Research On Mosaic Method Of Side-scan Sonar Image

With the continuous development of society,the demand for various resources is also increasing,and the ocean contains rich natural resources.Therefore,the exploration and exploitation of marine resources is the trend of development.However,compared with the land,the environment in the ocean is very harsh,and the visible and infrared detection methods are greatly limited in the sea,so the sonar detection technology has become the main means of ocean detection.Sonar equipment emits sound waves,and collects information by recording the intensity and time of the returned sound waves.We analyze sonar data,convert it into images,and then get a large submarine topography map through image mosaic.Aiming at the existing problems of side-scan sonar,this paper analyzes the sonar data and proposes a method of side-scan sonar image mosaic including image preprocessing,image matching and image fusion.And it was verified by experiment.The specific contents of this paper are as follows:Firstly,the preprocessing method of side-scan sonar image is studied.Due to the influence of marine environment,attenuation of acoustic signal,change of its own velocity and height,as well as the sonar data acquisition and recording mode and other factors,the collected side scan-sonar images have some problems such as noise,geometric distortion and gray level distortion.In this paper,the original side-scan sonar data is analyzed and sampled,and the image is filtered.Aiming at the geometric distortion of side-scan sonar image,geometric correction methods including correction of slant distance of side-scan sonar image,and velocity correction are studied.Based on the analysis of the current image gray-scale correction methods,a side-scan sonar image gray-scale correction method based on the Generative Adversarial Networks(GAN)is proposed,and the performance of the algorithm is verified through experiments.After preprocessing,the side-scan sonar image can reflect the seabed topography and landform more truly,and then match the image features.This paper studies common image matching methods,such as Sift,Surf and Orb.However,these methods are not suitable for the matching of side scan sonar images due to the large amount of noise in the images and the serious deformation of the target in the images.In this paper,an image matching method based on position information and grid structure is proposed.The longitude and latitude information recorded during the acquisition of side scan sonar image is used to divide the image by grid structure,and then the side scan sonar image is matched.And the effectiveness of the proposed method in sonar image matching is verified by experiments.Finally,the study of side-scan sonar image fusion methods,finally proposed a maximum multiband fusion and fusion methods of sonar image fusion,the algorithm is very good keep the details of the sonar image,and can solve the effects of shading image,the experiment proved that the method can effectively realize the side-scan sonar image fusion.In this paper,a series of image splicing methods including image preprocessing,image matching and image fusion are proposed,and the effectiveness of the proposed method is proved by experiments.