An adaptive background-frame based video coding method

Surveillance videos show a prominent feature of background redundancy. To full use this, we propose an adaptive background-frame based coding method for surveillance videos in this paper. In this algorithm, the adaptive background image is used as a long-term reference frame to provide better prediction for inter-frame coding. The background image is generated and updated dynamically by using layer information and reconstructed images, where layer information is obtained by background modeling and motion detection from the original sequences and encoded with Pattern Quadtree. Over the AVS2 test model RD7.0, this method can save averagely 9.7% of the coding bit-rate, equivalent to 0.25dB performance gain on the surveillance videos.