35results about How to "Improve face recognition performance" patented technology

A face imaging system for recordal and / or automated identity confirmation, including a camera unit and a camera unit controller. The camera unit includes a video camera, a rotatable mirror system for directing images of the security area into the video camera, and a ranging unit for detecting the presence of a target and for providing target range data, comprising distance, angle and width information, to the camera unit controller. The camera unit controller includes software for detecting face images of the target, tracking of detected face images, and capture of high quality face images. A communication system is provided for sending the captured face images to an external controller for face verification, face recognition and database searching. Face detection and face tracking is performed using the combination of video images and range data and the captured face images are recorded and / or made available for face recognition and searching.

A scalable face recognition method and apparatus using complementary features. The scalable face recognition apparatus includes a multi-analysis unit which analyzes a plurality of features of an input face image using a plurality of feature analysis techniques separately, compares the features of the input face image with a plurality of features of a reference image; and provides similarities as the results of the comparison, a fusion unit which fuses the similarities, and a determination unit which classifies the input face image according to a result of the fusion performed by the fusion unit.

A face image registration apparatus comprising: a face video image acquiring unit configured to acquire a front face image and at least one or more moving video images corresponding to predefined face pose changes; a similarity measurement unit configured to measure a similarity of the acquired front face image and each of the acquired at least one or more moving video images; an image selection unit configured to select at least images from the at least one or more moving video images based on the measured similarities;and an image storage unit configured to store the selected at least one or more images on the user basis.

The invention provides a human face identification method and apparatus, computer equipment and a readable storage medium. The method comprises the following steps: any human face image in a trainingset is input into a depth residual error network module group for training operation, a human force identification model is built; human face identification characteristics with specified dimensions are output, corresponding categories of multiple tasks for a corresponding human face image are predicted according to the human face identification characteristics with the specified dimensions basedon a full link layer classifier; a training loss value is calculated according to the predicted corresponding categories of multiple tasks for the corresponding human face image based on a loss function; whether to input any human face image in the training set into the depth residual error network module group for the training operation is determined according to the training loss value, and thehuman force identification model is built. Via the human face identification method and apparatus, the computer equipment and the readable storage medium disclosed in a technical solution of the invention, in-depth of training and construction of the human face identification model can be realized; multiple tasks can be subjected to the training operation, the human face identification model obtained based on the training operation is high in identification effects, multiple tasks of human face identification can be performed, and identification efficiency can be improved.

The invention discloses a video face identifying method which comprises the following steps of S1, carrying out face detection and trace on video to obtain face sequences, S2, screening the face sequences to obtain a typical face frame set, S3, optimizing the typical face frame set based on a front face generating technique and an image super resolution technique to obtain a reinforced typical face frame set, and S4, comparing the reinforced typical face frame set with a preset static face image matching base to identify or verify faces. Compared with an existing video face identifying method, the video face identifying method filters and compensates change of video face postures and resolutions through the reinforced typical face frame set. Thus, the robustness of video face identification is improved.

The invention belongs to the field of face recognition, in particular to a neural-network-based large-scale unbalanced data face recognition method and system, and aims to solve the problems of large-scale data optimization and improvement of face recognition efficiency. According to the method, the performance of model face recognition is improved by improving the loss function and the sampling mode, the loss function of the self-adaptive boundary margin is provided in the aspect of the loss function to cope with unbalanced face data, and an improvement scheme is provided in the aspect of sampling for data sampling and classification template sampling. According to the method, model training can be efficiently carried out on large-scale unbalanced face data, and the performance is improved.

The invention provides a face identification method and a face identification system. The method comprises the following steps: S101) carrying out segmentation on glasses in a glasses picture to obtain a glasses template; S102) defining face key points on the picture of human face without glasses, and defining glasses key points on the glasses picture and the glasses template; and S103) adjusting the glasses picture and the glasses template according to the face key points and the glasses key points so as to enable the glasses picture and the glasses template to be matched with the picture of human face without glasses, and the glasses in the glasses picture are superposed on the picture of human face without glasses through the glasses template to form a picture of human face wearing the glasses. The method can realize quicker, simpler and accurate identification of the human face wearing the glasses, and provides better identification performance. The face identification system also has the advantages above.

The invention discloses a robust human face image principal component feature extraction method and identification apparatus. The method comprises: by considering low-rank and sparse characteristics of training sample data of a human face image at the same time, directly performing low-rank and L1-norm minimization on a principal component feature embedded through projection, performing encoding to obtain robust projection P with good descriptiveness, directly extracting a low-rank and sparse principal component union feature of the human face image, and finishing image error correction processing; and by utilizing the embedded principal component feature of a training sample of a robust projection model, obtaining a linear multi-class classifier W* for classifying human face test images through an additional classification error minimization problem. When test samples are processed, a union feature of the test samples is extracted by utilizing a linear matrix P and then the test samples are classified by utilizing the classifier W*; and by introducing a thought of low-rank recovery and sparse description, the principal component feature, with better descriptiveness, of the human face image can be obtained by encoding, the noise can be eliminated, and the effect of human face identification is effectively improved.

The invention discloses a face recognition method, system and device based on centralized coordination learning, and the method comprises the following steps: obtaining a to-be-recognized face image,carrying out the face detection of the face image, and obtaining a first face image; After alignment processing is carried out on the first face image, a second face image with a preset size is obtained; inputting the second face image into a preset face recognition model based on centralized coordination learning for feature extraction, and obtaining a face feature vector of the second face image; and calculating cosine similarity by combining the face feature vector and a preset face database, and obtaining a face recognition result according to the cosine similarity. According to the invention, a face recognition model based on centralized coordination learning is adopted to carry out feature extraction on the face image, each feature is pulled to an original point and is respectively put into all quadrants, the inter-class distance is larger, the classification efficiency and recognition accuracy of the face are improved, and the method can be widely applied to the technical fieldof face recognition.

A face identification method includes the following steps. First, first and second sets of hidden layer parameters, which respectively correspond to first and second database character vectors, are obtained by way of training according to multiple first and second training character data. Next, first and second back propagation neural networks (BPNNs) are established according to the first and second sets of hidden layer parameters, respectively. Then, to-be-identified data are provided to the first BPNN to find a first output character vector. Next, whether the first output character vector satisfies an identification criterion is determined. If not, the to-be-identified data are provided to the second BPNN to find a second output character vector. Then, whether the second output character vector satisfies the identification criterion is determined. If yes, the to-be-identified data are identified as corresponding to the second database character vector.

The invention discloses a face recognition method and device based on deep learning, and the method comprises the steps: obtaining a face image training sample and a to-be-detected face image; extracting face training features from the face picture training sample, and extracting to-be-detected face features from the to-be-detected face picture; constructing a convolutional neural network model, and training the face training features of the face picture training sample by using the convolutional neural network model to obtain a face recognition model; and comparing the to-be-detected face features of the to-be-detected face picture according to the trained face recognition model so as to recognize the to-be-detected face picture. According to the method of the invention, the method can enable the inter-class spacing to be more uniform, can train more types of training data, can achieve the large-scale face data training, can improve the face recognition efficiency, and improves the face recognition performance.

The invention discloses a training method of a deep convolutional neural network for face recognition. The method comprises the following steps: 1) preparing a face image data set, dividing the face image data set into a training set and a verification set, and selecting the type, the structure, the hyper-parameter and the magnitude of a deep convolutional neural network model according to the scale and the complexity of the training set and the performance index of face recognition which should be achieved; 2) extracting features of face pictures input by the training set by using the model,and taking the features as input in the step 3); 3) constructing a loss layer, and iteratively calculating a loss value for the training; 4) comparing the loss value calculated in the step 3) with a preset threshold value, judging whether training is stopped or the gradient is calculated, and updating model parameters; and 5) verifying the model performance, and determining whether to stop training. According to the method, the human face features can be constrained by using a multivariate acting force from two aspects of an Euclidean space and an angle space during training, so that the deepconvolutional neural network model can learn the human face features with higher discrimination and robustness.

The invention relates to a face recognition method and device, computer equipment and a storage medium. The method comprises the steps of obtaining a to-be-identified image; extracting a complete faceimage and a specific area face image in the to-be-recognized image; based on a preset face recognition network, obtaining to-be-recognized face features corresponding to the complete face image and the face image in the specific area; and performing face recognition on the to-be-recognized image according to the to-be-recognized face features. According to the invention, in the face recognition process, the features corresponding to the complete face image and the face image in the specific area are extracted respectively, then feature fusion is carried out to obtain the to-be-recognized facefeatures, and face recognition is assisted based on the to-be-recognized face features, so that the normal face recognition performance is not influenced basically, meanwhile, the human face recognition performance of the partially shielded human face is improved.

The invention provides a low-quality face image enhancement method, system and device, and a storage medium. The method comprises the following steps: synthesizing a low-quality face image based on animage processing method; calculating a face weight matrix of the low-quality face image with respect to a five-sense-organ area of the low-quality face image; constructing a deep neural network, fitting parameters by using the low-quality face image, the real image and the corresponding face weight matrix as a training set, and generating an image enhancement model; and processing the input low-quality face image by using the image enhancement model to enhance the low-quality face image of the image. Low-quality face images caused by various reasons in a real environment are simulated by synthesizing the low-quality face images, so that the universality of the generated image enhancement model is improved; in the face recognition process, the GAN model is trained by using the supervisionsignal of the face feature similarity, so that the face recognition performance in a low-quality face image scene is improved.

A driver state monitoring system includes a first lighting module for driving a first lighting device, a camera for acquiring an image, a second lighting module for driving a second lighting device by synchronizing the second lighting device with the first lighting device wirelessly, and a controller for analyzing the image acquired by the camera to recognize a driver state.

A face recognition method based on multi-view collaborative complete discriminant subspace learning is provided. The method comprises the following steps: (1) using an objective function based on Cauchy loss and Fisher discriminant analysis to obtain complete feature representation as shown in the specification of the number as shown in the specification of training samples in a potential completesubspace, the number as shown in the specification of view generation functions as shown in the specification, and the number as shown in the specification of non-negative collaborative learning weights as shown in the specification; (2) given the non-convex nature of the objective function, obtaining two solutions as shown in the specification of the objective function by using the alternate solution method; (3) based on the solved view generation functions as shown in the specification and the non-negative collaborative learning weights as shown in the specification, solving complete feature representation of test samples in the complete discriminant subspace; and (4) based on the Euclidean distance between the test sample and the training sample in the complete discriminant subspace, classifying the test samples by using a nearest neighbor classifier. Compared with the existing multi-view face recognition method, the method provided by the present invention can more effectively fuse multi-view information and mine discriminant information, and is an effective multi-view face recognition method.

The invention provides a face video image quality optimization method. The face video image quality optimization method comprises the steps of collecting a video sequence containing a face; extractinga target face image in the video sequence; and carrying out no-reference quality evaluation on the target face image; and extracting a target face image in the video sequence by using an inter-framedifference method. According to the method, non-reference quality evaluation is carried out on the video sequence images, quality scores are given to the images of the same identity, and screening ofthe high-quality images and filtering of the low-quality images are achieved.

The invention discloses an access control machine based on face recognition, and relates to the technical field of access control machines, the access control machine based on face recognition comprises a fixed cross beam and a control terminal, a first servo motor is fixedly installed on the right side of the fixed cross beam, an output shaft of the first servo motor is fixedly sleeved with a driving lead screw located in the fixed cross beam, and the outer portion of the driving lead screw is sleeved with an electric telescopic rod in a threaded mode. In the access control machine based on face recognition, the infrared induction range finders on the left side and the right side are used for detection; when the infrared induction range finder detects a user, the infrared induction rangefinder sends a signal to the first-level controller, the first-level controller controls the first servo motor to drive the driving lead screw to rotate, and therefore the electric telescopic rod andthe face recognition camera are driven to move, face recognition is conducted on a user, automatic detection on the user is achieved, and the automation degree is improved.

The invention discloses a cross-pose face recognition method based on a progressive neural network and an attention mechanism. The cross-pose face recognition method comprises the following steps: performing face detection and key point alignment by using an MTCNN tool; estimating attitude information of each face image; sampling a front face image for each face image to serve as a reference image, forming image pairs, inputting the image pairs into a designed progressive neural network, extracting face feature pairs, and performing identity prediction; calculating an attention weighted mean square error loss function by using the feature pairs, and training a neural network in combination with a classified cross entropy loss function; and extracting face features by using the trained neural network and performing face verification. According to the invention, a new lightweight progressive network structure and an attention-based loss function are constructed, and the features of the side face image can be effectively adjusted in a feature space, so that the problem that the face recognition performance is reduced due to posture change in the face recognition field is solved.

The invention discloses a low-resolution face recognition method based on component and compressed dictionary sparse representation, and belongs to the fields of signal processing, mode recognition, machine learning and computer vision. When the dictionary is constructed, images capable of sparsely representing all video frames in the video are selected as representative frames, and then the HOG features of the representative frames and the mirror images of the representative frames are used for constructing the part dictionary. During testing, the dictionary is used for representing each frame of a tested video in a linear mode, a feedback mechanism is added to correct an abnormal recognition result, and finally voting is carried out to obtain a video classification result. According to the method, the sparse representation is applied to video face recognition, the robustness of the sparse representation to shielding and noise is kept, other steps are added to improve the effect and efficiency of the method in large-scale low-resolution video face recognition, and the defects of the method under the conditions of illumination change and the like are overcome.

The invention discloses a self-adaptive threshold selection method and a face recognition method. The method comprises the steps: carrying out face quality evaluation of a face recognition test set in an actual scene, averagely dividing a face quality evaluation score into n intervals, dividing the face images in a query set into the corresponding intervals according to the face quality evaluation score, selecting proper face recognition similarity thresholds a and b for the face pictures of the n intervals according to actual requirements and the face recognition performance of the used algorithm, and establishing a corresponding table of the face quality scores and the face recognition similarity thresholds. The problem of face recognition similarity threshold setting problem for different-size and different-quality faces is solved by the refined self-adaptive face recognition similarity thresholds. The face recognition similarity threshold values of different face qualities can be set in a detailed manner, and then the overall face recognition effect is improved.

The invention relates to the technical field of image processing, particularly provides a face image quality assessment method and device and a computer readable storage medium, and aims to solve the problem of how to accurately assess the image quality of a face image so as to accurately perform face recognition. In order to achieve the purpose, the method comprises the steps of obtaining a face recognition model trained by different types of face image samples, and extracting image features of the same to-be-evaluated face image for multiple times through the face recognition model; respectively calculating a feature distance between every two image features and obtaining an average value of all feature distances obtained by calculation; predicting the probability that the to-be-evaluated face image belongs to the face image through a face recognition model according to the average value; and determining the face quality score of the to-be-evaluated face image according to the probability. Based on the above mode, the image quality of the face image can be accurately evaluated, so that face recognition can be accurately carried out.

The invention relates to the technical field of face recognition, in particular to an intelligent access control management system based on face recognition. The intelligent access control management system comprises an access control module and a face recognition module. The intelligent access control management system further comprises a human face detection module, which is used for detecting whether a human face exists in the recognition area or not and can obtain corresponding human face image information when the human face exists in the recognition area; a dormancy module, which is used for controlling the system to enter a dormancy state; and a control module, which is used for starting the dormancy module to enable the system to enter a dormancy state when no human face exists in the recognition area, and which is also used for starting the face recognition module when the face detection module obtains the face image information so that the face recognition module performs face recognition on the face image information, and can correspondingly controls the working state of the access control module according to the face recognition result of the face recognition module. According to the intelligent access control management system, the face recognition efficiency and the energy consumption management effect can be both considered, so that the use effect of the access control management system is improved.

A face recognition method based on multi-view collaborative complete discriminant subspace learning, including the following steps: (1) using the objective function based on Cauchy loss and Fisher discriminant analysis to obtain a complete feature representation of a training sample in the potential complete subspace, View generating functions and non-negative collaborative learning weights; (2) In view of the non-convex nature of the objective function, the solution of the objective function and the sum are obtained through alternate solutions; (3) Based on the solution-based view generating function and non-negative Solve the complete feature representation of the test sample in the complete discriminant subspace by collaborative learning weights of ; (4) Classify the test sample using the nearest neighbor classifier based on the Euclidean distance between the test sample and the training sample in the complete discriminant subspace. Compared with the existing multi-view face recognition method, this method can more effectively fuse multi-view information and mine identification information, and is an effective multi-view face recognition method.

The invention discloses a method for extracting robust face image principal component features and a recognition device. By considering the low-rank and sparse characteristics of face image training sample data at the same time, the principal component features embedded in a projection are directly subjected to low-rank sum L1‑ Norm minimization, coding to obtain a descriptive robust projection P, directly extract the joint low-rank and sparse principal component features of face images, and complete image error correction processing at the same time; use the embedding of training samples of robust projection models Principal component features, through an additional classification error minimization problem to obtain a linear multi-class classifier W * , for the classification of face test images; when processing test samples, use the linear matrix P to extract their joint features, and then use the classifier W * Classify; by introducing the ideas of low-rank recovery and sparse description, the principal component features of face images with stronger descriptiveness can be encoded, which can remove noise and effectively improve the effect of face recognition.

The invention discloses a low-resolution face recognition method based on component and compressed dictionary sparse representation, and belongs to the fields of signal processing, mode recognition, machine learning and computer vision. When the dictionary is constructed, images capable of sparsely representing all video frames in the video are selected as representative frames, and then the HOG features of the representative frames and the mirror images of the representative frames are used for constructing the part dictionary. During testing, the dictionary is used for representing each frame of a tested video in a linear mode, a feedback mechanism is added to correct an abnormal recognition result, and finally voting is carried out to obtain a video classification result. According to the method, the sparse representation is applied to video face recognition, the robustness of the sparse representation to shielding and noise is kept, other steps are added to improve the effect and efficiency of the method in large-scale low-resolution video face recognition, and the defects of the method under the conditions of illumination change and the like are overcome.

The invention discloses a single-person close-up real-time recognition and automatic screenshot method for large-scale live broadcast scenes, specifically as follows: obtain the current video frame image of the live video in real time, and detect whether there is a human face in the video frame image; if the video frame image If there is a face in the frame image, the face detection module is used to obtain the face area, and then the face area is comprehensively evaluated for face size, clarity, position and angle, and then the optimal face is selected, and the optimal face is selected. The current video frame image of the face is saved as a screenshot; finally, the optimal face image is sent to the face recognition module for recognition, and the recognition result is output. The invention can be applied to identify and take screenshots of single-person close-ups in the video in large-scale live broadcast scenarios, and automatically save the screenshots when the screenshot indicators are met; at the same time, the screenshot indicators can not only obtain screenshots with better quality, but also avoid face recognition. A large number of repeated face snapshots bring about a large amount of back-end server and computing workload.

The invention discloses an interactive photographing system based on human face recognition, which belongs to the technical field of human face recognition, and comprises a photographing instruction uploading module, the output end of which is electrically connected with the input end of a photographing instruction acquisition module; the output end of the shooting instruction acquisition module and the output end of the deep learning module are electrically connected with the input end of a face recognition module. According to the method, the deep learning module is arranged, the deep learning module can learn a large number of human face features and feature selective extraction rules, needed distinguishing features can be accurately captured for rapid recognition during face recognition, the recognition precision and efficiency are greatly improved, meanwhile, the convolutional neural network is established, and the recognition accuracy is improved. The convolutional neural network is composed of multiple layers of neural networks, each layer has multiple different planes, and by introducing the convolutional neural network, the identification degree of feature information can be improved, meanwhile, the complexity of face recognition can be reduced, and the face recognition effect can be further improved.

The invention discloses a multi-band light source face recognition method based on a convolutional neural network, and the method comprises the steps of collecting face data under different band light sources, extracting related feature vectors based on different types of convolutional neural networks, carrying out the dimensionality reduction and feature selection of the feature vectors based on an arcface method, and finally, using a nearest neighbor classifier to obtain a face recognition result, so that the main problems existing in face recognition under complex illumination changes are solved, face image information of different light sources can be fused, and the recognition precision is improved. More accurate face recognition can be realized in an invisible light environment; in addition, the face is expressed by the low-dimension feature vector containing richer information, the recognition speed under the condition of extremely rich information can be improved, and the face recognition capability is further improved.