EP1864242A1 - Procede d'identification de visages a partir d'images de visage, dispositif et programme d'ordinateur correspondants - Google Patents
Procede d'identification de visages a partir d'images de visage, dispositif et programme d'ordinateur correspondantsInfo
- Publication number
- EP1864242A1 EP1864242A1 EP06708817A EP06708817A EP1864242A1 EP 1864242 A1 EP1864242 A1 EP 1864242A1 EP 06708817 A EP06708817 A EP 06708817A EP 06708817 A EP06708817 A EP 06708817A EP 1864242 A1 EP1864242 A1 EP 1864242A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- images
- learning
- face
- vectors
- representative
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/7715—Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods
Definitions
- a method of identifying faces from face images, a device and a corresponding computer program is a method of identifying faces from face images, a device and a corresponding computer program.
- the field of the invention is that of image processing and image sequences, such as video sequences. More specifically, the invention relates to a face recognition technique from a set of face images of one or more persons.
- the invention finds in particular, but not exclusively, applications in the fields of biometrics, video surveillance, or even video indexing, in which it is important to be able to recognize a face from a still image or a video sequence (for example to allow a recognized person to access a protected place).
- pp. 105 - 110 proposes, in the phase of recognition, to use either all the face images extracted from a video sequence, or a single key face image, namely that to which the face detector has assigned the highest confidence score.
- the invention particularly aims to overcome these disadvantages of the prior art.
- an object of the invention is to provide a face recognition technique from fixed face images or video sequences which has improved performance compared to the techniques of the prior art.
- an objective of the invention is to propose such a technique which gives satisfactory results, even when the face images to be treated are noisy, poorly framed, and / or have poor illumination conditions.
- Another objective of the invention is to propose such a technique that makes it possible to optimize the recognition capabilities of the statistical methods on which it is based.
- the invention also aims to provide such a technique that takes into account the quality of the face images used.
- Yet another object of the invention is to propose such a technique which is well suited to the recognition of several different persons, in the context of biometrics, video surveillance and video indexation applications for example.
- the invention also aims to provide such a technique that is simple and inexpensive to implement.
- a method of identifying at least one face from a group of at least two face images associated with the least one person comprising a learning phase and a recognition phase of said at least one face.
- the learning phase comprises at least a first step of filtering said images, implemented from a group of at least two learning face images associated with said at least one person, allowing selecting at least one training image representative of said face to be identified, the recognition phase using only said learning images selected during the learning phase.
- the filtering is performed from at least one of the thresholds belonging to the group comprising: a maximum distance (DRC 1nJ1x ) taking at least account of the membership of vectors associated with at least some of said images to a cloud constituted by said vectors; a maximum distance (DO 11181 ) between said vectors and vectors reconstructed after projection of said vectors on a space associated with said vector cloud.
- a maximum distance DRC 1nJ1x
- DO 11181 maximum distance
- the invention is based on a completely new and inventive approach to the recognition of faces, from still images or images extracted from video sequences. Indeed, the invention proposes to ignore all the face images available to identify the face of a person, but to filter images, to select only good quality images, it that is, those that are representative of the face to be identified (because the face is in frontal pose, that it is well framed, etc.).
- This filtering is performed by means of one or two filtering thresholds which are the centrally robust distance, or DRC, and / or the orthogonal distance, or DO.
- DRC centrally robust distance
- DO orthogonal distance
- the robust center distance takes into account the distance of a vector in the center of the vector cloud and the membership of the vector under consideration to that cloud.
- the orthogonal distance, or OD is the distance between a vector and the vector obtained after projection of the original vector into a space associated with the vector cloud, then reverse projection.
- the invention therefore proposes to select only a part of the training images, depending on their quality, so as to retain only those that are the most representative of facial images.
- At least one of said thresholds is determined from vectors associated with said training images.
- said learning phase also comprises a step of constructing a vector space for describing said at least one person from said representative training image (s).
- This construction step implements a technique belonging to the group comprising: a Principal Component Analysis technique; - a Linear Discriminant Analysis technique; a two-dimensional Principal Component Analysis technique; a Linear Discriminant Analysis technique with two dimensions.
- said recognition phase implements a second filtering step, from a group of at least two face images associated with said at least one person, called request images, and allows to select at least one representative request image of said face to identify, and at least one of said thresholds being determined during said learning phase, from vectors associated with learning face images.
- the recognition phase is thus complementary to the first filtering performed during the learning.
- at least one of said thresholds is determined during said recognition phase, from vectors associated with a set of images comprising at least two face images associated with said at least one person, called images. queries, and at least two training images representative of said face to be identified, selected during said learning phase, and said recognition phase implements a second filtering step, from said request images and makes it possible to select at least a representative query image of said face to be identified.
- the least noisy training images and the least noisy request images are selected, which greatly improves the face recognition performance compared with the prior art techniques.
- the request images are also filtered during the recognition phase by using the results of the learning phase, but this time in the form of training images representative of the face or faces to be identified, and no longer in the form of thresholds.
- said recognition phase also comprises a step of comparing projections, in a vector space of description of said at least one person constructed during said learning phase, of vectors associated with said at least one representative query image and with least one representative learning image selected during said learning phase, so as to identify said face.
- the notion of resemblance between two faces is then translated into a simple notion of spatial proximity between the projections of the faces in the space of description.
- this comparison step the projection of each of said vectors associated with each of said representative query images is compared with the projection of each of said vectors associated with each of said representative training images; determining, for each of said vectors associated with each of said representative query images, which is the nearest vector associated with one of said representative training images, and to which person, named designated person, it is associated; the face is identified as that of the person designated the greatest number of times.
- said first step of filtering said training images and / or said second step of filtering said request images implement said two thresholds, namely DO nJ3x and DRC 013x (calculated for all the images or sequence by sequence).
- the identification method of the invention also comprises a step of resizing said images, so that said images are all the same size. More precisely, in the presence of an image or a video sequence, a face detector makes it possible to extract a face image, of fixed size (all the images coming from this detector are thus of the same size). Then, during the processing of this face image of fixed size, it is a first resizing of the image during the filtering of the learning phase, so as to reduce its size, and thus avoid taking into account the details and remove the noise (for example, only one pixel out of three of the original image is retained). A second resizing of the image is also performed during the construction of the description space.
- said vectors associated with said images are obtained by concatenation of rows and / or columns of said images.
- said learning phase being implemented for learning images associated with at least two persons
- said thresholds associated with the learning images of each of said at least two persons are determined, and, during said recognition phase, said request images are filtered from said thresholds associated with each of said at least two persons.
- said learning phase being implemented for learning images associated with at least two persons
- said thresholds associated with the training images are determined. of the set of said at least two persons, and during said recognition phase, said request images are filtered from said thresholds associated with all of said at least two persons. Only two DCv thresholds 3x and DRC nJ3x are calculated for all the persons in the learning base.
- DRC 013x are determined after Robust Principal Component Analysis (RobPCA) applied to said vectors associated with said training images, to also determine a robust mean ⁇ associated with said vectors, and a projection matrix P constructed from eigenvectors of a robust covariance matrix associated with said vectors, and said thresholds are associated with the following distances:
- RobPCA Robust Principal Component Analysis
- X 1 is one of said vectors associated with said training images
- P dk is a matrix comprising the first k columns of said projection matrix P
- y y is the y ⁇ 1 " 6 element of a projection y t of said vector X 1 from said projection matrix and said robust average.
- the values of DCv 3x and DRC 013x are determined by analysis of the distribution of OD 1 and DRC 1 for the set of vectors x t .
- a nm is thus a matrix n lines, m columns); - lowercase letters (eg a, b) refer to vectors; for a matrix A nm, ⁇ refers to the ith row of A and tj refers to the element at the intersection of the ith row and / column A; det (A) is the determinant of matrix A;
- I n is the unit vector of dimension n;
- - diag (a lt ..., a n ) is the diagonal matrix with n rows, n columns, whose elements of the diagonal are a lt ..., a n ;
- a 1 is the transposed matrix of matrix A; a 'is the transpose of the vector a; Ivll is the Euclidean norm of vector v.
- the invention also relates to a system for identifying at least one face from a group of at least two face images associated with at least one person, said system comprising a learning device and a recognition device said at least one face.
- the learning device comprises means for determining at least one of the thresholds belonging to the group comprising: a maximum distance (DRC 1113x ) holding at least the membership of vectors associated with at least some of said images to a cloud constituted by said vectors; a maximum distance (DO 11181 ) between said vectors and vectors reconstructed after projection of said vectors on a space associated with said vector cloud; and first means for filtering said images, implemented from a group of at least two learning face images associated with said at least one person, for selecting at least one representative learning image of said face identifying, from at least one of said thresholds, the recognition device using only said learning images selected by said learning device.
- DRC 1113x holding at least the membership of vectors associated with at least some of said images to a cloud constituted by said vectors
- DO 11181 maximum distance between said vectors and vectors reconstructed after projection of said vectors on a space associated with said vector cloud
- the invention also relates to a device for learning a system for identifying at least one face, from a group of at least two learning face images associated with at least one person.
- a device for learning a system for identifying at least one face, from a group of at least two learning face images associated with at least one person comprises: means for analyzing said training images making it possible to determine, from vectors associated with said training images, at least one of the thresholds belonging to the group comprising: a maximum distance (DRC 11135 ) holding at least account of the belonging of said vectors to a cloud constituted by said vectors; a maximum distance (OD 1113x ) between said vectors and vectors reconstructed after projection of said vectors onto a space associated with said vector cloud; first means for filtering said training images from at least one of said thresholds so as to select at least one representative learning image of said face to be identified; means for constructing a vector space for describing said at least one person from said one or more representative learning image (s); so that only said learning images selected by said learning
- the invention also relates to a device for recognizing at least one face from a group of at least two face images associated with at least one person, called request images, said recognition device belonging to a control system. identifying said at least one face also comprising a learning device.
- Such a recognition device comprises: second means for filtering said request images from at least one threshold determined by said learning device, so as to select at least one request image representative of said face to be recognized; means for comparing projections, in a vector space of description of said at least one person constructed by said learning device, with vectors associated with said at least one representative query image and with at least one representative representative learning image selected by said learning device, so as to identify said face.
- said learning device comprising first filtering means implemented from a group of at least two learning face images associated with said at least one person, for selecting at least one representative learning image said face to be identified, said recognition device using only said learning images selected by said learning device.
- the invention further relates to a computer program comprising program code instructions for executing the learning phase of the at least one face identification method previously described when said program is executed by a processor.
- the invention relates to a computer program comprising program code instructions for performing the steps of the recognition phase of the method for identifying at least one face described above when said program is executed by a processor.
- FIG. 1 presents an example of facial images in frontal pose and well framed
- Figure 2 shows an example of face images which, unlike those of Figure 1, are noisy because poorly framed and / or non-frontal pose
- Figure 3 shows a block diagram of the face identification method of the invention
- FIG. 4 illustrates more precisely the processes performed during the learning phase of the method of FIG. 3, in a particular embodiment of the invention
- Figure 5 shows more schematically the learning phase of Figure 4
- FIG. 6 illustrates in more detail the processes performed during the recognition phase of the method illustrated in FIG. 3
- FIGS. 7 and 8 respectively show simplified diagrams of the learning and face recognition devices of the invention. 7. Description of an embodiment of the invention
- the general principle of the invention is based on the selection of a subset of images to be used, during the learning phase and / or the recognition phase, by using a robust Principal Components Analysis, or RobPCA.
- the invention makes it possible, in particular, to isolate the images of noisy faces during learning, and to deduce parameters for also filtering the face images during the recognition, which makes it possible to construct a description space without taking into account the noise, and to perform the recognition based on several examples images of faces also non-noisy.
- the proposed approach thus allows to considerably increase the recognition rates compared to an approach that would take into account all the images of the sequence.
- FIGS. 1 and 2 examples of face images are presented, on the one hand in frontal pose and well framed (FIG. 1), and on the other hand in non-frontal pose, or poorly framed, and therefore noisy ( Figure 2).
- the invention therefore makes it possible, in the presence of a set of face images, to select only face images of the type of those of FIG. 1, to perform the learning or recognition of faces, and to exclude all facial images of the type of those in Figure 2, which are considered noisy images.
- FIG. 3 presents a block diagram of the face identification method of the invention, which comprises three main steps: analysis 31 of the corpus of face images ((1 ! (1) , ... I M1 (1) ), ... (L 1 ⁇ , ...
- Each person 40 (also identified by the index j) is associated with a video sequence S ⁇ .
- a sequence S ⁇ can be acquired by filming the person 40 with the aid of a camera 41 for a determined duration.
- a face detector 42 By applying a face detector 42 to each of the images of the sequence S ⁇ (according to a technique well known to those skilled in the art which is not the subject of the present invention and will therefore not be described here in more detail) , a set of face images (I 1 *, ... I N ⁇ ) is extracted from the sequence S ⁇ .
- the invention then makes it possible to select only the images of faces which are in frontal pose and well framed and this, by analyzing the images of faces themselves.
- each image I, ⁇ is resized 43 so that all images have the same size: we then obtain a set of images (T 1 ⁇ , ... I ' N ⁇ ); a vector v ', ⁇ is associated with each of the images of faces I', red resized extracted from the sequence S ⁇ .
- the image I, ⁇ is considered as a representative face image, and is stored in the BA 51 learning base; the projection matrix P ⁇ , the robust average ⁇ ù) and the two decision thresholds ⁇ max t max DR C ⁇ ⁇ for each sequence S (J) are also saved in the training base BA 51.
- the projection matrix P ⁇ , the robust average ⁇ ù) and the two decision thresholds ⁇ max t max DR C ⁇ ⁇ for each sequence S (J) are also saved in the training base BA 51.
- all the images of faces extracted from all the learning video sequences S ⁇ are simultaneously considered.
- a single projection P, a single robust average ⁇ , a single decision threshold DO m ⁇ X and a single decision threshold DRC m ⁇ X are calculated during the learning phase.
- Learning Face The images are filtered using P, ⁇ , OD max etDRC max. An image / ', is filtered if:
- FIG. 5 more schematically shows these two constituent phases of the learning phase, namely the analysis of the video sequences of learning and the selection of the representative images ( ⁇ 7.1) and the construction of the description space ( ⁇ 7.2).
- a plurality of learning video sequences S 1 to S n are provided at the input, generally each associated with a distinct person that one wishes to be able to use. identify.
- a face detector 42 is applied to each of these sequences, in order to
- DRC nJ3x associated with the video sequence in question, and a projection method associated with the sequence (for example in the form of a projection matrix P and a robust average ⁇ associated with the images of the sequence);
- a request sequence S representing a person to be recognized (acquired for example by a camera of I ⁇ are first extracted from the sequence S with the aid of a detector 42.
- Each of these images I q can be considered as a request image and can therefore be used to identify the person sought, but as in learning, to increase the chances of identifying the person, we can chooses to select I q J for the identification
- it is chosen not to reuse the same procedure as in the learning phase ( ⁇ 7.1), since the acquisition of the video request is made under conditions that are generally less controlled (eg with the aid of a surveillance camera), and the assumption that the majority of the images extracted from the sequence are in frontal pose and well framed. 'is not always verified.
- two variants can be envisaged, depending on whether the selection of the image requests representative of the face to be identified is performed from the filter thresholds DCv x and DRC 013x calculated during the learning, or directly from the representative learning images.
- DO q ⁇ J) - ⁇ ⁇ - P $ ⁇
- and where p V is composed of k first column P®, and where y is the ith row of the matrix Y ° ⁇ projection matrix X 0 defined by Y n ⁇ k (X nx ⁇ - U ⁇ ') Pd x ⁇ -
- the image / is not selected if DOf> DO ⁇ x or DRC f> DRC ⁇ , V /. In other words, a face image is not taken into account during the recognition if the associated vector is considered aberrant by all the projections and the thresholds calculated for all the training video sequences.
- a single robust average ⁇ is considered during the learning process.
- a single decision threshold DO max and a single decision threshold DRC max the Face images queries are also filtered using P, ⁇ , max OD and max DRC during the recognition phase.
- a query / image is filtered (that is, considered aberrant) if:
- DO q and DRC q are respectively the orthogonal distance and the robust distance at the center of v '(where v' is the vector associated with / ', the resulting image of the resizing of I) using P and ⁇ .
- the representative learning images 53 from the learning phase are used.
- a filtering procedure similar to that used during the training is then applied to each of these sets by calculating the thresholds 013x and 013x that are associated with each of these sets.
- the face image I q is selected 80 if it is selected as the representative image by at least one of the filtering procedures applied (ie if for at least one of the sets there is a DO q ⁇ ⁇ DO m ⁇ K and DRC q ⁇ DRC max ).
- This selection procedure 80 of the representative query images can also be applied by inserting one or more images I q in the set of face images composed of all the representative learning images from the learning phase (all sequences of learning together). However, it is then desirable that the number of images I q inserted remains less than the number of representative training images. The filtering procedure is thus executed once and the image of faces I q is selected if it is retained as representative image.
- the identification of a request image q t is done in two steps. First, the representative query image q t is projected 81 in the description space 55 (calculated during learning) in the same manner as the images of the training database (step 54). Next, a search 82 of the nearest neighbor in the description space 55 is performed. It is a question of finding the projected vector among the projected vectors 56 corresponding to the images of the learning base which is the closest to the projected vector request. The request image q t is assigned to the same person as the person associated with the closest neighbor found. Each image q t and vote for a particular person, ie designates a person among those stored in the learning base. The results obtained for each of the representative query images of the set Q are then merged (83), and the face of the request sequence is finally recognized 84 as the person who has obtained the greatest number of votes.
- a set of face images is extracted I 1 , I 2 , ..., I n using an automatic face detector applied on each of the images of the video sequence.
- an automatic face detector applied on each of the images of the video sequence.
- the CFF detector described by C. Garcia and M. Delakis is used in "Convolutional Face-Finder: A Neural Architecture for Fast and Robust Face Detection", IEEE Trans. On Pattern Analysis and Machine Intelligence, 26 (11): 1408-1423, November 2004.
- These images are then resized to be all the same size (28x31). This resolution makes it possible to avoid taking into account the details in the images, because only the pose of the face (frontal or not) and its positioning in the image are important.
- the line y of the matrix corresponds to the vector associated with the image T 7 .
- This vector is built by concatenation of the lines of the image T 7 - after resizing.
- RobPCA allows to compute a robust mean ⁇ (vector of dimension d) and a robust covariance matrix C dxd considering only a subset of the vectors (namely vectors of dimension d associated with the images of faces. corresponds to a row of the matrix X). It also makes it possible to reduce the size of the images by projecting them into a smaller dimension space k (k ⁇ d) defined by the eigenvectors of the robust covariance matrix C. According to the RobPCA principle, and as detailed in the appendix 1 which forms an integral part of this description, if:
- the line i represents the projection of the line i of the matrix X. It is therefore the projection of the image I 1 .
- the calculation details of the matrix C and the robust average ⁇ by the RobPCA are given in appendix 1, which forms an integral part of the present description.
- two distances are calculated for each image /: it is the orthogonal distance (OD 1 ) and the robust distance in the center (DRC 1 ) . These two distances are calculated as follows: DO 1 , (2) where X 1 is the vector associated with /, (line i of the matrix X) and y t is the i th row of the matrix Y.
- the threshold of the orthogonal distance is on the other hand more difficult to fix because the distribution of the OD 1 is not known. The method proposed in the article by M. is used again
- Representative face images such as those of FIG. 1 are selected using the procedure presented here, from among a set of faces comprising images of the type of those of FIGS. 1 and 2.
- the proposed method therefore makes it possible to select only the images in frontal pose (figure 1) and to isolate the faces of profile or poorly framed (figure 2).
- the description space can be constructed by Principal Component Analysis (PCA).
- PCA Principal Component Analysis
- a learning base is first constructed in the form of a matrix.
- Each face image is resized so that all images are the same size.
- the chosen size is for example 63x57. This size can be the one obtained directly at the output of the face detector.
- Each image is then associated with a dimension vector 63x57 constructed by concatenation of the lines of the image.
- Each vector is then arranged in a row of the data matrix, denoted X md , where m is the number of images of selected faces and the size of the vectors (in this case d - 63x57).
- the description space is defined by the vectors of the matrix V which are also the eigenvectors of the covariance matrix of X.
- Y, ⁇ Qt V are saved for the recognition phase.
- the vector y t (the f line of the matrix Y) which is closest to it is found by calculating the distance between b t and all the vectors y t .
- the face image associated with b t is therefore recognized as being the person associated with the image represented by the closest neighbor found. Said that b t voted for the identified person. Once done for all b t , the face of the request sequence is finally recognized as the one with the highest number of votes.
- FIG. 7 shows the structure of a learning device of the invention, which comprises a memory M 61, and a processing unit 60 equipped with a ⁇ P processor, which is controlled by the computer program Pg.
- the processing unit 60 receives as input a set of training face images f 63, associated with one or more persons identified by the index j, from which the microprocessor ⁇ P realizes, according to the instructions of the program.
- Pg 62 a Robust Principal Component Analysis, or RobPCA.
- the ⁇ P processor of the processing unit 60 determines two filtering thresholds 68 of the images 63, called DCv 3x and DRC 013x , ie for each subset of images associated with each person.
- the data 68 also includes a robust mean ⁇ and a projection matrix P.
- the ⁇ P processor selects, from these thresholds, the mean ⁇ and the projection matrix P, and from the set 63 of images of one or more training images representative of the face or faces to be identified, (/ J- 0 ) * outputted from the processing unit 60.
- An analysis of ACP type also allows the ⁇ P processor to determine a description space, or model, 65 associated with each of the persons of index j, as well as a projection method 66 in this description space 65 of vectors associated with the training images, in the form of an average and of a projection matrix.
- FIG. 8 illustrates a simplified diagram of a face image recognition device of the invention. , which comprises a memory M 71, and a processing unit 70 equipped with a ⁇ P processor, which is controlled by the computer program Pg 72.
- the processing unit 70 receives as input: a set of face images requests 73, from which the recognition device must identify the face of a person; the filter thresholds DO- , and DRC 013x , as well as the robust mean ⁇ and the projection matrix P 68 delivered at the output of the learning device; the description space 65 constructed by the learning device; the projection method 66 used by the learning device; the vectors 67 associated with the representative training images and projected in the description space by the learning device.
- the ⁇ P processor of the processing unit 70 selects, according to the instructions of the program Pg 72, one or more request images representative of the face to be identified, among the set of request images 73, and from the DCv 3x and DRC thresholds. 013x , the robust average ⁇ and the projection matrix P 68. It then projects the vectors associated with these representative query images in the description space 65, by following the projection method 66. It then compares the vectors of Projected learning 67 and the projected motions vectors, to determine which face 74 is identified as the one shown on the query images 73.
- the thresholds 68 at the input of the recognition device are replaced by the representative training images 64, and the ⁇ P processor of the processing unit 70 performs a filtering identical to that performed by the device of FIG. learning from the set consisting of a request image 73 and the representative training images 64.
- RobPCA allows a principal component analysis, but considering only a subset of vectors. The idea is not to include in the analysis the noisy data which may affect the calculation of the mean and the covariance matrix (moments of order 1 and 2 known to be very sensitive to noise). For this, the RobPCA is based on the following property: a subset A is less noisy than another subset B if the vectors of A are less dispersed than those of B. In statistical terms, the set less noisy is the one whose determinant of the covariance matrix and the smallest.
- the Learning Base (BA) data is preprocessed using a PCA
- the Z matrix that is used in the following steps.
- the purpose of the second step is to find the least noisy vectors.
- a vector here refers to a line of the matrix Z, corresponds to a face image and is denoted z ,.
- h max ⁇ [an] [(n + Ic ⁇ x + 1) / 2] ⁇ , (4) where k ⁇ m ⁇ is the maximum number of principal components that will be used and a parameter included in 0.5 and 1. It represents the proportion of non noisy vectors. In the present case, this parameter corresponds to the proportion of the images of learning faces extracted from a sequence which are of good quality and which could be included in the learning base. The value of this parameter could therefore be set according to the acquisition conditions of the training sequences and the quality of the images of faces extracted from the sequences. The default value is 0.75.
- the method used to find the least noisy vectors is as follows. Firstly, for each vector z t , the noise level defined by:
- t MCD (z ] 'v) and s MCD (z ] ' v) are respectively the robust average and the robust standard deviation of the projection of all vectors in the direction defined by v. This is the mean and standard deviation of h projected values with the smallest variance.
- the outl noise level for all vectors is calculated and the h vectors with the smallest values of the noise level are considered.
- the indices of these vectors are stored in the set H 0 .
- H v orthogonal to v that contains h vectors In this case, all the vectors are projected on H v , which has the effect of reducing the size of the vectors by one, and the calculation of the sound effects is repeated. It should be noted here that this can possibly happen several times.
- K , ko ( z n, n ⁇ K TM) P 0 (r ⁇ , ko) , where ⁇ o (ri ⁇ ) is composed of the first k 0 columns of
- the covariance matrix of the Z * t vectors is estimated using a CDM estimator.
- the idea is to find the h vectors whose covariance matrix has the smallest determinant. Since it is virtually impossible to calculate the covariance matrices of all subsets containing h vectors, an approximate algorithm is used.
- This algorithm proceeds in 4 steps. 3.1 Let m 0 and C 0 respectively be the mean and the covariance matrix of the h vectors selected in step 2 (set H 0 ): (a) If det (C 0 )> 0 then calculate for each vector z * , the distance from Mahalanobis to m 0 :
- This procedure is therefore executed iteratively until the determinant of the covariance matrix of the selected vectors h no longer decreases.
- a weighted average and a weighted covariance matrix are calculated from m 4 and S 4 .
- S 4 is multiplied by one
- P 2 is a matrix kxk which contains the eigenvectors of S 5 and L 2 a diagonal matrix with the corresponding eigenvalues.
- the matrix P 2 is then projected in ffi by applying the inverse transforms of those applied throughout the preceding steps, which makes it possible to have the final matrix of the eigenvectors P dk .
- m 5 is projected in ffl, which allows to have ⁇ .
- the final covariance matrix C can be calculated using equation (1).
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
- Collating Specific Patterns (AREA)
- Image Processing (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
FR0503047A FR2884007A1 (fr) | 2005-03-29 | 2005-03-29 | Procede d'identification de visages a partir d'images de visage, dispositif et programme d'ordinateur correspondants |
PCT/EP2006/061109 WO2006103240A1 (fr) | 2005-03-29 | 2006-03-28 | Procédé d'identification de visages à partir d'images de visage, dispositif et programme d'ordinateur correspondants |
Publications (1)
Publication Number | Publication Date |
---|---|
EP1864242A1 true EP1864242A1 (fr) | 2007-12-12 |
Family
ID=35708620
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP06708817A Withdrawn EP1864242A1 (fr) | 2005-03-29 | 2006-03-28 | Procede d'identification de visages a partir d'images de visage, dispositif et programme d'ordinateur correspondants |
Country Status (6)
Country | Link |
---|---|
US (1) | US20080279424A1 (fr) |
EP (1) | EP1864242A1 (fr) |
JP (1) | JP2008537216A (fr) |
CN (1) | CN101171599A (fr) |
FR (1) | FR2884007A1 (fr) |
WO (1) | WO2006103240A1 (fr) |
Families Citing this family (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8085995B2 (en) | 2006-12-01 | 2011-12-27 | Google Inc. | Identifying images using face recognition |
FR2910668A1 (fr) * | 2006-12-21 | 2008-06-27 | France Telecom | Procede de classification d'une image d'objet et dispositif correspondant |
JP4986720B2 (ja) * | 2007-06-07 | 2012-07-25 | 株式会社ユニバーサルエンターテインメント | 個人識別データ登録装置 |
US8855360B2 (en) * | 2008-07-23 | 2014-10-07 | Qualcomm Technologies, Inc. | System and method for face tracking |
JP5524692B2 (ja) * | 2010-04-20 | 2014-06-18 | 富士フイルム株式会社 | 情報処理装置および方法ならびにプログラム |
JP5753966B2 (ja) * | 2010-08-05 | 2015-07-22 | パナソニックIpマネジメント株式会社 | 顔画像登録装置および方法 |
US8655027B1 (en) * | 2011-03-25 | 2014-02-18 | The United States of America, as represented by the Director, National Security Agency | Method of image-based user authentication |
US8965046B2 (en) | 2012-03-16 | 2015-02-24 | Qualcomm Technologies, Inc. | Method, apparatus, and manufacture for smiling face detection |
CN103870728B (zh) * | 2012-12-18 | 2018-06-12 | 富泰华工业(深圳)有限公司 | 控制系统、控制方法及电脑系统 |
US10002310B2 (en) * | 2014-04-29 | 2018-06-19 | At&T Intellectual Property I, L.P. | Method and apparatus for organizing media content |
KR102010378B1 (ko) * | 2014-09-24 | 2019-08-13 | 삼성전자주식회사 | 객체를 포함하는 영상의 특징을 추출하는 방법 및 장치 |
US9430694B2 (en) * | 2014-11-06 | 2016-08-30 | TCL Research America Inc. | Face recognition system and method |
US10839196B2 (en) * | 2015-09-22 | 2020-11-17 | ImageSleuth, Inc. | Surveillance and monitoring system that employs automated methods and subsystems that identify and characterize face tracks in video |
CN106557728B (zh) * | 2015-09-30 | 2019-06-18 | 佳能株式会社 | 查询图像处理和图像检索方法和装置以及监视系统 |
CN105678265B (zh) * | 2016-01-06 | 2019-08-20 | 广州洪森科技有限公司 | 基于流形学习的数据降维方法及装置 |
CN105760872B (zh) * | 2016-02-03 | 2019-06-11 | 苏州大学 | 一种基于鲁棒图像特征提取的识别方法及系统 |
KR102221118B1 (ko) * | 2016-02-16 | 2021-02-26 | 삼성전자주식회사 | 영상의 특징을 추출하여 객체를 인식하는 방법 |
CN106778818A (zh) * | 2016-11-24 | 2017-05-31 | 深圳明创自控技术有限公司 | 一种基于云计算的智能跟踪系统 |
CN107516105B (zh) * | 2017-07-20 | 2020-06-16 | 阿里巴巴集团控股有限公司 | 图像处理方法及装置 |
JP6997140B2 (ja) * | 2019-07-03 | 2022-01-17 | パナソニックi-PROセンシングソリューションズ株式会社 | 情報処理装置、判定方法、およびプログラム |
CN112069948A (zh) * | 2020-08-25 | 2020-12-11 | 辽宁工程技术大学 | 一种基于改进二维降维的人脸识别方法 |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5842194A (en) * | 1995-07-28 | 1998-11-24 | Mitsubishi Denki Kabushiki Kaisha | Method of recognizing images of faces or general images using fuzzy combination of multiple resolutions |
US6501857B1 (en) * | 1999-07-20 | 2002-12-31 | Craig Gotsman | Method and system for detecting and classifying objects in an image |
US6944319B1 (en) * | 1999-09-13 | 2005-09-13 | Microsoft Corporation | Pose-invariant face recognition system and process |
JP4161659B2 (ja) * | 2002-02-27 | 2008-10-08 | 日本電気株式会社 | 画像認識システム及びその認識方法並びにプログラム |
-
2005
- 2005-03-29 FR FR0503047A patent/FR2884007A1/fr active Pending
-
2006
- 2006-03-28 JP JP2008503505A patent/JP2008537216A/ja active Pending
- 2006-03-28 US US11/910,158 patent/US20080279424A1/en not_active Abandoned
- 2006-03-28 EP EP06708817A patent/EP1864242A1/fr not_active Withdrawn
- 2006-03-28 WO PCT/EP2006/061109 patent/WO2006103240A1/fr active Application Filing
- 2006-03-28 CN CNA2006800149452A patent/CN101171599A/zh active Pending
Non-Patent Citations (1)
Title |
---|
See references of WO2006103240A1 * |
Also Published As
Publication number | Publication date |
---|---|
JP2008537216A (ja) | 2008-09-11 |
FR2884007A1 (fr) | 2006-10-06 |
US20080279424A1 (en) | 2008-11-13 |
CN101171599A (zh) | 2008-04-30 |
WO2006103240A1 (fr) | 2006-10-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP1864242A1 (fr) | Procede d'identification de visages a partir d'images de visage, dispositif et programme d'ordinateur correspondants | |
Al-Allaf | Review of face detection systems based artificial neural networks algorithms | |
Nam et al. | Local decorrelation for improved pedestrian detection | |
EP3640843B1 (fr) | Procédé d'extraction de caractéristiques d'une empreinte digitale représentée par une image d'entrée | |
EP2321769B1 (fr) | Procédé de reconnaissance de formes et système mettant en oeuvre le procédé | |
WO2020050966A1 (fr) | Adaptation utilisateur pour authentification biométrique | |
EP1751689A1 (fr) | Procede pour la reconnaissance de visages, a analyse discriminante lineaire bidimensionnelle | |
WO2006103241A2 (fr) | Système et procédé de localisation de points d'intérêt dans une image d'objet mettant en œuvre un réseau de neurones | |
EP3018615B1 (fr) | Procede de comparaison de donnees ameliore | |
EP3620970B1 (fr) | Procédé d'extraction de caractéristiques d'une empreinte digitale représentée par une image d'entrée | |
EP3582141B1 (fr) | Procédé d'apprentissage de paramètres d'un réseau de neurones à convolution | |
EP2619713A1 (fr) | Procede de classification de donnees biometriques | |
FR3088467A1 (fr) | Procede de classification d'une image d'entree representative d'un trait biometrique au moyen d'un reseau de neurones a convolution | |
WO2010066992A1 (fr) | Procede d'identification ou d'autorisation, et systeme et module securise associes | |
Prasad et al. | Face recognition using PCA and feed forward neural networks | |
EP2517151B1 (fr) | Codage biometrique | |
WO2007077175A1 (fr) | Procede de classification d'images par reseaux neuroniques et classifieur d'images prealablement appris, dispositif et programme d'ordinateur correspondants | |
WO2008081152A2 (fr) | Procede et systeme de reconnaissance d'un objet dans une image | |
EP3929809A1 (fr) | Procédé de détection d'au moins un trait biométrique visible sur une image d entrée au moyen d'un réseau de neurones à convolution | |
FR3103045A1 (fr) | Procédé d’augmentation d’une base d’images d’apprentissage représentant une empreinte sur un arrière-plan au moyen d’un réseau antagoniste génératif | |
EP4292013A1 (fr) | Dispositif et procede de traitement de donnees videos pour detection du vivant | |
WO2024002618A1 (fr) | Procédé de reconnaissance biometrique | |
EP4163866A1 (fr) | Procédé, programme d'ordinateur et dispositif de traitement d'images par extraction de composante(s) principale(s) d'une représentation d'images | |
EP3825915A1 (fr) | Procede de classification d'une empreinte biometrique representee par une image d'entree | |
EP4099200A1 (fr) | Procede et dispositif d'identification et/ou d'authentification biometrique |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
17P | Request for examination filed |
Effective date: 20070921 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LI LT LU LV MC NL PL PT RO SE SI SK TR |
|
RIN1 | Information on inventor provided before grant (corrected) |
Inventor name: BERRANI, SID AHMED Inventor name: GARCIA, CHRISTOPHE |
|
RIN1 | Information on inventor provided before grant (corrected) |
Inventor name: BERRANI, SID AHMED Inventor name: GARCIA, CHRISTOPHE |
|
DAX | Request for extension of the european patent (deleted) | ||
17Q | First examination report despatched |
Effective date: 20081215 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN |
|
18D | Application deemed to be withdrawn |
Effective date: 20101001 |