Adaptive Underwater Image Compression with High Robust Based on Compressed Sensing

Saliency was overlooked by most of the conventional block compressed sensing (BCS). Actually, image blocks belong to salient region need more measurements while less measurements can meet the need of image blocks that are not salient when considering about observation in BCS. In this paper, an adaptive image compression based on compressed sensing is proposed. The observation matrix is designed according to the image saliency. The more important image block is, the more measurements this block will get. The saliency will be measured according to spatial domain (Image Activity Measure) and frequency domain (bispectrum). In addition, condition for underwater acoustic channel is poor than wireless channel, the need for a transmission scheme with high robust is greater. In this paper, the lost measurement can be located according to error detecting coding, and the lost measurement can be repaired by interpolation. Relevant experiments shows that the method proposed in this paper can improved the compression ratio and quality of reconstructed image in receiving end.