An Efficient Underwater Video Compression Algorithm for Underwater Acoustic Channel Transmission

This paper presents an efficient underwater video-coding algorithm suitable for the purpose of underwater acoustic channel transmission. The algorithm mainly consists of three modules, namely preprocessing of underwater images, global motion estimation, and hybrid global-local motion compensation. A wavelet transform based preprocessing approach is employed to effectively reduce the visual redundancy of underwater images. Global motion transform between neighboring frames are estimated by using a fast and robust feature-point based algorithm. Then a hybrid global-local motion compensation scheme is proposed to reduce the temporal redundancy of video sequences. The experimental results show that while maintaining the acceptable perceptual quality of reconstructed images, the proposed algorithm can provide high coding efficiency, especially applying to sea floor video compression and transmission at very low bit rates to meet the limited bandwidth requirement of underwater acoustic channels.