Research on Road Extraction from High-Resolution Remote Sensing Image based on Improved MRF

Abstract

Extracting roads from high-resolution remote sensing images have always been a challenging problem.In order to overcoming the shortcomings of current road extraction methods for high-resolution remote sensing images, this paper proposes an improved road extraction method which is based on MRF. Firstly, double threshold segmentation is used to preprocess remote sensing image that is in order to remove the shadow on the image. Nextly, the MRF feature field model is constructed which is based on the multi-feature of road, the energy function that is based on MRF characteristic field model is minimized by graph cutting method whose aim is to extracted road from the image segmentation object.Finally,the postprocessing is carried out by mathematical morphological operation.Through experimental verification and evaluation index calculation of road extraction effect, it shows that the algorithm can effectively extract road network from high-resolution remote sensing images.