Research On The Segmentation And Artificial Target Extraction Methods For The High Spatial Resolution Remote Sensing Images Based On Visual Perception Principle

With the continuous development of sensor technology, in recent years, the ground resolution of remote sensing images has been improved continually. As a result, the ground details description ability of remote sensing image has been improved, and more clear ground details could be shown. Duing to its high spatial resolution and few spectral resolution, the traditional remote sensing image processing technology can not provide a good analysis and processing for the high spatial resolution images, which have brought a new challenge for the classic remote sensing image analysis technology.However, with the huamn visual observation, the ground targets and complex details could be easily recognized. The huamn visual perception could perceive and cognize the high spatial resolution images from the perspective of color, texture, and shape. Inspired by this phenomenon, this paper manage to simulate the human visual perception for the analysis and recognition for the high spatial resolution images. In particular, the main research contents are as follows:1 This paper propose a segmentation method for high spatial resolution remote sensing images based on the classic visual perception theory. This method simulate the classic human visual perception on the basis of the research of visual neural mechanism. And then, by utilizing the unsupervise mathematic and image processing tools, our method analyze the remote sensing image by its color, texture and detail features. Acoording to the practical test, the laws of estimating the parameters' value are revealed here.Compared with the traditional segmentation approach, the segmentation experiment results have proven that our method is reliable and effective.2 This paper proposes a good road extraction method for high spatial resolution remote sensing images based on the unique shape characteristic of the roads. According to the linear shape characteristics of the road on the remote sensing image, the objects suspected to be roads are analyzed for road extraction with the principle of gestalt visual perception theory. Moreover, based on the practical test, parameters'value estimation laws are concluded. The road extraction experiments have proven that the proposed road extraction algorithm is fast, reliable, and effective.3 This paper propose a shape analysis method for the building in high spatial resolution remote sensing images. On the basis of human visual perception, by multiple shape parameters, the shape of buildings could be cognized in this paper. After the test of parameters' performance, the parameter selection method is summaried. Building recognition experiments using real city high resolution aerial images have demonstrated that the proposed method can simply and accurately analyze and identify the buildings with their shape charateristics for in-depth understanding.