The Research Of Several Key Technologies In Gait Recognition

The gait recognition is one new research topic in the field of computer vision and intelligent video surveillance, which makes use of human walking pattern to recognize and identify. In this paper, several key technologies in gait recognition are researched, including the shadow elimination of moving human in the phase of pretreatment, the solving of occlusion in multi-human tracking, the extraction of gait features and the gait recognition.To the shadow elimination of moving human, two methods are put. One is based on color information, one is based on color space conversion. In shadow elimination way of based on color, angle of color and blue hue are made use of doing shadow elimination; In shadow elimination way of based on color space conversion, HSV color space and rgb color space are made use of doing shadow elimination. These methods can eliminate the shadow of moving human. To solving of occlusion in multi-human tracking, the centroid of moving human is forecasted making use of gray estimate equation. This method can better solve the occlusion. To the extraction of human gait features, two methods are put. One is based on angle histogram of moving pixels on silhouette of human and the other is based on region feature of moving human. To the gait recognition in single view, it is done in one gait period. Then the gait features are extracted in one gait period. According to the two gait features, the nearest neighbor classifier is used and better recognition result is achieved. To recognition based on multi-feature fusion, some experiments are done and the two feature vectors are bunched to get a new feature vector. At last, the gait recognition is put into multi-human recognition and solve some problems in the background in which there are some people. It has become the basis of gait recognition in the future.After preliminary experiment, the method that we adopt has reached better result. This demonstrates the feasibility of the method we adopt.