A Review of Sonar Image Segmentation for Underwater Small Targets

Abstract

The existing image segmentation methods are various, but due to the particularity of sonar images, the ordinary image segmentation methods often fail to achieve ideal results when processing sonar images and have limitations. On the basis of studying a large number of image segmentation methods at home and abroad, the authors recommends the following methods with comprehensive characteristics that are more suitable for sonar image segmentation, such as thresholding, edge detection, MRF and clustering algorithm. For nearly a decade, many scholars have improved the traditional segmentation methods by combining the characteristics of sonar images, or improved the algorithms by combining various segmentation methods to make up for the shortcomings of the original segmentation methods applied to sonar images.