A Review of Underwater Acoustic Target Detection and Recognition Technology Based on Information Fusion

‍ ‍The competition for initiative in underwater space was becoming more and more fierce. In view of the complex Marine environment and diverse target types， the detection capability of the existing underwater platform and single sensor was limited and the stability was poor， which was difficult to meet the needs of accurate detection and identification of underwater acoustic targets in the future. With the increase of underwater target detection methods and means and the rapid development of multi-source information fusion technology， the research on underwater acoustic target detection and recognition technology based on information fusion theory can achieve more accurate state estimation and attribute discrimination of underwater acoustic targets， and thus improved the underwater detection and early warning ability. Firstly， the existing detection sensors and detection platforms above and below water were briefly introduced. On this basis， the network composition and structure characteristics of the stereo observation system of underwater acoustic target were discussed， and a cross-domain heterogeneous stereo observation system design was proposed. According to the existing fusion structure， the functional framework of the multi-information source based underwater acoustic target observation system was given. Then， according to the specific application of underwater acoustic target fusion detection technology， the fusion detection method based on target signal level and non-target signal level， the fusion location method based on space-time characteristics， motion characteristics and Marine physical field， and the underwater acoustic target tracking method based on classical nonlinear filtering， data association and random set theory were discussed and analyzed respectively. The problems faced by each method were analyzed briefly. Finally， the paper focused on the summary and analysis of the physical field characteristics of underwater acoustic targets. In view of the uncertainty of target information， three typical fusion recognition methods and frameworks based on Bayes theory， D-S evidence theory and deep learning algorithm were summarized， and their development status was analyzed. The advantages and prospects of underwater acoustic target detection and recognition technology based on information fusion are pointed out， and the problems and challenges in underwater stereo observation network， fusion structure， fusion detection and recognition algorithm and engineering applications are analyzed.