| Roads are the main body of modern transportation system. And they are very important to national economy and the foundation geographic information construction. In optical remote sensing, road extraction is a classical subject, and it has made many achievements. But SAR image cannot achieve good results when applying them directly. As a result, SAR image road extraction has become an important research content in image information extraction.According to the shape, gray scale and polarimetric coherence characteristics, this paper puts forward a SAR image road extraction method by intergrating polarimetric coherence optimization and angle information. This method combines the road information from polarimetric coherence optimization classification with the gray scale from angle information extraction as local features, and introduces trajectory model as global feature. Then use them to realize SAR image road extraction.In this paper the work and innovation mainly includes following aspects:(1) According to the polarimetric coherence optimization principle of PolInSAR, the process of polarimetric coherence optimization was done to analyze the polarimetric coherence optimization coefficient distribution of several typical features. Then compare with several kinds of typical classification methods to determine using polarimetric coherence optimization classification in this paper.(2) Choose the road center points for different mathematical model fitting experiments, through comparative analysis to determine using Fourier function as trajectory model in this paper.(3) Combine the road information getting from polarimetric coherence optimization classification with the gray scale of angle information extraction, then using it for road extraction in the process of rectangular template direction judgment.(4) Define the initialization, forward, rotating and end strategy, using Matlab software as the platform, then realize SAR image road extraction. |
PDF Full Text Request