A Method for Road Area Detection in High Resolution SAR Images

Automatic extraction of road from satellite images is one of the most important researches in the field of remote sensing. The method proposed in this study is based on a fuzzy method for detection of road areas from high resolution SAR images. In this method, the multiple features are extracted first based on the backscatter coefficient of each pixel and its neighbor pixels from the input image. The extracted features are combined with each other in the next step using a fuzzy algorithm and the desired road areas are selected separately in the last step considering the spatial and spectral criteria. The favorite results and root mean square of 78% were obtained by applying this algorithm on high resolution SAR images obtained from the TerraSAR satellite.