Adaptive scheme for Internet video transmission

Video streaming over the Internet is a growing user demand. Transmission of compressed video over the best effort Internet will inevitably result in serious degradations in video quality. In order to alleviate this effect, MPEG-4, the latest video standard, provides a number of error-resilient tools as well as the Fine Granularity Scalability (FGS) that adapts to the varying bit rates. Unfortunately, Error-resilient tools are not efficient when channel loss is high while FGS coding overhead is significant for low bandwidth channels. To guarantee better image quality during transmission, we propose an adaptive scheme that alternates between two different coding schemes depending on the channel status and underlying visual content. Furthermore, a Multi-Layer Perceptron (MLP) network is designed to provide an on-line decision mechanism with 86% accuracy. Experimental results suggest that our approach can achieve a superior video quality for a wide range of packet loss and video content.