Abstract: The serious noise pollution, low contrast of gray values of target and background areas, and weak edges in the side-scan sonar image are due to the limitations of side-scan sonar imaging mechanism and the abundant noise sources in the ocean. In response to the above problems, a partition enhancement method for the non-subsampled contourlet transform (NSCT) domain of side-scan sonar images is proposed. For the low-frequency part of sonar image, a nonlinear function enhancement method is used to improve the contrast of low-frequency image. For the high-frequency part of sonar image, the noise and texture edges are partitioned and the corresponding processing is made by analyzing the distribution rule of the difference between the maximum and minimum values of sub-band coefficients in different directions on the same scale. The proposed method was compared with the gamma enhancement method and the wavelet threshold enhancement method through experiment. The results show that the proposed method not only denoises noise well, but also can suppresses trivial textures and enhances weak edges.The enhancement effect for side-scan sonar image is more prominent.

Key words: side-scansonar, non-subsampledcontourlettransform, coefficientdifference, partitionenhancement