Urban Road Extraction High resolution remotely sensed imagery with Gabor texture and geometrical Features

Through the feature analysis of the frequency curve , The paper proposed a new method which integrated Gabor texture and geometrical features in the urban road Ction from high resolution remotely sensed imagery. in the experiment, the texture features in different frequencies and Different directions were obtained by a given bank of Gabor filters,andthen K-means clus Tering method is applied for imagery segmentation. The morphological methods were utilized to separate the road objects from Non-^oadobjects,and the geometrical features were applied to take out the non-road. Then the extracted roads were regulated by mathematical morphology. The result showed this proposed method could effectively extract the urban road information from high resolution RemoTely sensed imagery.