Methods Research Of Extract Road Information Based On Remote Sensing Image

With the development of aerospace science and technology level, the resolution of image obtained by remote sensing satellites is getting higher and higher, and the ground feature is also getting more and more. Extract and process the ground feature in the remote sensing image is the main research direction in the remote sensing application. Road as a ground feature has obvious feature in the remote sensing image, road extraction is always one of hot points in the application research, and the extracted road information can be applied to map update and automobile navigation and so on.This thesis first uses the method which combines the gray level histogram threshold segmentation with the mathematical morphology to extract road information when the gray level histogram of remote sensing image has obvious features. When the histogram does not have obvious feature and there are many different types of road in the image. Due to different types of road have different grayscale, use the method which combines Mean Shift smooth and segmentation with the gray level histogram multi-threshold segmentation to extract road information, obtain the initial road by choosing many grayscale ranges from the gray level histogram, and then get road information by Mathematical Morphology processing.In the high resolution remote sensing image, the differences in texture of road and the interferences which the road suffered are also become obvious, as starting points, this thesis studies the theory of fuzzy C mean(FCM), uses FCM algorithm to cluster and segment the ground features in the remote sensing image, and choose the grayscale value which represents the road to extract road information by threshold segmentation. Because the FCM algorithm has large dependence on the initial input parameters in the implementation process, and the difference of road texture distribution will make the road fracture after use FCM clustering segmentation process the image, this thesis uses the method which combines Mean Shift algorithm with FCM clustering segmentation to extract road. According to the grayscale distribution in the gray level histogram after Mean Shift segmentation to choose the clustering number of FCM clustering segmentation, and then get the road information by histogram threshold segmentation.