TRAFFIC AND VIDEO QUALITY WITH ADAPTIVE NEURAL COMPRESSION

Video sequences are major sources of traffic for broadband ISDN networ ks, and video compression is fundamental to the efficient use of such networks. We present a novel neural method to achieve real-time adapti ve compression of video. This tends to maintain a target quality of th e decompressed image specified by the user. The method uses a set of c ompression/decompression neural networks of different levels of compre ssion, as well as a simple motion-detection procedure. We describe the method and present experimental data concerning its performance and t raffic characteristics with real video sequences. The impact of this c ompression method on ATM-ceIl traffic is also investigated and measure ment data are provided.