Detection and analysis of moving objects using time series stereo difference images

Recently, methods for analyzing the motion and size of moving objects in various scenes, as in uninhabited surveillance systems, are searched for. The aim of this research is proposing a new method for moving objects extraction and analysis using time series stereo images, which is realizing high-speed and high-precision. At first creating the difference images from time series stereo images obtained from three CCD cameras, and the moving object´s portion of the image, can be extracted. Next applying the theory of stereo vision for extracted moving objects image, the depth image for only moving object portion is obtained. Furthermore, by using the extracted positional and depth information of moving objects, the project area and the volume as an index of size and the velocity vector as an index of movement are calculated, as the detected moving objects information. Finally, the proposed method is applied to the experimental images and the effect is confirmed.