The Segmentation And Realization Of High Spatial Resolution Remote Sensing Image Based On Region Growing Algorithm

High spatial resolution remote sensing image with characteristics of rich detail information and clear geometric structure has favorable expressive force of ground object,it is improper approaches for image processing of pixel-based,so object-oriented method emerges as the times require.Image segmentation was at a crucial stage from pixel-based to object-oriented,the segmeted regions are fundamental description of image objectification.But because of the uncertainty of image segmentation and complexity of high spatial resolution remote sensing image,there is no reliable model to guide the image segmentation and the study of remote sensing image segmentation field has been concerned as an challenge job until now.This thesis has studied the theory of region growing algorithm,then programmed to segment high spatial resolution remote sensing image.The work and contributions in this thesis are as follows:(1)The study and programming of region growing algorithm. On the basis of the principle of region growing ,region growing algorithm was designed and developed on MATLAB software platform.The algorithm consist of region growing and region merging. Carrying out working of region growing, the position of seeds were continuous selected, growing criterion was confirmed and growing pattern was designed;performing the working of region meging, the undesired regions produced by region growing were merged into the adjacent region according to similarity criterion.After segmenting the image, some attributes was selected and computed to describe every region.at last,the algorithm was apply to segment high spatial resolution forest zone remote sensing image and the results of segmentation was evaluated by contrasting realization of forest zone.(2)Object edge extraction and vectorization.In this thesis,there were three edge detecting and extraction approaches that are gradient-based,segmented region-based and pixel color sharp variation-based.These approaches were put into use to extract the edge of the segmented image,every approaches has extracted complete and right edge.At last,the edge was vectorize by the module of raster to vector which was developed by using COM component technology.The experiments demonstrate that segmented result was good by using region growing algorithm to segment high spatial resolution remote sensing image; the edge of segmengted image was extracted and vectorized subsequently,it is foundmental of future data analysis.