Object-based video coding using pixel state analysis

In archiving video for surveillance, frame-based coding has been used and it makes storage size large because the whole image is stored even if there is no object in the image. On the other hand, object-based coding has the capability to make storage size small, because it distinguishes between the foreground and the background regions of the image, and stores only foreground objects such as people. This paper describes object-based coding by pixel state analysis. In our method, pixel state analysis detects the foreground objects and background regions in video frames. Furthermore, it distinguishes foreground object pixels as stationary or transient pixels. For stationary pixels, it is possible to restore the color intensity by referring to the same pixel location in the last frame. Therefore, our method makes the storage size smaller. Additionally, the transient pixels of foreground objects are compressed using LZH codec. Since LZH codec uses lossless compression, the object region can be compressed with lesser loss in image quality. We have evaluated our system over 9 test sequences and obtained an improvement of 15% in compression ratio and better quality for the moving parts of the object region compared to MPEG-4.