The Research On The Method Of Bayesian Gait Recognition Based On Triangle Characteristic Of Human Leg

With the requirement of security in modern society enhancing, the traditional ID recognition method is limited, while the biological recognition technology which is regarded as the way of realization security, has been greatly explored and developed. Gait recognition as an emerging domain in biological recognition technology is becoming a leading edge in the field of computer vision recently for its difficult concealment, easy gathering and Long-distance Low resolution ratio. This thesis does some exploratory work in gait feature extraction, Classified, and Recognition system Construction.For the huge gait characteristic and complex computation of the defects of recent gait recognition research papers, this thesis proposes a kind of Bayesian gait recognition method based on Thigh Triangle Characteristic. An improved background subtraction algorithm is used for the gait sequences to subtract background, then binarization and morphologic method to eliminating the image noise, finally extracts essential frame which have the maximal or minimal stride according to stride variation curve. An approach is proposed to search the person's trunk of the essential frames, and divide the trunk into five parts. A triangle is applied to simulate the legs of essential frame, and extract its parameters as the gait feature, which are three edges of the triangle, the top angle and the coordinate of hip. KNN and a new N-best algorithm are used for test-data search the optimal numerical value from training data before Bayesian classification. The experimental results on the NLRP database show that the new method is not only simple and fast but also effective.At the same time, this thesis proposes and realizes a self-learning and domo system of gait recognition based on the algorithm proposed by ourselves, which adopts Matlab language as developing instrument and combining Web technique and its corresponding function. The system is mainly composed of gait pattern recognition technique, running through fundamental principles, methods and theories. The gait recognition system for self-learning and domo is practical application demonstration of pattern recognition, which can make students understand sensibly and enhance their ability of the self-learning.