| Since on October4,1957, the first satellite launch, the satellite remote sensing also appears,with the continuous development of the aviation and aerospace techniques, the Earth’s surfaceinformation acquisition has always been a constant pursuit direction of human, damage detectionafter disaster、urban planning、get urban road information and so on, of all these applications themost versatile is get urban road information. With the development of remote sensing technology,the resolution of remote sensing images becomes higher and higher, high-resolution remotesensing images provides a wealth of ground information, but it makes too much disturb becauseof the high resolution of remote sensing images, the driving car on the roads, pedestrians on theroads, the trees on both sides of the road, and the shadows of the building and trees, as well assome interference similar to the characteristics of the road surface, such as parking, wide roof,which gave great trouble on road detection,The main work of this article is to detect the edge information of the image, we studied somecommon image edge information extraction and classification algorithms, such as Roberts、Prewitt、Sobel algorithms based on first-order differential and canny algorithm、the watershedsegmentation edge information extraction method based on second-order differential. Simulationresults were compared and the advantages and disadvantages of the different algorithm of imageedge detection can be seen from the comparison results. Later we introduced a segmentationalgorithm based on k-means clustering and get the road information successfully, but the result isnot very good because of the interference near the road edge. Then we applied the kernel-basedlearning methods to K-means clustering, and get better results relative to K-means clusteringalgorithm. In the last work we get the exactly road information by the Mathematical morphologytheory and the Region characters of images methods. |
PDF Full Text Request