A low Complexity Change Detection Algorithm Operating in the Compressed Domain

This paper introduces a simple, fast, and efficient algorithm for image change detection in the compressed domain. The proposed technique operates directly on Discrete Cosine Transformed (DCT-ed) data which makes it suitable for processing compressed bit streams produced with DCT-based encoders such as Motion JPEG, MPEG-x, and H.26x. Its limited hardware requirements render this method compatible with low-cost, low-complexity, and low-power systems. There sults given in the frame of a video surveillance application demonstrate the excellent performance of this technique to detect the significant change in a scene while filtering out the noise. This method exploits the fact that the phase of the DCT coefficients of a transformed image contains a significant amount of information. By processing only the phase part of the DCT coefficients a simple, fast, efficient, and yet robust change detection method is achieved.