Spatio-Temporal Analysis for Moving Object Detection Under Complex Environment

Object detection is one of the main functions in automated video analysis. However, in reality, when the video was taken in the outdoor environment, noisy obstacles frequently perturb our effort to detect interest objects. This work tried to develop a method for the detection of moving objects under noisy environment by analyzing the motion pattern in spatio-temporal domain for large range. Moving object can be determined from motion pattern of object, which is represented by two kernels of temporal domain: vertical and horizontal. Information from temporal domain can be used to overcome the background noise and detect moving objects. The results of our experiments had achieved <(TPR)over bar> of 77.76% and (FPR) over bar of 3.14%.