Retinex-Based Bilateral Filter and Discriminative Dictionary Learning for Enhancing Sonar Image
Ping Zhou, Dongwei Yang, Zhixiang Yang, Song Guo · 2023
Although the sonar image has been corrected, there exist in some problems with low illumination, and unclear edge texture. Therefore, it is urgent to solve scattered noise, image blur and uneven illumination in sonar image enhancement. This article takes side scan sonar images as the research object and decomposes them into low and high frequency sub band images by non-subsampled Shearlet transform (NSST). Retinex-based bilateral filter is applied to enhance the contrast of low frequency image. Image denoising of high frequency sub bands is achieved with sparse dictionary learning. Experimental results show that the enhanced image obtained by proposed method can clearly display the target texture, terrain change and edge information, and the speckle noise can be reduced and suppressed better. The objective indicator values are the best, reflecting the contrast, clarity, information content, and filtering performance of the image. The implementation of this work is conducive to providing clear target images and texture information for subsequent application research.