Intelligent Extraction Method Based On Remote Sensing Images Of Natural Selection Algorithm On The Road

With the rapidly development of space techniques and information techniques, we have got large amount of geographic data, Target extraction from images has become the important means of the space information updating. Thus, automated information acquisition from images has become a bottleneck, which needs to be overcome. As one of the key issues, semi-automatic road feature extraction has attracted wide attention, in the national economic production and military target detection domain this research has a matter of great significance on both theory and praxis.The paper has studied the character of the road in the remote sensing image and the technique about the extraction of the road in our country and abroad. At the same time, it considers all kinds of cases generally. A new extraction model of the road is realized, whose key thought lies in the transformation from intelligent extraction according to road axis model to graphical theory model, and then, through searching the shortest path in graphical space, we get the road axis. In the large scale space we bring forward a new edge extraction algorithm and improve the self-adapting smoothing algorithm to pre-process the image, moreover, we discuss the transformation from intelligent extraction to graphical theory. In this paper, taking the advantage of GA and Bacterial algorithm we bring forward the natural selection algorithm to find the graphical optimization path. After rough-extraction, In the small scale space we optimize the road extraction result by snakes algorithm. The test shows the algorithm can extract the central line of the road from aviation image much faster and more robust.