| In order to better reflect the current situation of underwater structures,new requirements are put forward for the precise detection of underwater objects,and higher requirements are also put forward for the data processing.In the underwater target detection mode,the 3D scanning sonar technology based on the acquisition of point cloud data has more advantages than the traditional underwater measurement method,with high efficiency,reliable method and excellent results quality.By acquiring a large number of point cloud data and processing,it can accurately display the detailed characteristics of the measured object,which is more helpful for later research and analysis.But because of its huge data,the processing will become complex.Improve the detection technology and the accuracy of 3D point cloud data processing,so as to obtain more accurate shape and structure of underwater objects,so that underwater 3D scanning technology can be better used in marine engineering construction,has become an unavoidable problem.In this thesis,the following in-depth study and research are carried out on issue:(1)This thesis discusses the principle,work flow and the method of underwater fine detection of multibeam sounding system,3D real-time imaging sonar system and 3D scanning sonar system.(2)Based on the three-dimensional scanning sonar,the noise is divided into two types:large-scale noise and small-scale noise.Firstly,statistical filtering,cloth simulation filtering and radius filtering are used to remove the large-scale noise,and then fast bilateral filtering is used to smooth the small-scale noise;(3)By using the method of human interaction,the point cloud data is first roughly spliced,and then the data is finely spliced by ICP algorithm,which realizes the conversion between different coordinate systems.The method is simple and effective,which not only solves the problem of smooth denoising of data points,but also effectively maintains the typical geometric characteristics of point cloud data,and provides high-precision and efficient point cloud data for subsequent analysis.The method has been tested on Haihe estuary hydrological base and Wuhu rail-cum-road Changjiang river bridge.The results show that the precision of point cloud processing basically reaches the expected target of underwater fine detection. |
