A Novel Approach to Analyzing Object Motion Behavior

Object behavior analysis is important in traffic behavior analysis. In this paper, we propose a novel approach to analyzing object motion behavior. First we build the background probability model, then background subtraction, binarization, median filtering and morphological operation are implemented in sequence to obtain the candidate regions. Next, the sum of gradient is used for shadow removal. In the candidate region the true object is detect and their behavior is analyzed. The experimental results confirm the efficiency of proposed approach.