Research Of Efficient And Error-Resilient Underwater Video Coding Method

With the development of science and technology, ocean exploration and development has become the focus of scientific research. This includes tasks such as ocean environment measurement, submarine exploration, and submarine operation and so on. The widely used deep-sea observation equipment is ROVs or AUVs-based video surveying system, where optical video imaging is one of the key techniques for underwater short distance inspection and operation. However, how to transmit massive amount of video information acquired by ROVs or AUVs to the surface station is one of the challenges for deep-sea observation. Before transmitting the massive amount of underwater video, the video information must be compressed to meet the requirement of acoustic channels, which is characterized by narrow bandwidth and high error rate. According to the characteristic of underwater video and acoustic channels, efficient and error-resilient encoding algorithms for underwater video images are developed in this dissertation. The main research work finished includes the following two parts:(1) Research of efficient and error-resilient underwater image encoding algorithm. Firstly, the visual redundancy of underwater images is removed by a wavelet-based preprocessing technique according to the characteristics of underwater images. Secondly, the low frequency coefficients are quantized and encoded by a fixed length algorithm, and the values and the their corresponding positions of the important high frequency coefficients are coded separately, where the values are coded by a variable precision and fixed length algorithm, and their corresponding positions are coded by a position difference reduction–based RVLC (Reversible and Variable Length Coding) algorithm. Moreover, a packing and interleaving scheme and a placing scheme of the variable length coding blocks with respect to the fixed length acoustic channel packets are proposed to resist the transmission errors of the underwater acoustic channels. Finally, error detection and concealment algorithms for low and high frequency coefficients are presented respectively. The experimental results show that the proposed coding algorithm can resist package lost error less than 20%, random bit error less than 2%, and burst error less than 2.5%. Meanwhile, the image compression ratio can reach to 70:1, with efficient compression performance.(2) Research of efficient and error-resilient underwater video encoding algorithm. The proposed underwater video coding algorithm consists of the intra-frame coding module and the inter-frame coding module. In the intra-frame coding module, the first I frame in each video sequence group is compressed by the efficient and error-resilient underwater static image encoding algorithm mentioned above, thus providing an ideal reference frame for the reconstruction of frames remained in a video sequence group. In the inter-frame coding module, after the process of motion estimation and motion compensation, an error-resilient algorithm is developed for the obtained motion vectors and residuals, where a packing and interleaving scheme is introduced to prevent the diffusion of channel errors. In addition, the residual coefficients are coded by the position difference reduction based RVLC algorithm, and are protected from their importance. At the decoding end, an error detection and concealment algorithm is developed to reduce the influence of transmitting errors. Moreover, in order to allow the whole coding algorithm to adapt to various complex environments, the coding algorithm can detect and resist some abnormal situations in video sequences, such as scene switch, fast camera motion and so on. Experimental results show that the developed algorithm can resist package lost error less than 20%, random bit error less than 1.2%, and burst error less than 1.8%, and the compression ratio for video sequence can reach to 120:1.The research is one of the key technologies of underwater video based observation for underwater robots, possessing important research value and broad application prospect for deep sea exploration and development.