Research On Application Of Mathematical Morphology Algorithms For Image Processing

With the growth of the requirements for developing and exploration of the resources in the ocean, kinds of technology to detect objects underwater have been developed rapidly. As the most important information sources, video information became one of the focuses in the research of underwater objects' detection. The advanced imaging sonar has been the primary tool to do objects' detecting and recognizing underwater because of its wider imaging range and stronger anti-jamming capability.In this paper, the work is a part of acoustic vision in the Intelligent Autonomous Underwater Vehicle. The aim of the task is researching the algorithms for the preprocessing and recognition of underwater acoustic images. Even if numerous methods, but most of them are linear processing which have the side effect of bluring edges, so can not meet the case of detecting edges and suppressing noise. And the use of some nonlinear methods we have had are also limited for the lack of theoretical basis.With the characteristics of the task of preprocessing and recognition of underwater acoustic images, the nonlinear methods based on the mathematical morphology was proposed here. The performance of it is better than the conventional methods for the case of suppressing noise, detecting edges and recognition.In this paper, we introduced the binary morphology and its improvement, gray-scale and color morphology and analyzed their characteristics in filtering particularly. Then we simulated and compared the classical image processing and recognition algorithms adequately, obtain the useful conclusions and many processing results.Finally, we constructed a software which included some linear and nonlinear algorithms along with the mathematical morphology for the preprocessing and analyzing in underwater acoustic images.