| In recent years,the protection of maritime rights and interests has always been one of the hot issues in the world.The study of effective detection technology for seabed artificial targets has a very important practical significance to realize China’s maritime rights and interests.Effective interpretation of sonar image data is the main means of detecting seabed artificial targets.Currently,the identification of seabed artificial targets in sonar images mainly relies on the relevant technical personnel to judge through visual experience.However,with the development of large-scale sea sweeping,a large number of sonar images to-be-recognized will be produced in a short time on the relevant work platform.The mere interpretation of a large number of sonar images by manual visual observation is undoubtedly difficult to meet the needs of large-scale seabed detection operations at this stage.For this purpose,the paper designs a set of interactive sonar point cloud image analysis and processing process and studies data processing related denoised technology,simplification,hole detection and repaired,etc.on the original image of the seabed sonar point cloud.The method applied in this paper reduces the amount of point cloud data as much as possible while maintaining the original seabed features,reduces the impact of noise points on point cloud data,and effectively decreases the hole effect generated in the data.The point cloud processing method studied in the paper provides high quality data assurance for seabed artificial target detection algorithm.The main work can be summarized as follows:(1)The characteristics of side-scan sonar data and multi-beam sounding system sounding data are analyzed,and the two data is fused to form four-dimensional point cloud data.On this basis,the double-layer noise point cloud filtering algorithm is used to achieve effective removal of noise points.(2)A partition reduction method based on the characteristics of sonar point cloud is proposed.Firstly,the method uses the region growing method to divide the point cloud,and divides the data into flat regions and non-flat regions.Secondly,according to the geometric characteristics of the sonar point cloud,this paper calculates the characteristic parameters and feature thresholds required for data reduction.Finally,the effective reduction of point cloud data based on discrimination between non-flat areas and flat areas is achieved.The experimental results show that the method could effectively reduce the amount of point cloud data based on the effective maintenance of point cloud features.(3)The point cloud hole repairing algorithm based on local expansion of concentric circles is improved and applied to the repair of four-dimensional seabed sonar point cloud holes.Firstly,the point cloud boundary is identified based on the point-to-neighbor relationship method.Secondly,the point cloud clustering algorithm based on Euclidean distance is used to distinguish and homogenize the boundary and outer boundary of the point cloud.Finally,the improved concentric circular local expansion algorithm is used to repair the four-dimensional sonar point cloud hole.The experimental results show that the proposed method could effectively repair the different sizes of holes generated by the four-dimensional sonar point cloud data reduction process.(4)A seabed sonar point cloud data generation and processing system is designed and implemented,which combines with VS and MFC programming technology to complete the software platform construction,and applies VTK point cloud display tool to realize the rendering and display of point cloud.The system realizes the basic operations of point cloud processing and analysis such as rapid reading and display of seabed sonar point cloud,interactive control,point cloud editing and point cloud data batch generation,and further realizes point cloud streamlining and point cloud hole boundary identification and complex operations such as patching and point cloud surface reconstruction. |
PDF Full Text Request