Automatic Detection Techniques Of Underwater Small Targets Based On High-Resolution Sonar Image

In the application of economic construction and military activities,automatic detection of small underwater targets such as mine or frogmen is a popular research direction.This paper studies the automatic detection technology of underwater small targets based on high-resolution sonar image and realizes the automatic real-time detection of low false alarm and low miss detection of small underwater target.The technology is divided into four parts: preprocessing,multi-algorithm fusion detection,region of interest(ROI)segmentation and feature-based automatic detection of underwater small target.First,this article briefly summarizes the background and significance of automatic detection of underwater small targets,and introduces in detail the detection,segmentation and recognition methods of underwater small targets.Secondly,in order to ensure the real-time output of the detection results,downsampling technology is introduced to improve the computational efficiency of the algorithm.In order to reduce the missed detection rate,the Gamma transform algorithm is introduced to suppress the background noise of sonar image.In order to reduce the false alarm rate,the 2/2 fusion of Fuzzy c-means clustering and k-means clustering is introduced to reduce the inconsistent error detection caused by the different detection mechanisms of different algorithms;morphological operation and Hough transform were introduced to carry out spatial clustering of the detection results,and most of the non-target detection points were removed.In order to achieve ROIS automatic segmentation under low SNR,a pulse-coupled neural network combined with Fuzzy c-means clustering was introduced to separate the target and background with similar gray level.In order to further reduce the false alarm rate,the Fisher weighted scoring algorithm is introduced to unify the recognition and detection,and the identified features are used for detection.Through the above algorithm,the automatic detection of small targets under low signal-to-noise conditions is realized.In this paper,four data sets are obtained through field experiments and computer simulations,which verify that the algorithm can automatically detect underwater targets in real time under low signal-to-noise ratio conditions,and can achieve low alarm omission rate and low false alarm rate.To sum up,this article has certain innovation and engineering practice in the research of automatic detection technology of underwater small target high frequency imaging sonar.