Electronics Inc.|Method and device for coding residual signal in video coding system|
KR20180007680A|2016-07-13|2018-01-23|한국전자통신연구원|Method and apparatus for encoding and decoding image|
CN110622511A|2017-04-13|2019-12-27|Lg 电子株式会社|Image encoding/decoding method and apparatus thereof|
法律状态:
2015-09-08| B25D| Requested change of name of applicant approved|Owner name: BLACKBERRY LIMITED (CA) |
2015-09-29| B25G| Requested change of headquarter approved|Owner name: BLACKBERRY LIMITED (CA) |
2016-02-16| B03A| Publication of a patent application or of a certificate of addition of invention [chapter 3.1 patent gazette]|
2018-12-04| B06F| Objections, documents and/or translations needed after an examination request according [chapter 6.6 patent gazette]|
2019-12-10| B25A| Requested transfer of rights approved|Owner name: VELOS MEDIA INTERNATIONAL LIMITED (IE) |
2020-05-05| B06U| Preliminary requirement: requests with searches performed by other patent offices: procedure suspended [chapter 6.21 patent gazette]|
2021-08-24| B03H| Publication of an application: rectification [chapter 3.8 patent gazette]|Free format text: REFERENTE A RPI 2354 DE 16/02/2016,QUANTO AO ITEM (54). |
2021-11-03| B06A| Patent application procedure suspended [chapter 6.1 patent gazette]|
2021-12-21| B09A| Decision: intention to grant [chapter 9.1 patent gazette]|
2022-02-15| B16A| Patent or certificate of addition of invention granted [chapter 16.1 patent gazette]|Free format text: PRAZO DE VALIDADE: 20 (VINTE) ANOS CONTADOS A PARTIR DE 19/11/2012, OBSERVADAS AS CONDICOES LEGAIS. |
优先权:
申请号 | 申请日 | 专利标题
US201161561872P| true| 2011-11-19|2011-11-19|
US61/561,872|2011-11-19|
US13/446,186|US9743098B2|2011-11-19|2012-04-13|Multi-level significance map scanning|
US13/446,186|2012-04-13|
[返回顶部]