Forward-Looking Sonar Image Stitching Based on Midline Template Matching in Polar Image

Ocean perception has always been an important research content in ocean development. Forward-looking sonar (FLS) is widely used in marine perception due to its advanced performance, real-time properties, and good observation effect. Through the stitching of multiple FLS images, large-scale ocean geomorphology can be realized. However, most of the existing stitching methods of FLS images come from the stitching methods of optical images, which do not consider the unique characteristics of FLS images. Therefore, the result of FLS image stitching is prone to blur and misalignment. Through the analysis of the correspondence between the sonar motion and the pixel motion on the FLS polar image, it is found that the pixel motion in the midline area of the polar image has good stability. According to this characteristic, an FLS image registration method based on a midline template matching algorithm is proposed in this article. At the same time, the motion model of the sonar is introduced as a new constraint to the stitching process creatively, which makes the sonar image sequence stitch more accurately. In the end, the comparison experiments verify that the proposed method has better motion estimation accuracy and achieves a better stitching effect.