The Research Of 3D Sonar Image Mosaic Method

With the continuous development of human society,land resources begin to be increasingly scarce,and the ocean covers most of the earth’s surface,is a natural treasure house of mankind,the exploitation and utilization of Marine resources will become a new development opportunity for all countries in the world in the future.At the present stage,the remote imaging detection of the seabed is mainly conducted through sonar equipment.For the sonar data obtained by the side-scan sonar equipment,the waterfall image does not take into account the actual geographical position relationship.The sounding data obtained by the multibeam bathymetric equipment can reflect the detailed terrain information of the seabed.Most current studies are respectively for 2D sidescan sonar image or 3D depth image splicing independently,two-dimensional sidescan sonar image by color information to the bottom of the sea echo intensity,but unable to get the corresponding height information,and the three dimensional depth image can reflect the topography information,but could not judge information corresponding to the ocean floor.In view of the above problems,this paper studies the Mosaic algorithm of three-dimensional sonar images combined with side scan and sounding,so that the obtained images can simultaneously reflect the topography and geomorphologic information of the seabed.The main work of this paper is as follows:First of all,in the process of acquiring sonar image data,the data collected by the sonar equipment is not all information of seabed echo intensity,but also some other interferences in the ocean,such as environmental noise,self-noise and reverberation.Affected by these noises,some details in the sonar image will be covered up,the feature information will be relatively weak,the edge will be weakened or even discontinuous,thus reducing the image contrast.Therefore,this paper analyzes the collected original sonar data to obtain the sonar image,proposes an improved sonar image denoising algorithm based on 3D block matching algorithm,and gives corresponding experimental results.Secondly,after the pre-processing of sonar images,such as denoising and slant distance correction,the obtained sonar images will have fuzzy blocks in the middle region.Moreover,there may be some differences in the acoustic intensity obtained from the starboard and starboard sides of the sonar equipment,which will lead to more obvious gray difference in the middle region of the image.In order to solve this problem,this paper puts forward the method of deep learning to repair the middle region of side-scan sonar image after slant correction,so as to eliminate the fuzzy block in the middle region of sonar image,make the middle region of sonar image smoother,and ensure the integrity of the whole sonar image information.Thirdly,the location,course,sounding,attitude and other information obtained after the analysis of sonar data were used to study the construction of three-dimensional sonar point cloud,and the interpolation and fusion algorithm of side-scan image and sounding data was studied on the basis of each Ping,so as to construct the three-dimensional sonar image corresponding to the real seabed information.Finally,because in the underwater sonar equipment scanning characteristics,can cause two three-dimensional sonar point cloud is the area of overlap,and because the sonar equipment when measuring error or affected by external interference,when scanning the same area to get two points cloud has not completely aligned,could lead to can’t build a complete terrain,so to the point cloud of overlapping point cloud registration,the incomplete overlapping point cloud as aligned as possible,and then the fusion processing of point cloud overlap part,the final triangulation constitute a plane,so as to construct reflect seafloor topography three-dimensional sonar image.In this paper,the three-dimensional sonar image Mosaic method is systematically studied,the overall design of the Mosaic algorithm is completed,and the submarine topography is drawn,thus providing an effective method for seabed topography detection.