Abstract: A new automatic urban area extraction method from high-resolution remote-sensing imagery that exploits the unique local features of urban area is presented in this paper.The proposed algorithm contains the following steps: First,it obtains the filtering response images with Gabor filters grouped at various central frequencies and orientations;Secondly,we use the Ostu’s method to implement threshold segmentation,and then realize the logical and operation in the various orientations of every central frequency;Thirdly,we determine the optimal central frequency with the Gabor features distribution information;Finally,with the information above,we extract the urban area by forming the spatial voting matrix with the Gaussian function.Experimental results show that the approach is able to detect the urban areas in the high-resolution remote-sensing imagery.The results of a performance evaluation also support the high precision of this approach.