A Method for Background Suppression of Sonar Image Using Gaussian Mixture Model and Radon Transform

The background of sonar image of small underwater moving target fluctuates with space and time. The background suppression is one of the key processes in sonar detection. In this paper, the Gaussian mixture model is used to model the amplitude variation of each azimuth-range resolution cell of sequential sonar image, so as to suppress the main background of highlight area and highlight band in the sonar image. In order to solve the problem of residual noise from the Gaussian mixture model, the continuous change characteristics of the track from an underwater moving target are considered, the line features of the history-accumulated sonar image are extracted, and the residual noise is filtered out by Radon transform and inverse Radon transform. The analysis of multi-beam sonar data from measurement under the conditions of small underwater target’s linear motion and cross-azimuth curvilinear motion shows that the complex sonar background can be removed and the clean sonar back-ground can be obtained by using the Gaussian mixture model and the filtering method via Radon transform, which can be used in sonar detection of underwater moving targets.