The Research Of The Road Network Extraction For High Resolution Remote Sensing Images

With the rapid development of space technology and sensor technology, remote sensing information has become one of the main sources for geographical data. But our knowledge and application about it are way behind with collecting of the information sources of the space and aerial system. The development of high resolution remote sensing images makes their deep application possible. Road network extraction has been an important part of the extraction of remote sensing information and got wide attention, which has a matter of great significance in regional planning, traffic navigation, video matching and military spying.The paper introduces the background of research, the research status and development trend of the road extraction. At the same time, for the study of the rural road network, their features are analyzed and summarized, and then an effective automatic extraction method is presented based on feature models. Firstly, with the guidelines of multi-resolution technique, the linear features in the reduced resolution image are extracted by differential geometric approach, which indicate the existence of roads. Secondly, according to radiometric and edge features, road parts are built in high resolution image by the use of edges and parallel lines. Thirdly, road segments are made by combining road parts. Fourthly, the paper introduces briefly road detection based on snakes and through it road segments are connected. Finally, by using information of road segments and the junctions of ridge lines, the algorithm of extracting road junctions is established, which complete the extraction of entire road network. The experiment results show the method proposed by us can extract accurately rural road network and they can also be suitable for urban road extraction if they are studied further.