Patent•
Gait data-based identity recognition method
28 Nov 2017-
TL;DR: In this paper, a gait data-based identity recognition method is presented, which comprises the following steps of: firstly extracting gait profile curves of training samples and a to-be-recognized sample; processing the gait profiles curves by utilizing a line-by-line scanning method so as to obtain a high-dimensional gait feature matrix; carrying out dimensionality reduction on the high-dimensionality gait features matrix by utilizing an improved smooth auto-encoder; and finally judging which category of training sample is nearest to the to be-recognised sample
Abstract: The invention provides a gait data-based identity recognition method. The method comprises the following steps of: firstly extracting gait profile curves of training samples and a to-be-recognized sample; processing the gait profile curves by utilizing a line-by-line scanning method so as to obtain a high-dimensional gait feature matrix; carrying out dimensionality reduction on the high-dimensional gait feature matrix by utilizing an improved smooth auto-encoder; and finally judging which category of training samples is nearest to the to-be-recognized sample by utilizing a nearest neighbor algorithm. According to the identity recognition method provided by the invention, a new gait feature is adopted, and the improved smooth auto-encoder and a nearest neighbor theory are utilized to carry out feature dimensionality reduction and similarity judgement, so that structure information in two-dimensional gait images can be sufficiently utilized to describe gait differences between different persons, thereby improving the gait information-based identity recognition correctness.
Citations
More filters
Patent•
12 Oct 2018
TL;DR: In this paper, a history behavior concurrence-based data portioning method and system was proposed for unmanned attribute division carried out by utilizing history interaction data concurrence, and discloses a history behaviour concurrence based data componenting method.
Abstract: The invention relates to the field of machine classification, in particular to the field of unmanned attribute division carried out by utilizing history interaction data concurrence, and discloses a history behavior concurrence-based data portioning method and system. According to the method and system, information is obtained from mass data by utilizing long-term interaction data accumulation ofthe internet, and concurrent objects and information in interaction behaviors such as browsing and clicking carried out by most people through certain time windows in history data are classified through unsupervised technological means, so that the three difficulties that the object division standards are not uniform, no division rules exist or division is difficult and the artificial participation cost is high in the existing division are solved, the demanders are helped to solve the problems in the division besides saving the costs and creating values.
5 citations
Patent•
02 Nov 2018
TL;DR: In this article, a gait-based large-scale mobile phone user fast identity recognition method is proposed, where recognition of gait is realized through acquiring gait samples, constructing a gight searching sample set, performing Hash coding for the gait sample set and performing feature description for the Gait sample sets and recognizing unknown identity gait.
Abstract: The invention discloses a gait-based large-scale mobile phone user fast identity recognition method, wherein recognition of gait is realized through acquiring gait samples, constructing a gait searching sample set, performing Hash coding for the gait sample set, performing feature description for the gait sample set and recognizing unknown identity gait. In the method, by quickly searching out a batch of similar users and then performing detailed comparison, recognition time is reduced, moreover, by the designed Hash coding and feature description, accuracy rate is improved while characteristics of being speedy and convenient are realized.
1 citations
Patent•
17 Aug 2018
TL;DR: In this paper, a gait recognition method in combination with a perspective conversion model and a hidden Markov model is presented, which combines the advantages of the perspective conversion and the hidden MarkOV model to enhance the robustness of perspective change.
Abstract: The invention discloses a gait recognition method in combination with a perspective conversion model and a hidden Markov model. The method comprises selecting five representative gaits as key frames according to human walking habit characteristics; calculating and normalizing the distances from each frame in a gait period to the five key frames to construct a gait feature vector; constructing theperspective conversion model by using the gait feature vector in combination with truncated singular value decomposition and training the hidden Markov model parameters; converting, by the perspectiveconversion model, a test set observation vector into the same perspective as a registration set; finally, achieving the cross-perspective gait recognition based on the hidden Markov model. The test set converted by the perspective conversion model has higher commonality with the registration set. The hidden Markov model also facilitates the expression of a gait conversion process. The method combines the advantages of the perspective conversion model and the hidden Markov model to enhance the robustness of perspective change, can achieve better results in the case of cross-perspective recognition.
1 citations
Patent•
16 Jul 2019
TL;DR: In this article, a high-dimensional tensor analysis method is adopted to process electroencephalogram signals, and characteristics of the EEG signals of patients are obtained from multiple modes, including wavelet variation module, tensor decomposition module, data pattern prediction module, feature dimension reduction module and pattern classification prediction module.
Abstract: The invention discloses an electroencephalogram tensor pattern recognition technology and a brain-computer interactive rehabilitation system. The brain-computer interactive rehabilitation system comprises a signal acquisition module, a wavelet variation module, a tensor decomposition module, a data pattern prediction module, a feature dimension reduction module and a pattern classification prediction module. According to the present invention, a high-dimensional tensor analysis method is adopted to process electroencephalogram signals, and characteristics of the electroencephalogram signals ofpatients are obtained from multiple modes. In the aspect of model training, multi-layer deep network is adopted to train, and a robust model is obtained by applying a deep learning and training method. In rehabilitation training, the patients are provided with feedback signals during the training, so that the patients can adjust thinking activities by themselves through feedback intuitively. At the same time, the model is gradually fine-tuned in the rehabilitation training process, and updated according to newer electroencephalogram signal data, so that after a period time of running in, self-adaptive models for the patients can be trained, and the patients are better helped to carry out rehabilitation training and treatment recovery.
References
More filters
Patent•
20 Jan 2010
TL;DR: In this article, a method for extracting and processing gait feature information and identifying identification when people walk is proposed, which aims at reducing interference of external factors such as complicated background and the like, realizing better adaptivity to reality conditions.
Abstract: The invention relates to a method for extracting and processing gait feature information and identifying identification when people walk, and aims at reducing interference of external factors such as a complicated background and the like, realizing better adaptivity to reality conditions, more exactly extracting effective information which can reflect the walking feature of a motion human body so as to improve accuracy of gait identification. The technical proposal of the invention provides the method for gait information processing and identity identification based on a fusion feature. The method comprises the following steps: inputting a video sequence, segmenting profile information of a body object in a video image by target detection, synchronously extracting gait feature parameters by adopting boundary center distance and Radon conversion, carrying out corresponding post processing on the obtained feature parameters, taking a support vector machine as a classifier for classification identification, and evaluating the identification effect. The method is mainly applied to identity identification based on the gait feature information.
27 citations
Patent•
27 Jul 2012TL;DR: In this paper, a method of producing a gait representation for a subject, comprising the steps of: acquiring a sequence of images of the subject representing the gait of said subject, analysing each image of said sequence to identify one or more regions having a certain thickness based on a thickness characteristic function and a threshold value; for each image, removing said one OR more regions from said image to produce a modified image; and combining said modified images in the sequence to produce an image energy image.
Abstract: A method of producing a gait representation for a subject, comprising the steps of: acquiring a sequence of images of the subject representing the gait of said subject; analysing each image of said sequence to identify one or more regions having a certain thickness based on a thickness characteristic function and a threshold value; for each image, removing said one or more regions from said image to produce a modified image; and combining said modified images in the sequence to produce a gait energy image. Calculating and applying a thickness characteristic to the images allows a better identification of regions of the images which are most affected by covariate factors such as carrying an object or wearing heavy clothing. Such covariate factors have been found most often to be associated with the more static parts of the subject, i.e. the torso. The more dynamic parts of the subject, i.e. hands, legs and feet, are less affected by covariate factors and produce reliable gait information that can be used for identification purposes.
17 citations
Patent•
22 Oct 2008
TL;DR: In this article, a method for identifying sex based on walking, comprising two processes of training and identification, was proposed, which can help the intelligent vision monitoring control system identify sexes of people in monitoring scenes.
Abstract: The invention discloses a method for identifying sex based on walking, comprising two processes of training and identification. The walking video sequence marked with sex has characteristics picked up, and the characteristics picked up are trained for modeling so as to obtain sex classification model parameters; video data or camera data containing a walker have characteristics picked up, and the characteristics picked up are inputted in the model obtained by training so as to obtain the sex of the walker in the video. The sex identification based on walking is very important to improve the capability of the next generation intelligent monitoring control system in understanding monitoring scenes. When the invention is used for the intelligent vision monitoring control system, the invention can help the monitoring control system identify sexes of people in monitoring scenes, make the monitoring control system really understand what's going on in monitoring scenes and apply different safety levels according to different sexes of people in monitoring scenes; when the invention is used for the rough classification of people based on the walking identification, sexes of people can be judged, and the identification of a walker can be searched based on sex; the invention can also be used to analyze whether a customer is fond of specific products or specific entertainments.
9 citations
Patent•
06 Jun 2017
TL;DR: In this article, a variable-visual-angle gait recognition method based on static and dynamic feature fusion was proposed, which includes two types of gait features including the static gait feature based on regional mean distance and dynamic gait based on optical flow.
Abstract: The invention discloses a variable-visual-angle gait recognition method based on static and dynamic feature fusion. The method comprises the following steps: 1, extracting a semi-gait period based on the change of the number of pixels in a whole body region; 2, extracting a static gait feature based on regional average distance and adaptively extracting a dynamic gait feature by means of bidirectional optical flow prediction and error metric in combination with a semi-gait period extraction result; 3, fusing the static gait feature and the dynamic gait feature based on the idea of K proximity to classify and recognize the gait. The variable-visual-angle gait recognition method based on static and dynamic feature fusion uses two types of gait features including the static gait feature based on regional mean distance and dynamic gait feature based on optical flow. In the step of gait recognition, a feature fusion algorithm is used to classify the gaits and improve the gait recognition rate.
5 citations