• 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:
    The present invention relates to a method of identifying a sign on an image of a document capable of being deformed, comprising: - an acquisition (E1) of said digital image of said document; a determination (E2) in the acquired image of at least one candidate sign region using an image segmentation algorithm, for each candidate sign region, a calculation (E3) of a signature comprising information relating to the location in the acquired image of said candidate sign region and a region descriptor relating to local characteristics of the image in said region, - an identification (E4) of a sign on the image of the document from the calculated signatures jointly comprising a comparison (E41) of the computed signatures with reference signatures relating to document pattern sign regions, said comparison being made according to a geometric deformation model of said document, and an estimate (E42) as a function of said comparison of said geometric deformation pattern.
    公开号:FR3027136A1
    申请号:FR1459742
    申请日:2014-10-10
    公开日:2016-04-15
    发明作者:Alain Rouh；Jean Beaudet；Laurent Rostaing
    申请人:Morpho SA；
    IPC主号:
    专利说明:

    [0001] FIELD OF THE INVENTION The present invention relates to the field of the detection and identification of signs. More specifically, it relates to a method of identifying at least one sign on one or more images of an optionally distorted document, in particular on a game ticket such as a lottery ticket.
    [0002] STATE OF THE ART Many existing procedures may use documents comprising checkboxes or checkboxes to interrogate a person, for example to make him fill out a form, answer examination questions, or to enable him to participate in a game such as a lottery. Methods have thus been proposed to detect and identify the position and content of signs, such as boxes, on a document submitted by a user to an automated reading. Many of these methods, such as those disclosed by US Patent Application Nos. 5,140,139 or US2008 / 0311551, utilize OCR (Optical Character Recognition) or OMR (Optical Mark Recognition) techniques to recover the position of a box in the document from a document template and determine the contents of such a box. Such methods are capable of taking into account a certain hazard in positioning the submitted document against the known document model, such as rotation or enlargement. These methods are sensitive against possible deformations of the document submitted, such as folding or wrinkling, which are likely to vary the relative position of the boxes to be detected relative to each other. Systems have thus been proposed comprising a device enabling the mechanical degreasing of the document before its treatment. Such devices are effective but are generally complex, cumbersome and significantly increase the cost of the reading systems. They can also complicate the use of the system, for example by introducing a risk of paper jam in the steamer, and do not allow to capture an image of the document remotely, without inserting it into the reading system. In order to overcome these drawbacks, certain methods, such as that described in the patent application EP 2713315, propose to detect the signs of a document by template matching. Images of small signs such as a box or a portion of a box, prerecorded from a document template, are searched for in an image of the submitted document. The resemblance of an area of the document image with prerecorded sign images is estimated by a correlation calculation. Such methods are less sensitive than the above-mentioned processes with no mechanical wrinkle-free deformation affecting the document as a whole, such as folding, but are sensitive to variations in the result of correlation calculations such as differences in brightness, contrast or deformations changing the appearance of a box like crinkle or perspective effects. Other methods, such as that described in patent FR 2952218, propose to perform a degreasing of the document submitted by software, in order to determine a virtual image of the deformed document. Such methods nevertheless require a high computing power and may require a more complex system than a simple camera, for example in order to project on the crumpled document the patterns required for the calculation of software degreasing.
    [0003] There is therefore a need for a method for identifying signs on a wrinkled document, using a simple imaging device comprising neither mechanical unwinding nor projection system, without being sensitive to variations. bright and local deformations 5 suffered by the document. PRESENTATION OF THE INVENTION According to a first aspect, the present invention proposes a method of identifying at least one sign on at least one image of a document capable of being deformed, said method being implemented by a device of FIG. data processing capable of being connected to a first storage device and characterized in that it comprises steps of: - acquiring said at least one digital image of said document; determination in the acquired image of at least one sub-part of the acquired image, called candidate sign region, using an image segmentation algorithm, for each candidate sign region, calculation a signature 20 comprising information relating to the location in the acquired image of said candidate sign region and a region descriptor relating to local characteristics of the image in said region, - identification of at least one sign on said at least one image of said document from the calculated signatures, said identifying step further comprising a comparison of the calculated signatures with reference signatures relating to document pattern sign regions stored in the first storage device, said comparison being made according to a geometric deformation model of said document, and an estimate according to said comparison of said model of d geometric formation.
    [0004] Such an implementation makes it possible to locate and identify in the document signs from an image that can possibly be taken at a distance. Jointly performing the comparison of the signatures and the determination of the deformation model makes it possible to obtain a model that is well representative of the deformations of the document but also to discard the candidate sign regions which are not coherent with this model and therefore clearly not corresponding to a sign. to identify. The use of a complex deformation model makes it possible not to limit the method to taking into account simple deformations of the document, such as perspective deformations, and makes it possible to identify signs despite any wrinkling of the document, without having to degrease the document neither mechanically nor by software.
    [0005] According to one embodiment, for each candidate sign region: said comparison step may comprise: a comparison of the signature calculated for said region with said reference signatures stored in said first storage device; mapping the calculated signature to a reference signature based on said comparison and a geometric deformation model of the current document stored in a second storage device connected to said data processing device, so as to identify said region candidate sign, and said estimation step may comprise: an estimation of a new model of geometric deformation of said document from said current geometrical deformation model and said correspondence, a storage in the second device storing said new deformation model as a deformation model n current.
    [0006] Such a joint and iterative determination of the correspondences between regions and of a deformation model of the document makes it possible to minimize the risks of bad association by updating the model after each new mapping.
    [0007] Such a document may be a game ticket or an identity document. Such a sign to be identified may be a geometric figure, a character, a group of characters or a graphic element.
    [0008] Such a sign to be identified may also be a pattern delimited by a closed contour. Such a sign to be identified may in particular be a box, a circle, a star or an alphanumeric character or a specific reason for a country issuing the identity document. Thus, such an implementation makes it possible to automatically identify the boxes of a game ticket or the characters locating a particular piece of information in the document processed, such as an identity document, so as to read it in a automatic despite its deformations. Alternatively, a signature may further include scale information.
    [0009] When a similar pattern is repeated with different sizes in a document, this variant avoids a bad association between two regions of different sizes.
    [0010] A descriptor of a sign region may be relative to a contour and / or content information of said sign region. Such a descriptor makes it possible to take advantage of both the shape of the outline and the content of a region in order to identify it with respect to a reference document, for example taking into account the numbers appearing in the background of the boxes of boxes. a game ticket to differentiate them from each other. The use of such descriptors makes it possible to make the candidate sign region association process to a sign region of a document model relatively independent of the picture image's shooting conditions or the deformations of the document image. here, while helping to limit the number of erroneous associations.
    [0011] Such a deformation model can be determined using an inverse distance weighting or spline interpolation algorithm. Such algorithms can efficiently interpolate a deformation model of the document from the determined associations for a limited amount of sign region pairs. According to a second aspect, the invention relates to a computer program product comprising code instructions for executing an identification method according to the first aspect when this program is executed by a processor. According to a third aspect, the invention relates to a device 5 for identifying at least one sign on at least one image of a document liable to be deformed, characterized in that it comprises a data processing device capable of to be connected to a first storage device comprising: a module for acquiring said at least one digital image of said document; a determination module in the acquired image of at least one sub-part of the acquired image, called the candidate sign region, using an image segmentation algorithm; a calculation module; a signature for each candidate sign region comprising information relating to the location in the acquired image of said candidate sign region and a region descriptor relating to local characteristics of the image in said region, a module of identifying at least one sign on said at least one image of said document from the calculated signatures, said identification module comprising a computed signature comparison module with reference signatures relating to model sign regions of documents stored in said first storage device, according to a geometric deformation model of said document, and a module for joint estimation of said geometric deformation model according to said comparison. Such computer program product and identification device have the same advantages as those mentioned for the method according to the first aspect.
    [0012] Other features and advantages of the present invention will be apparent from the following description of a preferred embodiment. This description will be given with reference to the accompanying drawings in which: - Figure 1 shows an identification system according to one embodiment of the invention; FIG. 2 illustrates an exemplary lottery bulletin for which the method according to the invention is implemented; FIGS. 3a, 3b and 3c illustrate the different types of physical deformations that can be experienced by a document for which the method according to the invention is implemented; FIG. 4 is a diagram schematizing an implementation of a method for identifying a sign according to the invention; FIGS. 5a and 5b illustrate an example of a result of a segmentation algorithm that can be implemented by the method according to the invention; FIG. 6 illustrates another example of a result of a segmentation algorithm of boxes that can be implemented by the method according to the invention; FIG. 7 illustrates an example of a sign region descriptor comprising at least one local gradient orientation histogram in said region according to an embodiment of the invention; FIG. 8 illustrates the impact on the precision of description of a form of the number of Fourier transform coefficients used for the definition of a shape descriptor according to an embodiment of the invention. invention; DETAILED DESCRIPTION The present invention relates to an implementation of a method of identifying at least one sign on at least one image of a document capable of being deformed 1 by a processing device 2 included in a device identification 3 as shown in Figure 1 10 may be connected to a first storage device 4. Such documents are for example identity documents such as passports, identity cards or driver's license, forms Examination MCQs or game tickets, including lottery ballots including checkboxes. The document to be processed comprises at least one sign to identify. Such a sign may be a geometric figure such as a box, a circle or a star, or a character such as an alphanumeric character, or a group of characters, or even a graphic element such as a specific pattern. from a country issuing an identity document.
    [0013] The example of a lottery bulletin is illustrated in FIG. 2. Such a ballot may comprise decor elements 5, positioning blocks 6, grids 7, boxes 8 and a game code 9. One will hear per box a pattern having any closed outline defining a space to be filled by a user. Such an outline may be marked by a line or a difference in hue between the bottom of the box and the rest of the ballot. In the example shown in Figure 2, some boxes have a square closed outline and other boxes have a star-shaped outline. Boxes may also have content such as a number as illustrated in Figure 2.
    [0014] In the case of a document such as a passport, the signs to be identified may be alphanumeric characters or strings consisting of letters, numbers or special characters such as dashes, or a graphic element. document as a specific reason for the issuing country. The identification in such a document, by comparison with reference signatures, of signs such as recurring character sequences from one document to another, such as the labels (name, first name ...) of the fields of the document or 10 still birthplaces, allows to identify these signs with greater robustness than by traditional OCR methods. Such a document may exhibit physical deformations of several types as illustrated in FIGS. 3a to 3c resulting from the manipulation of the document by the user. Such deformations may for example be wrinkling as illustrated in FIG. 3a, folds as illustrated in FIG. 3b or curvatures as illustrated in FIG. 3c.
    [0015] The document liable to be deformed conforms to a predefined document template. Such a document model defines a geometric disposition of the component signs, that is to say the theoretical position of the signs, characters or boxes constituting it when the document is not physically deformed.
    [0016] FIG. 1 illustrates an identification device configured to identify at least one sign on at least one image of such a document by implementing an identification method described with reference to FIG. 4. This method proposes to identify these signs on the at least one image of the deformable document from an acquired image of the wrinkled document, without the need for wrinkling, and from a prerecorded document template corresponding to the wrinkled document. For this purpose, the identification device comprises an acquisition module 10 configured to acquire a digital image of said document during an acquisition step E1. Such an acquisition module may comprise an image pickup apparatus such as as a camera, a camera or a scanner. This acquisition module is configured to acquire a digital image of the document, for example in the form of a matrix of pixels, reproducing any physical geometric deformations of the document. Such an image may also include geometric deformations produced by the perspective projection performed by the acquisition module. Such an acquisition only requires presenting the document to be processed in the field of view of the acquisition module, and does not require that the document be put in contact with the identification device, for example by being inserted into the document. device or lying flat on an imaging surface.
    [0017] The identification device further comprises a module for determining candidate sign regions. Such a candidate sign region determination module is configured to determine in the acquired image at least one sub-part of the acquired image. , said candidate sign region, using an image segmentation algorithm during a segmentation step E2. Such a segmentation algorithm can implement a thresholding of the pixels of the image of said document, for example according to their brightness, their gray level or their color. Such thresholding may be a simple thresholding comparing the pixels of the single threshold image and separating them into two groups of pixels. Such segmentation may be represented as a black and white image as shown in FIGS. 5a and 5b. The segmentation algorithm may then include a related component analysis to determine in the image regions of related pixels of the image having the same characteristics. For example, two pixels may be considered related if they are horizontally or vertically adjacent in the pixel matrix of the image for the 4-connexity, or also diagonally for the 8-connectivity. Such a segmentation algorithm can also evaluate several segmentation hypotheses, for example according to the method of the component-related tree, as described in the document "A Baillard, C Berger, ... R Levillain, N Widynski, 2007 , Algorithm for calculating the component tree with applications to shape recognition in satellite imagery, 21 ° GRETSI Conference (11-14). Each segmentation hypothesis corresponding to a node of the tree can then be evaluated according to various criteria relating to the object to be segmented, such as colorimetric criteria or geometric criteria, estimated for the group of pixels corresponding to the connected component. represented by the concerned node of the tree. It is thus possible to select the best hypothesis of segmentation according to the criteria considered, or to consider several hypotheses of segmentation. The regions of related pixels determined in the image by the segmentation algorithm constitute the candidate sign regions determined during the segmentation step E2.
    [0018] The segmentation algorithm can also determine the candidate sign regions using an MSER ("Maximally Stable Extreme Regions") algorithm, region growth segmentation, merging and decomposition, or detection of contour. The segmentation algorithm may also perform additional filtering operations in addition to the operations mentioned above, for example in order to enhance the contrast of the image before it is processed or to make a selection of related components. based on geometric criteria such as area or characteristic radius, or colorimetric criteria to remove related pixel regions of no interest for sign recognition.
    [0019] In the case of a document with boxes, it is possible to adapt the segmentation algorithm in order to try to detect candidate box regions corresponding to the boxes of the document according to their outlines but also according to their content. , that is to say as a function of the pixels situated within the boundaries of the outline of a box, as represented in FIG. 6. Such an approach can particularly be implemented when the inside of the boxes presents a particular feature, such as a distinctive fill color or a box number in the case of a lottery ticket. This will make the case detection more robust.
    [0020] Such a segmentation algorithm can lead to the selection of candidate sign regions which do not correspond to signs, for example to boxes or characters, but corresponding to other elements of the document such as decor elements. the positioning blocks of a lottery ticket or dirt deposited on the surface of the document. Candidate sign regions are not necessarily disjoint and some pixels or groups of pixels may belong to several candidate sign regions. The identification device also comprises a calculation module 12 configured to calculate, during a calculation step E3, a signature for each candidate sign region determined during the segmentation step E2. Such a signature of a region may include information relating to the location of said region in the acquired image as well as a region descriptor relating to local characteristics of the image in said region. Such a signature may further include scale information.
    [0021] The role of such a descriptor of a region is to describe the local characteristics of said region in a synthetic manner. Such a descriptor is preferably chosen so as to be invariant to changes of scale and to rotations and to be robust with regard to disturbances of the geometry or photometric characteristics of the image in the region such as a local affine transformation, a change of point of view, a noise addition, a modification of the illumination, etc. Such an invariance can be obtained by normalizing the image in the determined candidate regions or by normalizing the images. descriptors themselves, usually at the cost of loss of information. Such descriptors make it possible to effectively distinguish candidate regions from each other robustly to disturbances in the image acquisition of the document. These descriptors thus make it easy to put the candidate regions in correspondence with signs of a document model comprising various signs of known signatures stored in a storage device, with a low probability of error.
    [0022] The descriptor of a sign region may be relative to a contour and / or content information of said sign region. Such a descriptor may be relevant especially when the signs to be identified are boxes.
    [0023] The descriptor of a sign region may in particular be of the region descriptor type, that is to say taking into account information of the entire sign region including its content, and include at least one orientation histogram of local gradients in said region as shown in Figure 7. Such a histogram has in different directions the accumulation of local gradients in each of these directions 3027136 15 in the candidate region. The calculation of the descriptor of a region may be accompanied by a subdivision of this region as presented in the document "Lowe D. G., 1999, Object recognition from local scaleinvariant features, Proceedings of the International Conference on Computer Vision, vol. 2, p. 1150-1157. " The descriptor of a sign region may also be of form descriptor type and include a shape descriptor such as a Fourier descriptor of said form, as described in document Burger W. & Burge MJ, 2013, Fourier shape-descriptors, principles of digital image processing, 169-227, London Springer ", or a shape context descriptor descriptor as described in the document" Belongie S., Malik J. & Puzicha J. , 2002, Shape matching and object recognition using pattern contexts, Pattern Analysis and Machine Intelligence, IEEE Transactions on, 24 (4), 509-522. This type of descriptor can be interesting with identifiable signs with little ambiguity thanks to their shape like barcodes or one-dimensional OCR. Such a Fourier descriptor stores the coefficients of a Fourier transform describing in a more or less approximate manner the shape of the region. The more the number of stored coefficients is important, the more precisely the shape is described, as shown in Figure 8, at the cost of a loss of robustness. Specific descriptors may be employed when the signs to be identified are geometric figures characterized exclusively by their outline, such as positioning blocks of certain game tickets. The descriptors employed may then not take into account the content information of a candidate sign region and be mainly based on the outline of such a region.
    [0024] The identification device finally comprises an identification module 13 of at least one sign, configured to identify at least one sign on at least one image of the document during an identification step E4 from the signatures. calculated during the calculation step E3. To do this, the identification module comprises a comparison module 14 configured to compare, during a comparison step E41, signatures of candidate sign regions calculated during the calculation step E3 with stored reference signatures. in the first storage device. These reference signatures relate to sign regions of various document models. The comparison module attempts to find for each region of sign candidate a region of a model of the document having a signature close to that of the candidate region. If such mapping to a reference sign region is performed, the candidate sign region is considered identified and corresponding to the sign of the reference sign region. Such mapping is called "bitter". The identification module further comprises a model estimation module 15 configured to estimate, during an estimation step E42, a model of geometric deformation of said document as a function of said comparison performed by the comparison module 14. Such a deformation model may be: an affine model determining an affine application connecting the candidate sign regions and the reference sign regions of the document, such as translation, rotation or homothety. Such a model makes it possible to preserve dot alignments and distance ratios between document points, a homographic model determining a homographic application connecting the candidate sign regions and the reference sign regions of the document. Such a model makes it possible to match a plane of a plane surface seen by a camera with the plane of the same surface in another image, an interpolation model determined using a distance weighting algorithm. reverse, as proposed in the document "Franke R., 1982, Scattered data interpolation: 10 tests of some methods, mathematical of computation, 38 (157), 181-200", and / or spline interpolation. The use of such deformation models and the determination of the bitter according to such models makes it possible not to limit the method to the taking into account of simple deformations such as perspective deformations due to a bad positioning of the treated document, by example in a plane inclined with respect to the imaging plane. Thus, the method described makes it possible to take into account much more varied and much more complex deformations such as local deformations due to crumpling of the document. Such a model can be determined by successively implementing at least two algorithms and / or applications among the examples cited above. Different algorithms can thus be tested to determine the most suitable or the most suitable combination to describe the deformation of the document. Different algorithms can be adapted and recorded so that at least one algorithm corresponds to each type of deformation such as crumpling, folding, curvature, etc.
    [0025] The estimation step E42 and the comparison step E41 are carried out jointly so that the determination of the bitter is carried out as a function of the corresponding deformation model.
    [0026] According to a first variant, the deformation model is determined globally from all the determined combinations of bitters. The determination of the correct bitter can then be carried out so as to minimize a criterion defined according to the deformation model chosen.
    [0027] According to a second variant during the step of identifying signs E4, the deformation model can be determined iteratively. To do this, for each candidate sign region determined at the segmentation step E2, the comparison step E41 can then comprise: a comparison of the signature calculated for said region with said reference signatures stored in said first device storage, - matching the calculated signature with a reference signature according to said comparison and a geometric deformation pattern of the current document stored in a second storage device 16 connected to said data processing device 2 so as to identify said candidate sign region, and the estimation step E42 may comprise: an estimate of a new geometric deformation model of said document from said current geometric deformation pattern and said correspondence, storage in the second storage device of said new strain model as a model the deformation current.
    [0028] Each mapping of a candidate sign region to a reference sign region then results in an update of the current deformation pattern. Each mapping can be performed to minimize bad association indicators, such indicators being computable from the current deformation model as a function of the consistency between the association and the current model, and / or the similarity of the descriptors of the mapped regions. If, for a given candidate sign region, no reference sign region provides an indicator less than a predetermined acceptance threshold, the candidate sign region can be discarded. The possible associations for this candidate sign region then seem hardly compatible with the current deformation pattern, and this candidate sign region then likely corresponds to a region having no sign 15 and should not be mapped to a sign region. a document template. This can for example be the case for a lottery ticket if the sign region candidate matches a decoration element and not a box.
    [0029] In such iterative processing, the first candidate sign regions may, for example, be the sign regions located at the location of the positioning blocks of a reference model. The following candidate sign regions may be processed in a random order or step by step, depending on their proximity to a previously processed candidate sign region. Such an implementation thus makes it possible to identify in the image of a document capable of being deformed signs such as characters or boxes, from an image of the document possibly taken at a distance, without requiring a mechanical degreasing or software, without this identification being sensitive to the shooting conditions of the image of the document.
    权利要求:
    Claims (11)
    [0001]
    REVENDICATIONS1. A method of identifying at least one sign on at least one image of a deformable document (1), said method being implemented by a data processing device (2) capable of being connected to a first storage device (4) and characterized in that it comprises steps of: - acquisition (El) of said at least one digital image of said document; determination (E2) in the acquired image of at least one sub-part of the acquired image, called the candidate sign region, using an image segmentation algorithm, for each sign region candidate, calculation (E3) of a signature comprising information relating to the location in the acquired image of said candidate sign region and a region descriptor relating to local characteristics of the image in said region, - identification (E4 ) at least one sign on said at least one image of said document from the calculated signatures, said identifying step further comprising a comparison (E41) of the calculated signatures with reference signatures relating to regions of documents stored in the first storage device (4), said comparison being made according to a geometric deformation model of said document, and an estimate (E42) according to said comparison its said geometric deformation model.
    [0002]
    2. Identification method according to the preceding claim, wherein for each sign region sign: - said comparing step (E41) comprises: a comparison of the signature calculated for said region with said reference signatures stored in said first storage device; - a mapping of the calculated signature to a reference signature according to said comparison and a geometric deformation model of the current document stored in a second storage device (16) connected to said device data processing method (2), so as to identify said candidate sign region, - and said estimating step (E42) comprises: - an estimation of a new geometric deformation model of said document from said deformation model current geometry and said correspondence, - storage in the second storage device (16) of said new mod is deformation as a model of current deformation. 15
    [0003]
    3. Identification method according to one of the preceding claims, wherein said document is a game ticket or an identity document. 20
    [0004]
    4. Identification method according to one of the preceding claims, wherein a sign to be identified is a geometric figure, a character, a group of characters or a graphic element.
    [0005]
    5. Identification method according to one of the preceding claims, wherein a sign to be identified is a pattern delimited by a closed contour.
    [0006]
    6. Identification method according to one of claims 3 to 5, wherein a sign to be identified is a box, a circle, a star or an alphanumeric character or a specific pattern of a country issuing the identity document. . 3027136 23
    [0007]
    7. Identification method according to one of the preceding claims, wherein a signature further comprises a scale information. 5
    [0008]
    8. Identification method according to one of the preceding claims, wherein a descriptor of a sign region is relative to a contour information and / or content of said sign region.
    [0009]
    9. Identification method according to one of the preceding claims, wherein the deformation model is determined using an inverse distance weighting or spline interpolation algorithm.
    [0010]
    A computer program product comprising code instructions for performing an identification method according to any one of the preceding claims when the program is executed by a processor.
    [0011]
    11. Device for identifying (3) at least one sign on at least one image of a document liable to be deformed, characterized in that it comprises a data processing device (2) capable of be connected to a first storage device (4) comprising: - an acquisition module (10) of said at least one digital image of said document; a determination module (11) in the acquired image of at least one sub-part of the acquired image, called the candidate sign region, using an image segmentation algorithm, - a module computing (12) a signature for each candidate sign region comprising information relating to the location in the acquired image of said candidate sign region and a region descriptor relating to local characteristics of the image in said region An identification module (13) of at least one sign on said at least one image of said document from the calculated signatures, said identification module (13) comprising a comparison module (14) of the signatures calculated with reference signatures relating to document model sign regions stored in said first storage device, as a function of a geometric deformation model of said document, and an estimation module (15) of said document geometric deformation model according to said comparison.
    类似技术:
    公开号 | 公开日 | 专利标题
    EP3007105B1|2020-12-02|Method for identifying a sign on a deformed document
    US20210209335A1|2021-07-08|Object change detection and measurement using digital fingerprints
    US9036905B2|2015-05-19|Training classifiers for deblurring images
    CN108463828A|2018-08-28|The signature authentication of feature based
    JP2020030858A|2020-02-27|Image processing apparatus and image processing method
    WO2011048302A1|2011-04-28|Method, computer program, and device for hybrid tracking of real-time representations of objects in image sequence
    EP3382605A1|2018-10-03|Method for analysing a deformable structured document
    Benlamoudi et al.2017|Face antispoofing based on frame difference and multilevel representation
    CN108109164B|2021-04-13|Information processing method and electronic equipment
    WO2012079996A1|2012-06-21|Method of comparing images of irises by intelligent selection of textured zones
    WO2018042127A1|2018-03-08|Method and system for generating a surface signature
    EP3153991A1|2017-04-12|Method for analysing a content of at least one image of a deformed structured document
    WO2016062937A1|2016-04-28|Method of comparing digital images
    WO2010142897A2|2010-12-16|Method and device for calibrating an image sensor using a real-time system for following up objects in an image sequence
    FR3047832B1|2019-09-27|METHOD FOR DETERMINING A COLOR VALUE OF AN OBJECT IN AN IMAGE
    FR2986890A1|2013-08-16|METHOD FOR INSERTING A DIGITAL MARK IN AN IMAGE, AND CORRESPONDING METHOD FOR DETECTING A DIGITAL MARK IN AN IMAGE TO BE ANALYZED
    CN107240184B|2020-03-27|Method, device and equipment for identifying versions of plastic coins
    WO2012107696A1|2012-08-16|Methods, device and computer programs for recognising shapes, in real time, by means of an appliance including limited resources
    WO2019148442A1|2019-08-08|Template optimization method and device, electronic device and computer program product
    CN111784675A|2020-10-16|Method and device for processing article texture information, storage medium and electronic equipment
    CN113222963A|2021-08-06|Non-orthographic infrared monitoring sea surface oil spill area estimation method and system
    WO2020240124A1|2020-12-03|Shot-processing device
    EP3908968A1|2021-11-17|Method for processing digital images
    FR2998391A1|2014-05-23|Method for identification and/or authentication of individual, involves implementing identification and/or authentication process such that image in comparison zone is compared with image in positioning zone after registration of images
    FR2912827A1|2008-08-22|Biometric verification method for e.g. identification of animal, involves realizing association between characteristics of current and reference irises, and determining presence or absence of identity between irises
    同族专利:
    公开号 | 公开日
    US10025977B2|2018-07-17|
    CA2908210A1|2016-04-10|
    FR3027136B1|2017-11-10|
    US20160104039A1|2016-04-14|
    EP3007105A1|2016-04-13|
    EP3007105B1|2020-12-02|
    AU2015238872B2|2019-11-14|
    AU2015238872A1|2016-04-28|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US5940534A|1995-07-17|1999-08-17|Nippon Telegraph And Telephone Corporation|On-line handwritten character recognition using affine transformation to maximize overlapping of corresponding input and reference pattern strokes|
    WO2001005140A1|1999-07-09|2001-01-18|Hewlett-Packard Company|Curled surface imaging system|
    US8345979B2|2003-07-22|2013-01-01|Cognex Technology And Investment Corporation|Methods for finding and characterizing a deformed pattern in an image|
    WO2006130012A1|2005-06-02|2006-12-07|Lumex As|Method, system, digital camera and asic for geometric image transformation based on text line searching|
    US20140247470A1|2008-05-06|2014-09-04|Compulink Managment Center, Inc.|Camera-based document imaging|WO2021116427A1|2019-12-11|2021-06-17|Carrus Gaming|Method, device and computer program product for decoding a game ticket|
    WO2021123209A1|2019-12-18|2021-06-24|Carrus Gaming|Method for segmenting an input image showing a document containing structured information|US5140139A|1989-11-13|1992-08-18|Cognitronics Corporation|Preparing mark/read documents with markable boxes and locating the boxes from the document scan data|
    US6606421B1|2000-05-25|2003-08-12|Hewlett-Packard Development Company, L.P.|Geometric deformation correction method and system for dot pattern images|
    US20080311551A1|2005-08-23|2008-12-18|Mazer Corporation, The|Testing Scoring System and Method|
    WO2008033329A2|2006-09-15|2008-03-20|Sciammarella Cesar A|System and method for analyzing displacements and contouring of surfaces|
    US8055901B2|2009-03-17|2011-11-08|Scientific Games International, Inc.|Optical signature to enable image correction|
    FR2952218B1|2009-10-30|2012-03-30|Sagem Securite|METHOD AND DEVICE FOR OBTAINING AN IMAGE OF A DOCUMENT DEFROSSE FROM AN IMAGE OF THIS DOCUMENT WHEN IT IS FROZEN|
    US8582819B2|2011-11-18|2013-11-12|Xerox Corporation|Methods and systems for improving yield in wanted vehicle searches|
    US8588470B2|2011-11-18|2013-11-19|Xerox Corporation|Methods and systems for improved license plate signature matching by similarity learning on synthetic images|
    EP2713315B1|2012-09-26|2015-06-24|Lotsys|Method and device for reading a document|FR3064783B1|2017-03-30|2020-06-26|Peripheriques Et Materiels De Controle P M C|METHOD FOR IDENTIFYING THE CHECK BOXES OF A BET TICKET|
    FR3064782B1|2017-03-30|2019-04-05|Idemia Identity And Security|METHOD FOR ANALYZING A STRUCTURAL DOCUMENT THAT CAN BE DEFORMED|
    FR3081244B1|2018-05-17|2020-05-29|Idemia Identity & Security France|CHARACTER RECOGNITION PROCESS|
    US10311556B1|2018-07-02|2019-06-04|Capital One Services, Llc|Systems and methods for image data processing to remove deformations contained in documents|
    法律状态:
    2015-09-28| PLFP| Fee payment|Year of fee payment: 2 |
    2016-04-15| PLSC| Publication of the preliminary search report|Effective date: 20160415 |
    2016-09-21| PLFP| Fee payment|Year of fee payment: 3 |
    2017-09-21| PLFP| Fee payment|Year of fee payment: 4 |
    2018-09-19| PLFP| Fee payment|Year of fee payment: 5 |
    2019-09-19| PLFP| Fee payment|Year of fee payment: 6 |
    2020-09-17| PLFP| Fee payment|Year of fee payment: 7 |
    2020-12-18| CA| Change of address|Effective date: 20201110 |
    2020-12-18| CD| Change of name or company name|Owner name: IDEMIA IDENTITY & SECURITY FRANCE, FR Effective date: 20201110 |
    2021-09-22| PLFP| Fee payment|Year of fee payment: 8 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1459742A|FR3027136B1|2014-10-10|2014-10-10|METHOD OF IDENTIFYING A SIGN ON A DEFORMATION DOCUMENT|FR1459742A| FR3027136B1|2014-10-10|2014-10-10|METHOD OF IDENTIFYING A SIGN ON A DEFORMATION DOCUMENT|
    EP15188820.3A| EP3007105B1|2014-10-10|2015-10-07|Method for identifying a sign on a deformed document|
    CA2908210A| CA2908210A1|2014-10-10|2015-10-07|Identification process for a sign on a deformed image|
    US14/878,837| US10025977B2|2014-10-10|2015-10-08|Method for identifying a sign on a deformed document|
    AU2015238872A| AU2015238872B2|2014-10-10|2015-10-08|Method for identifying a sign on a deformed document|
    [返回顶部]