ROI-based joint source-channel coding for wireless video transmission

In this paper, a novel method is proposed for real-time video coding and transmission over wireless channel. The method optimally allocates the resources between source and channel coder by taking the characteristic of the human visual system and the condition of the time-varying wireless channel. Specifically, we first propose our method to extract the region of interest (ROI) from video sequences. Then we present how to estimation the condition of wireless channel, such as efficient throughput and error pattern, etc. Lastly, ROI-based joint source-channel coding method is proposed. The experimental results show that the proposed method can significantly improve video quality, especially subjective video quality, in the presence of wireless bursty errors.