Algorithm Research Of Digital Image Processing Based On Gray System Theory

With the image technique more increasingly extensive application, the image segmentation technique is becoming more and more important. The image segmentation is the most basic and the most important technique, it is an important and difficult in the image-processing realm, it also is one of the bottlenecks of image theories development. The quality of image segmentation directly influences the quality of image analysis, image understanding and image recognition. So image segmentation is still an essential and important link to any system of comprehension and system of automatic object identifies. Introduction depict that image segmentation have a important station in digital image processing and this paper's significance of research, summaries general situation of technique development and present condition and existent problem of image segmentation. Then, this paper introduced the basic knowledge of the gray correlation analysis, this paper emphasized to expatiate the definitions and theorems of absolute worth correlation degree. The knowledge of basic theories establish basement to application of image enhancement and image segmentation. The paper detailed introduced filter methods, put forward improved medium filtering algorithm efficiently, give a new gray correlation degree mean filter algorithm and realizes, the result of experiment express the new algorithm not only is more effective but also can keep the more details of image than the traditional algorithm. This is an important innovation of this paper. In forth chapter, this paper introduce the development and classification of the artificial nerve network and emphasize a new segementaion technology of combing gray and artificial nerve network. In recent years, with putting forward new methods of intelligence calculation, evolving calculation, creature calculation, algorithms of image segmentation combining the particular theories are becoming research hotspot. A lot of particular theories will appear shortage in applied process like this or so, the paper put forward to adjusting the number of implicit node and realize optimization ability to get better segmentation result. Aimed at implicitnode of error Back Propagation network is too large or too small to cause the network optimization to descend, and the shortage of image segmentation result that affected, the paper put forward to use the thought of gray correlation into network train process to regulate the number of the implicit node to realizes the ability of superior of the network to attain the better segmentation result. At the same time the superior ability and the stick property of network oneself are strengthened consumedly. Then, many examples show new algorithms are better than traditional algorithm. And the reasons to cause shortages of algorithms are analyzed in the paper. This chapter is important and innovation.At last, the paper summaries the fruits and shortages of this paper. The author discussed the viewpoint of research in this realm and talked about the experience of oneself in period of doing this paper.