Detection of the moving targets of forward-looking sonar based on background elimination method

Forward-looking sonar (FLS) is widely used in the industry of underwater moving targets detection, both in sea and lake environments. Aiming to detect the moving targets in the FLS images for fixed scene, a background elimination method was proposed. Firstly, choosing the sonar beam data as the processing object, a background model was established by the mid-value of each echo point in the time dimension. Secondly, the moving targets difference image was generated by background elimination algorithm. Finally, the moving targets binary image was segmented from the difference image by Otsu threshold method. In the second step, compared with the mean-value background used in single gaussian background model, the mid-value background was less disturbed by moving targets and more consistent with the actual background. Compared with the inter-frame diff8erence method, the proposed method could accurately segment the moving target from the FLS images. Therefore, the proposed method can be used in underwater moving targets detection, also it has the potential application in the underwater monitoring work.