关键词: 合成孔径声纳, 图像均衡, Retinex全变分, 乘性噪声, 平均距离向

Abstract: To address the issue of low visual resolution caused by imbalanced grayscale in synthetic aperture sonar images, a synthetic aperture sonar image equalization method based on the total variation framework is proposed by combining the sonar echo model, speckle noise statistical characteristics, and decomposition model prior assumptions. Firstly, the synthetic aperture sonar image decomposition model considering background noise is derived based on the characteristics of the Rayleigh multiplicative distribution of noise in the image and maximum likelihood estimation; Then, in response to the terrain and beam patterns of the illumination component, the sediment angle response of the reflection component, and the influence of noise, the objective function for the decomposition of synthetic aperture sonar images is established through segmented smoothing constraints on the illumination component, sediment echo intensity information constraints on the reflection component, and anisotropic total variation constraints; Finally, after enhancing the contrast component, the balanced image is obtained by multiplying it with the reflection component. The experimental results show that the model solves the problem of uneven brightness in SAS images, greatly improves brightness and contrast, and outperforms existing methods in noise suppression and image equalization.

Key words: synthetic aperture sonar, image equalization, retinex variational, multiplicative noise;average distance direction