The Research On Recognitionmethod About Gait Behavior Based On Frequency Feature

Moving objects detection and behavior identification in video sequences are topics related with image processing, computer vision, pattern recognition, and artificial intelligence and so on. They have been more and more important in recent years because of their widely application in intelligent video surveillance, human-computer interaction, video indexing. Since human behavior is diverse and soft, as well as video scene is complex, the study on human activity recognition is a very hard issue.In this paper, moving object detection and human activity recognition are the focus of our research. For moving object detection, we summarize the existing algorithms on moving object detection and analyze the advantages and disadvantages of them. In the procedure of moving object, it will introduce various types disturb factor. For example, moving shadow, the gaps in moving object are common disturbance; they will be affected human behavior identification in the following, so we offer corresponding solutions to deal with the disturbance. To shadow removal problem, we adopted a robust and efficiently computed background subtraction algorithm that is able to cope with local illumination changes, such as shadows and highlights, as well as global illumination changes. The algorithm is based on a proposed computational color model which separates the brightness from the chromaticity component. Using the computational model detect and remove shadow region. In order to overcome the shortcoming of the present shadow elimination algorithm, we take advantage of two-directional scan line algorithm to fill the gaps on the moving object in binary image. Try our best to retrieve the character of moving object. In addition, in the process of moving object detection, it’s more likely that complete component divide. Directed against the situation, we make use of component labeling method to extract the main region. Experiments validate the effectiveness of our approach.We develop a novel approach named functional fuzzy recognition technology, which can solve the taster that it’s difficult to find appropriate cut-off frequency in the process of low-pass filtering. My paper is based on star skeleton algorithm. We adopt functional fuzzy recognition technology to cope with the disturb factors in dataset which is consisted of moving object’s center-boundary distance. The robustness of behavior identification is improved and the difficulty of threshold selection is decreased.