Detection Technology for Dim and Small Targets Based on Forward-Looking Sonar Images

Due to the complex and changeable characteristics of the underwater environment, the low contrast between the weak target and the background in the forward-looking sonar (FLS) image makes conventional target detection algorithms ineffective at detecting weak targets. This paper proposes a high-precision target detection method that effectively combines the k-means clustering algorithm with the optimized Average Absolute Gray difference (AAGD) algorithm to detect weak targets in a high-intensity background noise environment. The proposed method introduces a multidirectional gradient composite window and adaptive threshold to the AAGD algorithm, overcoming the performance attenuation associated with high-intensity sharp edges. By processing the acquisition FLS of the water tank and the lake experiments, the results show that compared with a number of conventional detection algorithms, the algorithm in this paper is less affected by high background noise and can effectively capture more weak targets, confirming the significant performance improvement of the algorithm in weak target detection.