Semi Automatic Road Detection from Satellite Images

Automated extraction from high resolution satellite images provide an opportunity to extract road area, which are useful for vehicle navigation. However, factors like occlusion due to trees along the road, building shadows and vehicles, variation in road surface characteristics and changing road geometry, which become more apparent in high resolution satellite images, make road extraction a challenging problem. In this text three different road detection methods are studied. These method are Clustering, Directional Morphology, and Adaptive threshold, and are compared on the basis of different parameters like completeness, correctness, and quality. In terms of time processing the directional morphology method take more time to process the image as it is pixel based operation, it considers 8 direction for segmentation but produces higher value of correctness. The clustering method takes less time compared to other method, it's time for segmentation is depends on the number of iteration it take and varies according to the type of image set. The large iteration count produces better results but takes more time to segment image. The adaptive thresholding method requires very less time.