The Study On Image Post-processing Scheme For 3D Imaging Sonar

With the development of marine, the technologies of underwater exploration obtains more and more attention and imaging sonar is the one of the most important technologies, which is usually used for underwater objects detecting and terrain mapping. In order to improve the detection precision and visual effect, three dimension imaging sonar emerges as the times require.Hangzhou Applied Acoustics Research Institute developed a new three dimension imaging sonar based on linear array synthetic aperture technologies. As a novel imaging sonar, it has many advantages over other imaging sonar, but it is still lack of pertinent post-processing and visualization scheme, which makes such an advanced imaging sonar not put into application. According to this practical need, this thesis designed and implemented a fast and effective post-processing scheme by studying the traditional image processing methods and combining them with the characteristics of three dimension sonar images.Firstly, according to the characteristics of three dimension imaging sonar data, the design concept of the whole system was set up, and the detailed description of design module was-built. In order to handle the three dimension imaging sonar data properly and achieve better system application interface, this system was divided into three modules, such as post-processing module, display module and file read-write module. The interactive operation was designed a good interface for integrated application.Secondly, the image segmentation, enhancement and visualization algorithm involved in post-processing module were described in detail. A note about whether the algorithm suited for being accelerated by GPU programming and how to accelerate it was appended in the end of each algorithm.Finally, the function and performance of this system was completely tested via experimental data. The results shows the system designed and implemented in this thesis can handle image data generate by three dimension image sonar correctly, and the operation efficiency of visualization is fast enough for real-time sonar imaging.