• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    An active pixel image sensor comprising a matrix (M) of pixels (P) organized by rows and columns, and a read circuit (CL) having a separate read channel (VL) for each column of pixels, wherein each pixel comprises: a photodiode (PPD), a storage node (FD), a transfer transistor, at least one reset transistor of the storage node, a line selection transistor and a voltage follower transistor; and each reading channel comprises a subtraction block (CND) connected to receive on a first input the voltage (VP) at the terminals of the storage node of one of the pixels of the corresponding column and on a second input a reference voltage (VREF) substantially equal to the reset voltage of the pixels of the matrix seen at the input of the read channel; and in that said sensor also comprises a controller (CTR) configured to drive the pixel transistors and the read circuit for performing image acquisition in global shutter mode with subtraction of the reset noise and non-destructive readout of the pixels . Method of acquiring images by means of such a sensor
    公开号:FR3039319A1
    申请号:FR1557073
    申请日:2015-07-24
    公开日:2017-01-27
    发明作者:Gregoire Chenebaux
    申请人:PYXALIS；
    IPC主号:
    专利说明:

    ACTIVE PIXEL IMAGE SENSOR WITH GLOBAL SHUTTER MODE OPERATION, RESTRICTED NOISE CANCELLATION AND NON-DESTRUCTIVE READING The present invention relates to an active pixel type image sensor and an image acquisition method. using such a sensor. The invention aims in particular to allow the acquisition of images in global shutter mode with subtraction of the reset noise in combination with the non-destructive reading of the pixels.
    Active pixel image sensors, typically made using CMOS (metal, oxide, complementary semiconductor) technology, are commonly used in electronic imaging systems. In these sensors, each pixel comprises a photodiode, optionally a node for storing the photogenerated charges by the photodiode ("floating diffusion") in the case of using a pinned photodiode ("pinned photodiode" in English), and a plurality of transistors (typically between 3 and 5 per pixel). A general introduction to these sensors is provided, for example, by the article by Abbas El Gamal and Helmu Rlyoukhy "CMOS Image Sensors", IEEE Circuits & Devices Magazine, May-June 2005, pages 6 to 20.
    It is known that active pixel image sensors can operate without the need for a mechanical shutter. The function of the shutter is then obtained purely electronically, by appropriate control of the transistors of each pixel. The electronic shutter thus produced may be of the "rolling" or "global" type. In the case of a rolling shutter, the integration periods of the different pixel lines are out of phase with each other. In the case of a global shutter ("global shutter" in English), on the other hand, all the pixels integrate the incident light simultaneously, the reading of the pixels being done later and line by line. Operation in rolling shutter mode is simpler to perform and maximizes the integration time for a given image acquisition rate, and therefore the sensitivity of the sensor, but produces harmful artifacts (distortion of the scene ...) if the scene changes quickly (moving object, sudden change in lighting conditions, ...). Thus, some applications require a global shutter.
    It is also known that active pixel image sensors are affected by different types of noise, among which may be mentioned the so-called "kTC" noise, or resetting noise. Before acquiring an image, an active pixel is reset to a voltage (potential difference with respect to a mass) assumed to be known; then it begins to accumulate photogenerated charges, which leads to a decrease in its voltage. If the reset voltage was well known, measuring the voltage at the pixel output at the end of the integration would determine the incident light intensity. In fact, the reset voltage varies unpredictably from one image to another due to thermal fluctuations of the charge carriers. To remove this noise source it is necessary to sample twice the output voltage of the pixel - just after the reset and at the end of the integration period for a three-transistor (3T) pixel, or just after the reset and just after a charge transfer for a pixel with four or more transistors (4T) and to subtract the first sample from the second. This technique is known as Correlated Double Sampling (CDS). In its simplest form, the correlated double sampling is not compatible with the use of a global shutter for matrix sensors having a consequent number of lines (of the order of ten or more). In this case, it is known to use a technique known as "quadruple correlated sampling" (CQS), which consists in acquiring two images for each image, one in the absence of light ("black image"). representative of the resetting voltages of all the pixels) and the other after having illuminated the pixel during the desired integration time ("integration image" obtained after an overall transfer of the charges in the matrix). The acquisition of the black and integration images is made possible by the fact that, in the pixels compatible with this technique, a so-called transfer transistor makes it possible to isolate the source of the photogenerated charges (pinched photodiode) of a storage node. unlit. Both the black image and the integration image are obtained by difference between two voltage samples. The black image is the result of the difference between a reference voltage and a voltage acquired after the end of the reset of the storage node. The integration image is the result of the difference of a voltage acquired at the end of the integration time, after a transfer operation performed globally between the black image and the integration image, and the same reference voltage. More precisely, the storage nodes are reset, line by line, and the resetting voltage is concomitantly acquired, which is a constant and low-noise voltage level which serves as a reference voltage; after a predefined time, the resetting of the storage node is stopped and a new acquisition of the voltage of the pixels representative of the noise sampled on the storage node is made, the difference between the two acquisitions constituting the black image. During this time, the photodiodes of all the pixels integrate electric charges which, after the acquisition of the black image, are transferred globally to the corresponding storage nodes. To obtain the integration image, a first acquisition of the voltage at the terminals of the storage nodes is carried out, then during the reinitialization of said nodes, at a second voltage acquisition, in order to obtain still the reference level, low noise, and finally to the subtraction of the two tensions thus acquired. The black image obtained in this way is essentially a measurement of the post-reset noise (random difference between the reset voltage and the voltage at the terminals of the storage node after reset), which noise itself affects the integration image. because the storage node is not reset between the two acquisitions. Therefore this noise can be removed by subtracting the two images. The article by B. Fowler et al. "A 5.5 Mpixel 100 Frames / sec Wide Dynamic Range Low Noise CMOS Image Sensor for Scientific Applications", SPIE Proceedings 7536 - Sensors, Cameras, and Systems for Industrial / Scientific Applications XI, describes an active pixel image sensor implementing the CQS technique combined with a global electronic shutter. A disadvantage of such a sensor is that the reading of the voltages of the pixels is necessarily destructive, because it is accompanied by a reset of the storage nodes in order to acquire the reference voltage. However, non-destructive reading would be preferable in some applications because it would allow a plurality of pixel readings while continuing to integrate photogenerated charges to stop integration when the exposure level is optimal. The invention aims to provide an active pixel image sensor for combining operation in global shutter mode with noise subtraction reset (or KTC) and non-destructive readout of the pixels. Incidentally, the invention also aims to minimize the overall noise level and the complexity of the pixel reading electronics.
    An idea underlying the invention that makes it possible to achieve this goal is to use an external voltage reference generated, for example, at the readout circuit of the pixel array, instead of using the reference level as the reference level. reset voltage of the storage nodes. This allows this reset only once, during the reading phase of the black image which is thus non-destructive. Moreover, the reference voltage generated in the read circuit can be much less noisy than the reset voltage of the storage nodes; indeed, a high performance voltage level generation circuit may be provided outside the array, but can not reasonably be integrated with the pixels. In addition, this reference voltage is permanently available, which simplifies the design of the read circuit compared to the case described in the aforementioned article by B. Fowler et al.
    An object of the invention is therefore an active pixel image sensor comprising a matrix of pixels organized by rows and columns as well as a reading circuit comprising a distinct reading channel for each column of pixels, in which each pixel comprises a photodiode, a storage node of the electric charges integrated by said photodiode, a transfer transistor for selectively enabling the transfer of said electrical charges from the photodiode to the storage node, a resetting transistor of the storage node for connecting selectively said storage node to a reset voltage source, o a line selection transistor for selectively connecting said pixel to the read channel of the column to which it belongs, and o a voltage follower transistor for transferring to said read path the voltage across the storage node through said transistor of line selection; characterized in that each read channel comprises a subtraction block having two inputs and an output, said subtraction block being connected to receive on a first input the voltage across the storage node of one of the pixels of the corresponding column by means of the voltage follower transistor and the line selection transistor of said pixel and on a second input a reference voltage of value substantially equal to the reset voltage of the matrix pixels seen at the input of the reading channel, and to provide at its output a signal representative of a difference in the voltage levels present at its inputs; and in that said sensor also comprises a controller configured to drive the pixel transistors and the read circuit for: performing a global reset of the photodiodes; perform a reset of the storage nodes of the pixels of the matrix and acquire a first digital image, called black image; then perform a global charge transfer of the photodiodes to the storage nodes of said pixels; then acquire a second digital image called integration image; to perform image acquisition in global shutter mode with subtraction of the reset noise and non-destructive pixel readings of the integrated charges.
    According to different embodiments of such a sensor:
    The sensor may also comprise at least one generator of said reference voltage, outside said pixel matrix.
    Each said subtracter block may be a differential digitizing string configured to provide at its output a digital signal representative of the difference of the analog voltage levels present at its inputs.
    The controller may be configured to drive the pixel transistors and the read circuit to reset the pixel storage nodes line by line.
    A first and a second sample - and - hold device may be present at the first and second inputs of each said subtraction block, respectively, and said controller may be configured to: a) enable the reset transistor (s) of all the pixels to reset their photodiodes and their storage nodes; b) activating the first sampled - blockers of each read channel so that a sample of the reference voltage is present on the first input of each subtraction block; c) for each pixel column, and after resetting the corresponding storage nodes: c1) successively activate the line selection transistors, c2) in correspondence of each activation of a line selection transistor, activate the second sampler -blocker the read path corresponding to said column such that a voltage sample is present on the second input of the corresponding subtraction block; then c3) controlling the analog - to - digital converter of said read channel, whereby a first digital image called black image is acquired line by line; and then: d) activating the transfer transistors of all the pixels, so as to perform a global charge transfer from the photodiodes to the storage nodes of said pixels; e) activating the first sampled - blockers of each read channel so that a sample of the reference voltage is present on the first input of each subtracter block; f) for each pixel column, and after charge transfer to the corresponding storage nodes: f1) successively activate the line selection transistors, f2) in correspondence with each activation of a line selection transistor, activate the second sampler -blocker of the read channel corresponding to said column, then f3) control the analog-digital converter of said read channel, whereby a second digital image called integration image (IIM) is acquired line by line; no reset of photodiodes or storage nodes being performed between operations b) and f3).
    Said controller may also be configured to repeat operations d) to f3) one or more times, depending on a signal level of said integration image (s), before performing operations a) to c) again. , so that one or more integration images are acquired for the same black image.
    The sensor may also include a digital image processor configured to subtract from said or said integration image the corresponding black image.
    Said pixels may comprise a reset transistor of the photodiode and a reset transistor of the separate storage node.
    Each pixel may also include a photodiode reset transistor for selectively connecting the photodiode to a reset voltage source.
    Another object of the invention is a method of acquiring images by means of an active pixel image sensor comprising a matrix of pixels organized by rows and columns as well as a reading circuit comprising a reading channel. distinct for each column of pixels, in which each pixel comprises: o a photodiode, o a storage node of the electric charges generated by said photodiode when it is illuminated, o a transfer transistor for selectively authorizing the transfer of said electric charges of the photodiode at the storage node, o at least one reset transistor for selectively connecting the photodiode and the storage node to a reset voltage source; o a line selection transistor for selectively connecting said pixel to the read path of the column to which it belongs, and o a transistor mounted as a voltage follower for transferring to said read path l a voltage across the storage node through said line selection transistor; the process comprising the following steps; A) enabling the reset transistor (s) of all the pixels to reset their photodiodes and their storage nodes; B) sampling a reference voltage, generated inside said read circuit and of value substantially equal to the reset voltage of the pixels of the matrix; C) for each pixel column, and after resetting the corresponding storage nodes: successively activate the line selection transistors, sample the voltage at the terminals of the storage node of the corresponding pixel, subtract the reference voltage sampled during the step B) of the voltage thus sampled and digitizing the result, so as to acquire line by line a first digital image called black image; and then: D) activating the transfer transistors of all the pixels, so as to perform a global charge transfer from the photodiodes to the storage nodes of said pixels; E) sampling a reference voltage, generated inside said read circuit and of value substantially equal to the reset voltage of the pixels of the matrix; F) for each pixel column, and after charge transfer to the corresponding storage nodes: successively activate the line selection transistors, sample the voltage across the storage node of the corresponding pixel, subtract the sampled reference voltage when step E) of the voltage thus sampled and digitize the result, so as to acquire line by line a second digital image called integration image.
    According to different embodiments of such a method:
    Steps D) to F) can be repeated one or more times, depending on a signal level of said integration image (s), before performing again said steps A) to C), so that one or more integration images are acquired for the same black image.
    The method may also include a step of subtracting from said or said integration image the corresponding black image. Other characteristics, details and advantages of the invention will emerge on reading the description given with reference to the accompanying drawings given by way of example and which represent, respectively: FIG. 1, an overall functional diagram of a image sensor according to an embodiment of the invention; FIG. 2, the electrical diagram of a five transistor active pixel that can be used in the image sensor of FIG. 1; FIG. 3 is a simplified electrical diagram of a reading channel of the image sensor of FIG. 1; Figure 4 a timing diagram illustrating a first mode of operation of the image sensor of Figure 1; and FIG. 5 is a timing diagram illustrating a second mode of operation of the image sensor of FIG. 1.
    As illustrated in FIG. 1, an image sensor according to one embodiment of the invention comprises a matrix M of active pixels P which are arranged in rows and columns. The matrix M used for acquiring images may be a subset of a larger matrix, some of whose lines and / or columns may not be used in a given application.
    One embodiment of the pixels of the matrix M is illustrated in FIG.
    Conventionally, all the pixels of the same column are connected, via selection transistors (not shown in FIG. 1, reference T4 in FIG. 2) to a respective read channel VL, a simplified diagram of which is provided in FIG. 3. When acquiring an image, the selection transistors of all the pixels of the same line are activated at the same time; thus an image is acquired line by line. The matrix M is connected to a source VSrst of DC voltage Vrst- This voltage is called "reset voltage" because, as will be explained with reference to FIG. 2, the storage nodes are reset by connecting them to the source VSrst for bring them to the potential Vrst-
    The VL reading channels are connected to a VSref source of DC voltage VREf · This voltage Vref, called the reference voltage, is substantially identical to the VRSt col image of the reset voltage Vrst on the column connector CCi linking the pixels of the matrix to VL reading channels. By "substantially identical" is meant that the voltage Vref is included in the possible distribution of the images on the column connector of the reset voltages of all the pixels of the matrix.
    In the embodiment of FIG. 1, the read channels and the reference voltage source are gathered in a read circuit CL. This circuit receives as input analog signals from the matrix M and provides at its output "raw" digital images, which are processed by a digital image processor DIP. The latter, in particular, subtracts the black images from the corresponding integration images in order to suppress the reset noise.
    The sensor of FIG. 1 also comprises a controller CTR which generates control signals of the transistors of the pixels of the matrix (R_PD, R_FD, CH_T, L_Si) and reading channels (SH_REF, SH_PIX). Timing diagrams of these signals are shown in Figure 4 and will be discussed in detail later. Optionally, the controller CTR may also receive an IMAX signal from the read circuit or the image processor and representative of a signal level of the acquired integration images. For example, IMAX may correspond to the highest signal level among the pixels of the array. The use of this signal will be detailed below, with reference to FIG. 4.
    The digital image processor DIP and the controller CTR may consist of (or include) one or more suitably programmed microprocessors and / or dedicated digital circuits.
    Figure 2 illustrates the architecture of a pixel P of the matrix Μ. This is a conventional architecture, called "5T", that is to say, 5 transistors. This pixel comprises a PPD pinch photodiode, which integrates electric charges when the pixel is illuminated. A transistor T1 is connected between the photodiode and a so-called "floating" pn junction ("floating diffusion"), identified by the reference FD, which is protected from light by an LSFD screen. When the transistor T1 is inactive, the photodiode PPD accumulates the charges of photonic origin; when the transistor - driven by the signal CFI_T generated by the controller CTR - is active, the charges stored in the photodiode PPD are transferred to the junction FD, which then serves as a load storage node. Two transistors T2, T3, driven by the signals R_PD and R_FD, respectively, make it possible to reset the photodiode and the charge storage node through the voltage source VSrst. The voltage at the terminals of the charge storage node is read by a transistor T5 connected as a voltage follower (the drain of this transistor being at the voltage Vrst), and transmitted to the reading channel of the "i" column to which the pixel belongs by means of a transistor T4 of line selection, driven by the L_Si signal (where "i" identifies the column).
    FIG. 3 illustrates the simplified electrical diagram of a reading channel VL according to one embodiment of the invention. This reading channel comprises a current source LP distributing a bias current to all the pixels of the same column, depending on the selection of the line L_Si. This reading channel also comprises a first sample - and - hold device SH1 controlled by the signal SH_REF generated by the controller CTR and whose input is connected to the reference voltage source VSref, and a second sample - and - hold device SH2 driven by the SH_PIX signal generated by the controller CTR and whose input is connected to the pixels of the same column of the matrix M via the line selection transistors of these pixels; the voltage applied to the input of SH2 is indicated by Vp when one of these selection transistors is activated. The outputs of these sample and hold devices are connected to the inputs of a differential digitizer chain CND, which can typically perform a single ramp analog-to-digital conversion. The digital signal generated by this converter, proportional to the digitized difference between the sampled values of VP and VREf, is inputted to the digital image processor DIP.
    FIG. 4 illustrates a mode of operation of the sensor of FIG. 1, and more precisely an embodiment of its image acquisition sequence.
    The sequence begins with a global reset (that is to say concerning all the pixels) of the photodiodes (signal R_PD). The global reset of the load storage nodes (R_FD signal) takes place shortly before the reading of the DIM image. The falling edge of the signal R_PD, ie the end of the global reset operation of the photodiodes, triggers the start of the time of integration of the light, which has a duration Ai and ends on the falling edge of the CH_T signal which triggers the stop of the global charge transfer from the photodiodes to the charge storage nodes.
    In parallel with the integration time Ai, acquisition of the black image DIM is performed, line by line. The insert in the lower left of the figure illustrates the operations that lead to the acquisition of a pixel of the black image; these operations are performed simultaneously for all the columns (and therefore all the reading channels) and successively for all the rows. First, the signal SH_REF activates the sample-and-hold device SH1 so that it acquires a sample of the reference voltage Vref. Next, the signal L_Si activates the line selection transistor of a pixel of the ith line. matrix M; during the activation time of this transistor, the signal SH_PIX activates the sample and hold circuit SH2 to acquire a sample of the voltage Vp at the terminals of the charge storage node of this pixel. The three operations L_Si, SH_PIX and SH_REF could also be done simultaneously for all the lines. Since there has been no load transfer since the storage node reset operation (R_FD), the voltage Vp is equal to Vrstcol, the image of the reset voltage Vrst to the driver Vp, to which is added - algebraically - a thermal noise V "tc: Vp = VRSt_col + Vktc · The output signal of the differential digitizing chain CND is a digital image of the difference of the two sampled values: sdDiM = G (VRST_coL + V« tc-Vref), where G represents a gain of the VL reading chain. Since Vref ^ Vrst_col and has almost no noise, we can write sdDiM ^ G-VKTc, where the index "DIM" reminds us that it is the signal relating to the black image. It can thus be seen that the black image essentially constitutes a mapping of the reset noise of the matrix M. The acquisition of the integration image IIM (illustrated by the insert in the lower right part of the figure) comprises the same operations, but which are implemented after the overall load transfer. Thus, in this case VP = VRSt_col + Vktc + Vint where V | NT (generally less than 0) is the useful signal on the conductor Vp, proportional to the light energy intercepted by the photodiode of the pixel during the integration time At Remembering that VREf ^ Vrst_col can be written sdiNT ^ G- (V | NT + VKtc) · Therefore, the reset noise V "tc can be suppressed by calculating the difference between sdiNT and sdoiM: sdiNT - sdoiM = GVint Advantageously, this subtraction is calculated numerically by the DIP processor.
    FIG. 5 illustrates an alternative mode of operation of the sensor of FIG. 1, and more precisely another embodiment of its image acquisition sequence, in which the storage nodes are not reinitialized simultaneously, but line per line during the acquisition (also performed line by line) of the black image. In this case, the single reset signal of the storage nodes R_FD is replaced by a plurality of reset signals R_FDi, one per line. For each line "i", this reset signal R FDi must precede the sampling of the signal Vp, and therefore the falling edge of the signal SH_PIX. On the other hand, the reset of all the photodiodes remains global, as well as their transfer of charge, so that the integration phase is simultaneous for all the pixels of the matrix. Resetting the storage nodes line by line reduces the fluctuations of the reset voltage, and therefore the static noise level of the image.
    In the sensor described by the aforementioned article by B. Fowler et al., The reference voltage is acquired by reading each pixel during the reset of its charge storage node. Thus, the reference voltage (or reset) is not available permanently, and must be stored in an analog memory to be subtracted from the useful signal. To implement this analog subtraction, in addition, it is necessary to provide a switching module for inverting the two inputs of the differential amplifier arranged at the input of its analog-to-digital converter to change the signal of the signal. exit ; the invention avoids this additional complexity. In addition, the choice of a reference outside the matrix allows a reference to very low noise, essential element of the previous operation. Thus, the invention allows an appreciable simplification of the architecture of the read circuit accompanied by an improvement in performance in terms of noise.
    But the main advantage of the invention is that the reading of a pixel does not require the reset of its storage node, or its photodiode; it is said to be nondestructive. Thus, after the first global transfer of charge (CH_T 1 in FIG. 4) and the reading of the pixels line by line, the photodiodes continue to integrate photogenerated charges. It is therefore possible to carry out subsequent load transfers - pulses CH_T2, CH_T3 ... - followed by other non-destructive readings. Thus, for a single black image, a plurality of integration images IIM, IIM2, IIM3 corresponding to increasing integration times are obtained (for example Atmi, 2 · Δϊιντ and 3 · Δϊιντ, respectively.
    The integration times are not necessarily identical if R_PD pulses should appear between CH_T1 and CH_T2 or CH_T2 and CH_T3 for example). This allows in particular to optimize the exposure of the final image. For example, after acquiring the first integration image, the DIP processor can determine the IMAX signal intensity corresponding to the brightest pixel of that image. This value is passed to the CTR controller. If IMAX is sufficiently close to a predefined value, corresponding to the saturation of the pixel, the acquisition sequence stops; otherwise, we proceed to another non-destructive reading, and so on. The integration time is thus adapted to the lighting conditions, within the constraints imposed by the acquisition rate requirements. Such a mode of operation would not be possible in the case of the sensor of the aforementioned article by B. Fowler et al., In which the reading of a pixel is necessarily destructive, always accompanied by its reinitialization. To achieve the same cumulative signal level with the sensor of B. Fowler et al. multiple acquisitions should be made, each with its own noise-generating reset. In the case of the invention, on the other hand, only two images are subtracted in all cases, which considerably reduces the noise level. The invention has been described in connection with a particular embodiment, but many variants can be envisaged. In particular :
    The sensor may not include the DIP processor, whose functions can be implemented by an external device. The same goes for the CTR controller.
    A single processor and / or dedicated digital circuit can be used to accomplish the functionality of the DIP processor and the CTR controller. Or, the DIP processor and the CTR controller may be physically distinct devices.
    The read circuit, the DIP processor, the CTR controller, the voltage sources VSrst and VSref can be co-integrated with the matrix M, or not. These elements can be made in completely integrated form or use discrete components.
    Several known schemes can be used to realize the VSrst and VSref-
    The active pixels may have architectures different from that ("5T") illustrated in FIG. 1. For example, it is possible to use "4T" pixels which do not include the transistor T2, so that the Resetting the photodiode is done through T1 and T3. In any case, each pixel must comprise at least one photodiode, a storage node, a transfer transistor, at least one reset transistor, a line selection transistor and at least one amplifier-mounted transistor, for example as a voltage follower.
    The diagram of Figure 3 is very simplified. Reading circuits known from the prior art can be used for the implementation of the invention.
    In the diagram of Figure 3 we considered the case where each reading channel comprises a differential scanning channel. Alternatively, the read channels may include analog subtractor circuits and output analog differential signals at their output, digitized with an analog-to-digital converter; the latter may be common to several reading channels, or all of them, and need not be integrated into the read circuit (it may, for example, be part of the image processor). One could even consider a digitization of sampled Vp and Vref signals, followed by a subtraction operation implemented digitally. In any case, each reading channel must include a subtraction block - made in analogue, digital form or in the form of a differential analog-to-digital converter.
    The image acquisition sequence may differ from that described with respect to FIG. 4. For example: the photodiodes may be reset before, after or at the same time as the storage nodes; the reference voltage can be acquired before, after or at the same time as the voltage at the terminals of each storage node.
    A sensor according to the invention is adapted to be used in global shutter mode while allowing a non-destructive reading of the pixels. Nevertheless, it can also be used in rolling shutter mode and the acquisition of multiple integration images is only optional.
    In case of acquisition of multiple integration images for the same black image, the elementary integration times separating two successive charge transfers may not be the same.
    权利要求:
    Claims (12)
    [1" id="c-fr-0001]
    An active pixel image sensor comprising a matrix (M) of pixels (P) organized by rows and columns and a read circuit (CL) having a separate read channel (VL) for each column of pixels, wherein each pixel comprises: o a photodiode (PPD), o a storage node (FD) of the electric charges integrated by said photodiode, o a transfer transistor (T 1) for selectively enabling the transfer of said electrical charges from the photodiode to storage node, o a storage node reset transistor (T3) for selectively connecting said storage node to a reset voltage source (VSrst) (VRSt), o a line selection transistor (T4) for selectively connecting said pixel to the reading channel of the column to which it belongs, and o a transistor (T5) mounted as a voltage follower for transferring to said reading channel the voltage at the terminals of the storage node by the integer intermediate of said line selection transistor; characterized in that each read channel comprises a subtraction block (CND) having two inputs and an output, said subtraction block being connected to receive on a first input the voltage (Vp) across the storage node of one pixels of the corresponding column via the voltage follower transistor and the line selection transistor of said pixel and on a second input a reference voltage (VREf) of value substantially equal to the reset voltage of the pixels of the corresponding column; the matrix seen at the input of the read channel, and to provide at its output a signal representative of a difference in the voltage levels present at its inputs; and in that said sensor also comprises a controller (CTR) configured to drive the pixel transistors and the read circuit for: performing a global reset of the photodiodes; performing a reset of the storage nodes of the pixels of the matrix and acquiring a first digital image, called black image (DIM); then perform a global charge transfer of the photodiodes to the storage nodes of said pixels; then acquire a second digital image called integration image (IIM); to perform image acquisition in global shutter mode with subtraction of the reset noise and non-destructive pixel readings of the integrated charges.
    [2" id="c-fr-0002]
    2. An image sensor according to claim 1 also comprising at least one generator (VSref) of said reference voltage (Vref), outside said pixel matrix.
    [3" id="c-fr-0003]
    3. An image sensor according to one of the preceding claims, wherein each said subtractor block is a differential digitizing chain configured to provide at its output a digital signal representative of the difference of the analog voltage levels present at its inputs.
    [4" id="c-fr-0004]
    4. An image sensor according to one of the preceding claims wherein said controller (CTR) is configured to drive the pixel transistors and the read circuit to reset the storage nodes of the pixels line by line.
    [5" id="c-fr-0005]
    The image sensor according to one of the preceding claims wherein a first and a second sample-and-hold (SH1, SH2) are present at the first and second inputs of each said subtraction block, respectively, and said controller is configured to : a) enabling the reset transistor (s) of all the pixels to reset their photodiodes and their storage nodes; b) activating the first sample-and-hold of each read channel so that a sample of the reference voltage is present on the first input of each subtraction block (CND); c) for each pixel column, and after resetting the corresponding storage nodes: c1) successively activate the line selection transistors, c2) in correspondence of each activation of a line selection transistor, activate the second sampler -blocker the read path corresponding to said column such that a voltage sample is present on the second input of the corresponding subtraction block; then c3) driving the analog to digital converter of said read channel, whereby a first digital image called black image (DIM) is acquired line by line; and then: d) activating the transfer transistors of all the pixels, so as to perform a global charge transfer from the photodiodes to the storage nodes of said pixels; e) activating the first sample-and-hold devices of each read channel, such that a sample of the reference voltage is present on the first input of each subtractor block; f) for each pixel column, and after charge transfer to the corresponding storage nodes: f1) successively activate the line selection transistors, f2) in correspondence with each activation of a line selection transistor, activate the second sampling -blocker of the read channel corresponding to said column, then f3) driving the analog-to-digital converter of said read channel, whereby a second digital image called integration image (IIM) is acquired line by line; no reset of photodiodes or storage nodes being performed between operations b) and f3).
    [6" id="c-fr-0006]
    An image sensor according to claim 5 wherein said controller is also configured to repeat operations d) to f3) one or more times, depending on a signal level (IMAX) of said at least one integration image. , before performing again operations a) to c), so that one or more integration images (IIM, IIM2, IIM3) are acquired for the same black image.
    [7" id="c-fr-0007]
    7. An image sensor according to one of claims 5 or 6 also comprising a digital image processor (DIP) configured to subtract from said or said integration image the corresponding black image.
    [8" id="c-fr-0008]
    The image sensor according to one of the preceding claims wherein said pixels comprise a photodiode reset transistor (T2) and a separate storage node reset transistor (T3).
    [9" id="c-fr-0009]
    The image sensor according to one of the preceding claims, wherein each pixel also comprises a photodiode reset transistor (T2) for selectively connecting the photodiode to a source of reset voltage (Vrst).
    [10" id="c-fr-0010]
    10. A method of acquiring images by means of an active pixel image sensor comprising a matrix (M) of pixels (P) organized by rows and columns and a read circuit (CL) comprising a path for each column of pixels, in which each pixel comprises: a photodiode (PPD), a storage node (FD) of the electric charges generated by said photodiode when it is illuminated, a transistor of transfer (T 1) to selectively allow the transfer of said electrical charges from the photodiode to the storage node, o at least one reset transistor (T2, T3) for selectively connecting the photodiode and the storage node to a source (VSrst) of reset voltage (VRSt), o a line selection transistor (T4) for selectively connecting said pixel to the read channel of the column to which it belongs, and o a transistor (T5) connected to a voltage follower for transferring erase at said read path the voltage across the storage node through said line select transistor; the process comprising the following steps; A) enabling the reset transistor (s) of all the pixels to reset their photodiodes and their storage nodes; B) sampling a reference voltage, generated inside said read circuit and of value substantially equal to the reset voltage of the pixels of the matrix; C) for each pixel column, and after resetting the corresponding storage nodes: successively activate the line selection transistors, sample the voltage at the terminals of the storage node of the corresponding pixel, subtract the reference voltage sampled during the step B) of the sampled voltage and digitize the result, so as to acquire line by line a first digital image called black image; and then: D) activating the transfer transistors of all the pixels, so as to perform a global charge transfer from the photodiodes to the storage nodes of said pixels; E) sampling a reference voltage, generated inside said read circuit and of value substantially equal to the reset voltage of the pixels of the matrix; F) for each pixel column, and after charge transfer to the corresponding storage nodes: successively activate the line selection transistors, sample the voltage across the storage node of the corresponding pixel, subtract the sampled reference voltage when step E) of the voltage thus sampled and digitize the result, so as to acquire line by line a second digital image called integration image.
    [11" id="c-fr-0011]
    11. The method of claim 10 wherein steps D) to F) are repeated one or more times, depending on a signal level of said integration image or images, before performing again said steps A) to C), so that one or more integration images (IIM, IIM2, IIM3) are acquired for the same black image.
    [12" id="c-fr-0012]
    12. Method according to one of claims 10 or 11 also comprising a step of subtracting from said or said integration image the corresponding black image.
    类似技术:
    公开号 | 公开日 | 专利标题
    EP3122035B1|2019-10-16|Active pixel image sensor with operation in global shutter mode, reset noise subtraction and non-destructive reading
    EP2398230B1|2017-05-03|Linear image sensor using CMOS technology
    KR102212100B1|2021-02-03|Split-gate conditional-reset image sensor
    EP2567541B1|2019-12-25|Image sensor having a sampler array
    EP1265291A1|2002-12-11|CMOS image sensor and method for driving a CMOS image sensor with increased dynamic range
    EP2770731A1|2014-08-27|Image pickup device and camera system
    US7420154B2|2008-09-02|Pixel circuit with non-destructive readout circuit and methods of operation thereof
    JP2011082852A|2011-04-21|Solid-state imaging apparatus
    FR2973162A1|2012-09-28|VERY HIGH DYNAMIC IMAGE SENSOR
    FR2867936A1|2005-09-23|Image sensor, for digital camera, has averaging circuit carrying out averaging operations with reset charges and image charges to generate averaged reset and image charges, where average operations are performed in analog domain
    FR3042912A1|2017-04-28|IMAGE SENSOR WITH A HIGH DYNAMIC RANGE
    US9386240B1|2016-07-05|Compensation for dual conversion gain high dynamic range sensor
    US20090321799A1|2009-12-31|Method and apparatus for increasing conversion gain in imagers
    EP3332548B1|2020-04-15|Method for controlling an active pixel image sensor
    EP2408194B1|2014-12-24|Image sensor using CMOS technology with high video rate
    FR2906081A1|2008-03-21|CMOS LINEAR IMAGE SENSOR WITH CHARGE TRANSFER TYPE OPERATION
    FR2957475A1|2011-09-16|CMOS TYPE IMAGE SENSOR HAVING EXTENDED LINEAR DYNAMIC
    EP1290869B1|2012-08-01|Method for converting an analog signal into a digital signal and electromagnetic radiation sensor using same
    EP2486724B1|2015-09-16|Method for reducing noise in the image signal of an image sensor
    EP3026890B1|2019-06-05|Control of the integration time in a photosensitive device
    WO2014195622A1|2014-12-11|Cmos sensor with standard photosites
    EP3629573A1|2020-04-01|Quick image sensor with pixel binning
    EP1265290A1|2002-12-11|Method for driving a CMOS image sensor with increased sensitivity
    FR3064869A1|2018-10-05|IMAGE SENSOR
    EP2664132A1|2013-11-20|Cmos linear image sensor with motion-blur compensation
    同族专利:
    公开号 | 公开日
    EP3122035B1|2019-10-16|
    US20170026595A1|2017-01-26|
    US9781364B2|2017-10-03|
    ES2765877T3|2020-06-11|
    FR3039319B1|2018-06-15|
    EP3122035A1|2017-01-25|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US5818898A|1995-11-07|1998-10-06|Kabushiki Kaisha Toshiba|X-ray imaging apparatus using X-ray planar detector|
    US20030076431A1|2001-10-24|2003-04-24|Krymski Alexander I.|Image sensor with pixels having multiple capacitive storage elements|
    US20070045681A1|2005-08-30|2007-03-01|Micron Technology, Inc.|Efficient charge transferring in CMOS imagers|
    WO2010066850A1|2008-12-12|2010-06-17|E2V Semiconductors|Image sensor with double charge transfer for large dynamic range and method of reading|
    FR2939950A1|2008-12-12|2010-06-18|E2V Semiconductors|Image capturing matrix pixels controlling method for image sensor, involves controlling conduction of reinitialization transistor of node for selected line transistors by application of gate potential, and sampling potential of conductor|
    WO2012088338A2|2010-12-21|2012-06-28|Sionyx, Inc.|Photodetecting imager devices having correlated double sampling and associated methods|
    WO2012105259A1|2011-02-04|2012-08-09|パナソニック株式会社|Solid-state image capture device and method of driving same|
    GB2492387B|2011-06-30|2017-07-19|Cmosis Nv|Pixel array with individual exposure control for a pixel or pixel region|
    US9191598B2|2011-08-09|2015-11-17|Altasens, Inc.|Front-end pixel fixed pattern noise correction in imaging arrays having wide dynamic range|
    US9094612B2|2012-09-25|2015-07-28|Semiconductor Components Industries, Llc|Back side illuminated global shutter image sensors with back side charge storage|
    GB2516971A|2013-08-09|2015-02-11|St Microelectronics Res & Dev|A Pixel|
    JP6377947B2|2014-04-21|2018-08-22|ルネサスエレクトロニクス株式会社|Solid-state imaging device and electronic device|
    US9602750B2|2014-11-25|2017-03-21|Semiconductor Components Industries, Llc|Image sensor pixels having built-in variable gain feedback amplifier circuitry|
    US9502457B2|2015-01-29|2016-11-22|Semiconductor Components Industries, Llc|Global shutter image sensor pixels having centralized charge storage regions|
    US9843797B2|2015-06-11|2017-12-12|Semiconductor Components Industries, Llc|Imaging systems having column readout circuitry with test data injection capabilities|US9887225B2|2016-05-27|2018-02-06|Taiwan Semiconductor Manufacturing Co., Ltd.|Pixel with spacer layer covering photodiode|
    CN108632546B|2017-03-17|2021-06-04|豪威芯仑传感器有限公司|Pixel acquisition circuit, optical flow sensor and image acquisition system|
    CN109244088B|2017-07-10|2022-02-01|中芯国际集成电路制造有限公司|Semiconductor device, preparation method thereof and electronic device|
    CN108827461B|2018-04-25|2019-05-03|上海芯仑光电科技有限公司|Pixel Acquisition Circuit and light stream sensor|
    法律状态:
    2016-06-28| PLFP| Fee payment|Year of fee payment: 2 |
    2017-01-27| PLSC| Search report ready|Effective date: 20170127 |
    2017-06-28| PLFP| Fee payment|Year of fee payment: 3 |
    2018-06-28| PLFP| Fee payment|Year of fee payment: 4 |
    2020-06-25| PLFP| Fee payment|Year of fee payment: 6 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1557073A|FR3039319B1|2015-07-24|2015-07-24|ACTIVE PIXEL IMAGE SENSOR WITH GLOBAL SHUTTER MODE OPERATION, SUBTRACTION OF RESET NOISE AND NON-DESTRUCTIVE READING|
    FR1557073|2015-07-24|FR1557073A| FR3039319B1|2015-07-24|2015-07-24|ACTIVE PIXEL IMAGE SENSOR WITH GLOBAL SHUTTER MODE OPERATION, SUBTRACTION OF RESET NOISE AND NON-DESTRUCTIVE READING|
    US15/216,496| US9781364B2|2015-07-24|2016-07-21|Active pixel image sensor operating in global shutter mode, subtraction of the reset noise and non-destructive read|
    EP16180798.7A| EP3122035B1|2015-07-24|2016-07-22|Active pixel image sensor with operation in global shutter mode, reset noise subtraction and non-destructive reading|
    ES16180798T| ES2765877T3|2015-07-24|2016-07-22|Image sensor with active pixels with global shutter mode operation, reset noise subtraction and non-destructive readout|
    [返回顶部]