Research On Gait Recognition Technology And System Realization

With the rapid development of computer vision technology,there is a growing demand for security.Biometrics,as a common authentication technology for security situations,has received extensive attention and research.Face,fingerprints and other biological features not only require close contact with the camera device,but achieve higher costs to realize the technology.And gait recognition,with non-contact,concealment and other characteristics,overcome the traditional biometric identification that needs high-resolution image defects.Based on the OpenCV open source function library,this paper takes C and C ++(based on MFC)as the development tool to develop a complete gait recognition access control system,which is based on the application of gait recognition system.This paper designs a complete set of gait recognition technology,and that is developed by Matlab with a large CASIA-B gait database provided by the Chinese Academy of Sciences Institute of automation of samples for testing,and the test results show that the gait recognition technology scheme achieves high recognition rate.The experimental results show that this gait recognition scheme achieves a very high recognition rate.On this basis,Completing the following three aspects of work:First,using C and C++ to achieve the entire gait recognition technology and forming various functional modules for independent use;Second,building a complete gait image collection environment.The image acquisition environment is composed of a camera module,a single color background and a light,forming a fixed collection environment to reduce the noise caused by external factors;Third,proposing a background detection and removal algorithm and rejection algorithm.After adding the camera module,due to the actual environment noise interference,the image preprocessing stage will produce a lot of noise(shadow),which is not conducive to feature extraction;After adding the camera module,due to the actual environment noise interference,the image preprocessing stage will produce a lot of noise(shadow),that is not conducive to feature extraction;the channel Y,Cr or Cb of the target image and background image were differenced while the image is converted to YCrCb,finding the shadow position and achieving shadow removal through the set of frame difference threshold.In addition,the access control system not only needs to meet the high recognition rate in the practical application,but also have a rejection function,that not only can accurately identify the real sample in the training sample library,but also refuse to identify the false sample which does not belong to the training sample library.Therefore,this paper studies the recognition module in gait recognition,and proposes a rejection algorithm based on extreme value optimization.Only in the premise of training samples,finding the projection direction minimum dispersion of all samples after the projection geometry space and determining the rejection interval;it achieves rejection function when the false test sample is projected in the projection direction,and fall in the rejection interval.The experimental results show that the rejection algorithm can achieve a high rejection rate and confirm the effectiveness of the algorithm.With real-time strong and user-friendly interface,this paper finally implements a complete and available gait recognition system based on OpenCV.