Sonar Image Segmentation With Three-dimensional THC Entropy Based On Genetic Algorithm

The role of sonar technology in the field of ocean development has become more and more important,and more and more researchers have devoted themselves to this direction.Sonar image processing is one of the key technologies,and the process can be divided into filtering of sonar image,sonar image segmentation,sonar image feature extraction and target recognition.How to accurately segment sonar images has become one of the important research contents in sonar image processing field.Sonar images can be divided into target bright area(bright area),target dark area(dark area)and submarine reverberation zone(reverberation zone)according to its own characteristics.The bright area and dark area are commonly studied by people.The reverberation zone is generally the background area formed by complicated seabed environment.The purpose of sonar image segmentation is to divide the bright area,dark area and reverberation zone completely and accurately.In this work,the sonar image segmentation method is studied.The main contents are as follows:Firstly,this paper briefly introduces some common methods of filtering and denoising images,and uses Wiener filtering to reduce the noise contained in sonar images.The core idea of Wiener filtering is the minimum mean square error criterion.Wiener filtering is a traditional linear filtering method,which can keep the details and edge information of the image completely.Secondly,the THC entropy segmentation method based on histogram is used to segment sonar images.THC entropy is extended on the basis of Shannon entropy,and the parameter introduced shows that the degree of non expansion of the system,which is non additive.THC entropy considers the interaction between the two subsystems.In this paper,the THC entropy image segmentation method is extended from one dimension histogram and two-dimensional histogram to the threedimensional histogram,and the local spatial gray information of the image is further utilized,but the computation exponentially increase.Then,the THC entropy image segmentation method based on three-dimensional histogram is time-consuming.In order to reduce its operation time,this paper uses genetic algorithm to optimize the THC entropy sonar image segmentation method based on three-dimensional histogram.Finally,the two-dimensional and three-dimensional histogram are constructed with different features to segment the sonar image.The influence of histogram constructed by different features on histogram THC entropy image segmentation is discussed.