Automatic Detection Technology Of Sonar Image Target Based On The Three-Dimensional Imaging

In recent years,countries around the world have intensified their research and exploration of the ocean.Sonar technology has received increasing attention.Compared with the so-called sonar equipment,the related technologies of sonar target detection and recognition cannot be matched.China’s research on the detection of sonar targets is relatively late,but it is also catching up.Hangzhou Institute of Applied Acoustics has developed a three-dimensional imaging sonar device based on linear array synthetic aperture.The amount of data has increased dramatically,from two-dimensional to three-dimensional.If only relying on manual detection,it is not only inefficient,but also conducive to the storage and maintenance of data..Therefore,it is particularly important to study the correlation algorithm based on three-dimensional imaging for automatic detection of sonar image targets and to develop an automatic detection system for sonar targets.In this paper,the algorithm of sonar image based on 3D imaging is studied,including preprocessing,image segmentation and target recognition.Finally,the automatic detection system of sonar target is designed and implemented.The research contents and results are as follows:(1)Study the method of sonar image preprocessing.The preprocessing of sonar images mainly includes denoising and whitening.In the aspect of denoising the sonar images,three classical image denoising algorithms are introduced,and the applicable scenarios and advantages of these three denoising algorithms are analyzed.At the same time,the sonar image is whitened,and the background is equalized,which inhibits the interference of the seabed reverb to some extent.Aiming at the characteristics of images acquired by linear array synthetic aperture sonar equipment,this paper adopts median filtering denoising to denoise the three-dimensional sonar image,and gives the algorithm principle and basic steps.Through the comparative analysis of experiments,it is used.Value filtering denoising can get good results.(2)The method of sonar image segmentation is studied.In this paper,the segmentation of sonar image mainly includes two aspects.One is to separate the water body and the stratum from the sonar image,and divide the two parts with larger difference.The second is to study the image segmentation algorithm for water image and stratigraphic image.The OTSU algorithm is used in the water part of this paper,and the two-dimensional maximum entropy algorithm is adopted in the stratum,which lays a foundation for effective feature extraction.(3)The method of target recognition of sonar is studied.This paper presents three classical target recognition methods,including sonar target recognition based on multi-dimensional Bayesian classification model,target recognition based on artificial neural network and target recognition based on support vector machine,and experiments on three recognition methods.analysis.(4)Design and implement an automatic detection system for sonar targets.Through the research and analysis of this paper,a set of algorithm based on three-dimensional imaging sonar image data is integrated,and the sonar target automatic detection system is designed and implemented strictly according to the relevant specifications of software engineering.Through actual verification,it has the characteristics of simple interface and simple operation.It has the advantages of high detection rate and low false alarm rate in sonar target detection and recognition.