An efficient moving object detection and distance measurement algorithm using correlation window

Moving object extraction and depth perception is one of the complex tasks that restrict the real time suitability of any video surveillance system. This paper presents a new method of extracting any moving object and its depth calculation. A modified background subtraction algorithm is used to detect the moving object. While a correlation window based method is used for solving the correspondence problem. The main proposal of this paper lies in searching a suitable window in the left image that has the least correlation with its neighboring windows which actually reduces the possibility of finding a false corresponding window in the right image. The difference between the positions of the similar windows gives rise of disparity which is later used to calculate the depth. This method shows better result than other existing methods.