Gait recognition based on various viewing angles

A method for analyzing and recognizing person gait based on various viewing angles was proposed. Firstly, motion human body was detected by using adaptive background updating algorithm. Then, motion track period, which has been briefed as centroid, was extracted by using 4-directions chain code. Finally, by different trained periods to get corresponding fuzzy sets, and in addition with membership function, three gait types such as jogging, walking and running was recognized. Experimental results show that the method is performed well at moving speed estimation and gait recognition, and it is remarkable robustness for background noise and varying viewing angles.