Spatio-temporal video filtering for video surveillance applications

Video surveillance cameras are inexpensive but the manpower required to monitor and analyze is costly. Therefore, the captured videos are usually monitored sparingly. Nevertheless, only a very small part of the recorded information is really relevant. To avoid huge data stream transmission, we propose to pre-analyze the video at the sensor to extract information of interest. Then, compress and send out valuable parts to be analyzed more precisely at the control station. In this article, we present a spatio-temporal video filtering method in the context of video surveillance applications aiming to filter as possible the captured data without altering the interesting information for surveillance purposes. The proposed approach ameliorates the H.264/AVC encoding phase by reducing the prediction error values during the motion estimation step which leads to a faster and better compression.