An Underwater Target Enhancement Method for Three-Dimensional Sonar Images Based on Time Accumulation

Abstract

Abstract Three-dimensional sonar (3DS) can get the image of the entire scene by emitting sound signal once time. However, the image quality is poor manifested as sparse and missing of target surface points, owing to complex underwater environment and hardware limitation of sonar. To obtain dense and complete object structure, we propose a target enhancement method based on time-accumulation. This method includes two crucial steps: 1) Space mapping and registration. The operation convert the 3DS images from polar coordinate system to the Cartesian coordinate system, which can acquire object point cloud. Then the spatial deviation between image frames as well as relative motion between sonar and object would be corrected. 2) Point data fusion. The operation can covering different dynamic environment, time, and target morphology by fusing at the data level, thus going the result of enhancement. Experimental results demonstrate that our proposed enhancement method can effectively enhance the object result, improve the quality and understandability of the 3DS image.