Underwater Image Processing Based On Compressed Sensing

The ocean become important areas of human exploration gradually, and peoplecarry out marine scientific research work mainly through underwater imagingunderwater target information. With the increasing demands of underwaterinformation, the pressure which underwater imaging system’s hardware facilitiesundertake also increases. In recent years, Compressed Sensing theory which proposedby Donoho and Candès brings a new idea for people collecting signals. CompressedSensing theory makes signals sampling and compression coding happened in the samestep, the sampling process and the compression were completed at the same time. Theamounts of sample data will reduce greatly and avoid be wasted. All those could helpto alleviate hardware pressure caused by huge amounts of information. The mainworks of paper are as follows:Firstly, the paper introduces the application present situations and background ofthe compressed sensing theory, and the characteristics of underwater imaging. Thepaper introduces the basic knowledge of compressed sensing theory in detail, andanalysis the sparse representations, measurement matrixes, reconstruction methods ofcompressed sensing deeply. The paper introduces the structures and advantages offive kinds of commonly measurement matrixes in detail. The specific steps andperformance analysis about several kinds of typical minimum κ norm method,matching pursuit algorithms are introduced.Secondly, the paper introduces the basic theory of wavelet transform and waveletdenoising methods briefly. This paper puts forward an improved method based on theidea of wavelet threshold denoising in order to improve the application ofcompression perception in noisy images. The new method processes the coefficientsof noisy images based on sparse matrix, improves the sparse degree of noisy images.The experimental results show that this method increases the reconstruction effect ofnoisy image significantly, and shortens the operation time greatly. Finally, the paper introduces the underwater imaging technology and thedevelopment of related devices briefly and applies the compressed sensing theory tothe underwater imaging. The applications of sparse bases, measurement matrixes,reconstruction methods which commonly used are researched in underwater imagesand analyzed of simulation experiment under different sampling rates. Theexperimental simulation results shows that， the wavelet transform base of sparsetransform, parts of Hadamard matrix of measurement matrix, the OMP algorithm andSP algorithm of reconstruction algorithms have better effects in underwater imageprocessing. The paper puts forward the problems to be solved and further researchemphasis in this field, and look to the development of the field at the same time.