44 results about "Eigenface" patented technology

A face recognition system and process for identifying a person depicted in an input image and their face pose. This system and process entails locating and extracting face regions belonging to known people from a set of model images, and determining the face pose for each of the face regions extracted. All the extracted face regions are preprocessed by normalizing, cropping, categorizing and finally abstracting them. More specifically, the images are normalized and cropped to show only a persons face, categorized according to the face pose of the depicted person's face by assigning them to one of a series of face pose ranges, and abstracted preferably via an eigenface approach. The preprocessed face images are preferably used to train a neural network ensemble having a first stage made up of a bank of face recognition neural networks each of which is dedicated to a particular pose range, and a second stage constituting a single fusing neural network that is used to combine the outputs from each of the first stage neural networks. Once trained, the input of a face region which has been extracted from an input image and preprocessed (i.e., normalized, cropped and abstracted) will cause just one of the output units of the fusing portion of the neural network ensemble to become active. The active output unit indicates either the identify of the person whose face was extracted from the input image and the associated face pose, or that the identity of the person is unknown to the system.

A face recognition system and process for identifying a person depicted in an input image and their face pose. This system and process entails locating and extracting face regions belonging to known people from a set of model images, and determining the face pose for each of the face regions extracted. All the extracted face regions are preprocessed by normalizing, cropping, categorizing and finally abstracting them. More specifically, the images are normalized and cropped to show only a persons face, categorized according to the face pose of the depicted person's face by assigning them to one of a series of face pose ranges, and abstracted preferably via an eigenface approach. The preprocessed face images are preferably used to train a neural network ensemble having a first stage made up of a bank of face recognition neural networks each of which is dedicated to a particular pose range, and a second stage constituting a single fusing neural network that is used to combine the outputs from each of the first stage neural networks. Once trained, the input of a face region which has been extracted from an input image and preprocessed (i.e., normalized, cropped and abstracted) will cause just one of the output units of the fusing portion of the neural network ensemble to become active. The active output unit indicates either the identify of the person whose face was extracted from the input image and the associated face pose, or that the identity of the person is unknown to the system.

A method and system for concatenating eigenvoice vector data with eigenface vector data to obtain a new basic eigenvector to more positively and accurately identify a person. More specifically, for a specific person, the data of an eigenface vector is concatenated with the data of an eigenvoice vector to form a single vector, and this single vector is compared to reference vectors also obtained by concatenting the data of eigenface vectors and eigenvoice vectors to obtain a single vector of each persons in a defined group of people to determine if a person that is a member of a defined group of people. Using this single eiginvector gives a higher success rate for identifying a person than an eigenface vector or an eigenvoice vector used separately or together as separate vectors.

The invention discloses a method for human face authentication. The method comprises the following steps: the fisher coefficient of a target to be authenticated is calculated with the fisher eigenface method based on principal component analysis; the similarity matching degree corresponding to the fisher coefficient is calculated by utilizing a lookup table classifier; the similarity matching degree and a preset threshold are compared to get the authentication result. The invention also discloses a device for human face authentication. The device comprises a coefficient computing module, which computes the fisher coefficient of a target to be authenticated; an output computing module, which computes the similarity matching degree corresponding to the fisher coefficient with a lookup table classifier; and an authenticating module, which compares the similarity matching degree and the preset threshold to get the authentication result. The method and the device for the human face authentication of the invention, by introducing the lookup table classifier with more powerful expressing ability than a single threshold, can perform better classification on training samples, thereby improving the recognition rate of human face authentication and reducing the mis-detection possibility.

A method for detecting and recognizing human face in vivo comprises the steps of: a video human face image is acquired; the face part of the video frame image is segmented to get the optical flow modulus sequence and direction sequence; three kinds of feature information are extracted from the sequence, and the extracted feature is trained to get an SVM classifier to judge whether it is a living face or not; according to the living face image detected by infrared image, the face part in the color image is segmented, the color face image is extracted by using PCA algorithm, and the collected face image is represented by one-dimensional vector, and the covariance matrix C of the vector is obtained; the eigenface space W is obtained from the covariance matrix C, and the eigenface space W is used as the input of the neural network; the neural network is optimized and trained; after the training, the face images to be recognized are input into the network for face recognition. In order to solve the cheating behavior of photographs in face recognition method, the difference between living face and image video face can be judged effectively.

In blendshape-based facial animation, two main approaches are used to create the key expressions: manual sculpting and statistically-based techniques. Hand-generated expressions have the advantage of being intuitively recognizable, thus allowing animators to use conventional keyframe control. However, they may cover only a fraction of the expression space, resulting in large reproduction animation errors. On the other hand, statistically-based techniques produce eigenfaces that give minimal reproduction errors but are visually non-intuitive. In the invention the applicants propose a technique to convert a given set of hand-generated key expressions into another set of so-called quasi-eigen faces. The resulting expressions resemble the original hand-generated expressions, but have expression space coverages more like those of statistically generated expression bases. The effectiveness of the proposed technique is demonstrated by applying it to hand-generated expressions.

The invention discloses a low-rank matrix and eigenface-based human face identification method, which relates to the technical fields of digital image processing, mode identification, computer vision, physiology and the like, and is used for solving the problem in human face identification based on static images or video images in various scenes. A low-rank matrix is applied to preprocessing of a human face picture based on a low-rank matrix concept, and the influence of changes of illumination, expressions and the like is reduced through low-rank processing on a training picture, so that the algorithm robustness and the identification accuracy are improved. According to the key points of the technical scheme, the method comprises the following steps of firstly acquiring a human face sample picture and establishing a sample library; secondly during a training stage, constructing an eigenvector space through operations of calculating a sample mean value, an eigenvalue, an eigenvector and the like, and projecting the eigenvector to obtain an eigenface; and finally during a test stage, performing PCA projection on a test sample to obtain an eigenvector, calculating a distance between the eigenvector and the eigenface, taking a shortest distance as an identification result, and outputting the identification result.

The invention provides a human face recognition method based on multiple-feature space sparse classifiers. The human face recognition method includes the following steps that original training samples, namely X1....XN, are projected onto an Eigenface feature space, a Laplacianface feature space and a Gabor feature space respectively to form a sub-dictionary OE, a sub-dictionary OL and a sub-dictionary OG; the genetic algorithm is used for carrying out joint optimization and training on the three sub-dictionaries to obtain a sub-dictionary NE, a sub-dictionary NL and a sub-dictionary NG; the sub-dictionary NE, the sub-dictionary NL and the sub-dictionary NG are used for training the sparse classifier SRCs. Each sparse classifier carries out sparse representation on a sample to be tested and obtains residual errors corresponding to the ith training sample, wherein the residual errors are RiE, RiL and RiG respectively and then, the mean value of the residual errors corresponding to the ith training sample are calculated. A category corresponding to the minimum value of the residual error mean value E[Ri] is the category which the human face sample to be tested belongs to. According to the human face recognition method based on the multiple-feature space sparse classifiers, the adopted dictionary training method can select a sample with the best separating capacity from each sub-dictionary, so that the human face recognition accuracy based on the sparse classifiers with the dictionaries is improved.

The present invention relates to a system and method for identifying of a pattern by comparing the pattern with two or more identified patterns. The system comprises an input unit for capturing a pattern for identification, a processing unit for determining eigenfaces corresponding to the captured pattern and the two or more identified patterns, and determining orientation vectors corresponding to each determined eigenface, and a comparison unit for comparing the determined orientation vector corresponding to the captured pattern with each of the determined orientation vectors corresponding to the identified patterns. The method comprises determining eigenvectors corresponding to each of the identified patterns, determining eigenfaces corresponding to each of the identified patterns and the pattern being identified, determining orientation vectors corresponding to each of the identified patterns and the pattern being identified, comparing an orientation vector corresponding to the pattern being identified with each of the orientation vectors corresponding to the identified patterns, and identifying the pattern. The present invention further provides a method of clustering a plurality of patterns into a predetermined number of clusters by using orientation vectors.

The invention discloses a face recognition method and device and an intelligent security and protection management system, and the method comprises the steps: calculating the feature face weight of ato-be-recognized face image, and then calculating the Euclidean distance between the feature face weight of the to-be-recognized face image and each value in a feature face weight vector of a trainingset, and if a certain Euclidean distance is less than or equal to a preset similar face threshold, judging that the face is the same face, if all Euclidean distances are greater than or equal to a preset non-face threshold, considering that the face is a non-face, and if all Euclidean distances are between the similar face threshold and the non-face threshold, considering that the face is a new face. Non-human faces and human faces without entrance guard permission can be effectively recognized, the intelligent security and protection management system can achieve the functions of remote monitoring, automatic alarming, entrance guard management, intelligent human face recognition, intelligent safety helmet recognition, safety early warning, video capture and the like, the working pressureof people is relieved, and the working behaviors of warehouse personnel are standardized.

The invention provides a joint training method of sub-dictionaries in multiple characteristic spaces and for face recognition. The joint training method of the sub-dictionaries in the multiple characteristic spaces and for face recognition comprises the steps that original training samples{X[1] to X[N]} are projected in the Eigenface characteristic space, the Laplacianface characteristic space and the Gabor characteristic space to form the sub-dictionary OE, the sub-dictionary OL and the sub-dictionary OG respectively, and joint optimization training is carried on the three sub-dictionaries by using a genetic algorithm to obtain the sub-dictionary NE, the sub-dictionary NL and the sub-dictionary NG. According to the joint training method of the sub-dictionaries in the multiple characteristic spaces and for face recognition, a sample, with the highest distinguishing capacity, in each sub-dictionary can be selected, and thus accuracy of face recognition on the basis of a sparse classifier using the sub-dictionaries can be improved.

The invention discloses an online passing system based on a face feature recognition technology. Comprising an acquisition module, an acquisition module, an extraction module, an initialization module, a calculation module, a matching module, a detection module and an optimization module. A feature face matching algorithm is adopted, so that face recognition with higher accuracy is realized; on-line/off-line living body detection service is adopted to identify whether a user is a real person or not, and cheating tools such as photos, videos and 3D molds are effectively resisted; the space and time complexity of the system is optimized by adopting a system optimization algorithm, the response time of the system is reduced, and the fluency of the system is improved. The face feature recognition system can prevent lawbreakers from cheating by using tools such as photos, videos and 3D molds of others, can effectively improve the face recognition accuracy and passing efficiency of an existing face recognition system, can be widely applied to the fields of financial industry, retail industry, security passing and the like, and the business requirements of identity verification, face attendance checking, gate passing and the like are met.

The invention provides a geological target recognition method based on face recognition, which is used for recognizing geological targets such as fractures and holes of different sizes and shapes. Themethod comprises the specific steps of loading the electric imaging logging data, preprocessing and processing the electric imaging logging data, generating dynamic and static images, etc., loading core data containing fractures or cavities, structured by standard geologic target feature faces; carrying out the feature extraction of training sample image; carrying out the feature extraction of test sample image; carrying out the distance function selection; carrying out the pattern matching of geological targets based on eigenface face recognition. Because the dynamic micro-resistivity imageis formed firstly and then the face recognition technology is used to recognize the geological targets of different sizes and shapes, the method has the characteristics of high accuracy and strong reliability, and can improve the recognition rate of the geological targets.

The embodiment of the invention relates to the technical field of target detection, in particular to a method for training a color spot detection model and a related device, and aims to reduce the interference of area features (such as background features and facial features) without color spots on a training model by intercepting images in an image sample set into candidate area images. Therefore, the accuracy of model training can be improved; resolution enhancement processing and size adjustment are performed on each candidate region image containing color spots to obtain a plurality of training images with consistent sizes, so that interference of the sizes on model learning can be effectively avoided, and the features of the color spots are enhanced; besides, the number of the color spots is introduced in the training process to restrain the positions of the color spots, the prediction positions of the color spots can be trained in the direction the same as the positions of the real color spots, and the model training precision is further improved; the improvement of the multi-training precision enables the detection precision of the color spot detection model obtained through the training of the method to be higher.

The invention discloses a face recognition method under an illumination change condition. The method comprises steps as follows: step one, acquiring a neutral network through training; step two, acquiring virtual face images of each person under target illumination conditions; step three, acquiring a simulation image of each person under each target illumination condition, and establishing N neutral networks; step four, acquiring simulation face images of each person under all target illumination conditions; step five, establishing a face training library comprising real face images and simulation face images; step six, matching to-be-tested images with images in the training library under corresponding illumination conditions. According to the face recognition method under the illumination change condition, eigenface features of all persons under rest view angle conditions are simulated through training of a radial basis function network on a single image set, recognized view angle invariance is realized to a certain extent, and the good recognition effect can be realized.

The invention relates to a face recognition method, which comprises the following steps of 1) elongating each human face image of a training set by one column as a vector; 2) adding up all the faces on the corresponding dimensions and then averaging them to obtain an average face; 3) subtracting that average face image from each image to obtain a data matrix of the difference image; 4) calculatingthat covariance matrix, and then decomposing the covariance matrix into eigenvalue, and obtaining a desired eigenvector (eigenface); 5) projecting that images of the train set and the test set onto these eigenvectors, and then finding the nearest neighbor of the training set for classification and recognition for each image of the test set. The method has the characteristics of accurate recognition, simple realization, safety and practicability.

The invention discloses a face recognition method and system of an SVM based on PCA and ReliefF, a storage medium and equipment, applied to the field of image recognition. The method comprises the steps: converting a face image into a preset dimension vector, and storing the preset dimension vector into a face vector set; calculating a vector accumulation average value to obtain an average image; calculating a difference value between the face image and the average image, calculating a feature vector of a covariance matrix according to the difference value, and forming a feature face space; respectively calculating the feature weight of each feature face space sample according to a ReliefF algorithm, and constructing a feature weight vector; constructing and solving an optimal solution according to a preset penalty parameter and the feature weight vector, obtaining a decision function, and determining a support vector machine; and training and verifying based on a support vector machine to obtain a face recognition model, and performing face recognition by using the face recognition model. According to the technical scheme, the defect that a single ReliefF algorithm cannot remove redundant features is overcome, the data dimension is effectively reduced, and SVM training is accelerated.

The invention discloses a face recognition method based on a multi-dimensional Taylor network. The construction process of the multi-dimensional Taylor network comprises the following steps: performing Taylor expansion on an activation function of a neural network structure, performing linear combination on the Taylor expansion of the activation function to obtain network output, and finally obtaining an optimal solution of a target function by minimizing a loss function. Through a simulation test of an ORL face data set, a feature face extracted by PCA and 2DPCA algorithms is used as networkinput, and the effectiveness of the method is verified.

The invention relates to an animal identity recognition system and method for increasing recognition rate through nose prints. As the nose print of an animal is generally unique like a human fingerprint, the animal is recognized by utilizing the characteristic and combining body characteristics and facial characteristics as comparison bases. Compared with a traditional chip implanted animal recognition means, the invention has the advantages that by means of a non-implanted means, the will of a feeder to actively log in the pet identity database can be improved, and meanwhile, when the invention is used for managing the epidemic animals (such as epidemic dogs and epidemic cats), needed expenditure is low, and recognition degree is quite excellent.