The Automatic Extraction Algorithms Used in Remotes Sensing Images about Changes in Wetland Edge

Abstract: In recent years, the wetland resources have been seriously damaged by natural factors and human factors, facing with tremendous pressure from that. At the same time, with the high-resolution satellite remote sensing has been widely applied, the latest remote sensing technology can be used on the management, protection and dynamic monitoring of the wetland resources. But how to get the features extraction of the massive remote sensing images automatically or semi-automatically has become a key issue in remote sensing applications. Because of the complexity and diversity of automatic extraction in remote sensing, it involves computer vision, artificial intelligence, pattern recognition and image understanding and so on. So by combining the human's "recognition" capability and the computer's "quantity test" and "positioning" capability, the human-computer interaction method for automatic extraction of remote sensing images is an effective way.In general, features extraction is generally divided into three categories, namely: point-like feature extraction, linear feature extraction and facial feature extraction. Especially for some surface-like surface features in wetland remote sensing images (lakes, artificial reservoirs, marsh meadow wetland, etc.) Because of the color characteristics of image is relatively uniform in image, by image segmentation and edge extraction, target objects can be separated from the background and be extracted finally.In this paper, by using the single and multi-channel remote sensing image segmentation method, the color remote sensing image can be changed into binary color images, then get separation of the background and target objects finally. On this basis, by image subtraction of two remote sensing images in the same location to obtain the changes polygons, using multi-structure elements of mathematical morphology algorithms to get the information of the changed edge.As to improvements in the algorithm, in the single-channel image segmentation algorithm, using an improved RGB transfer to HSV color space algorithm, to make up the shortcoming of original algorithm, which are the individual points are not defined and the color level is not obvious. In the multi-channel segmentation algorithm, based on Quaternion complex moment invariant, image segmentation processing effects can be significantly improved. At the same time, the proposed multi-angle multi-structure elements Morphologic Edge algorithm has an universal generalization in the edge extraction of remote sensing images…
Key words: remote sensing; image segmentation; edge extraction