Object-Oriented For Road Extraction From High Resolution Remote Sensing Images

With the rapid development of remote sensing technology in recent years,the way to obtain high-resolution remote sensing images has become easier.High-resolution remote sensing images contain rich details of ground features,and it has become an important way to obtain geographic information of ground features from high-resolution remote sensing images.As one of the basic geographic information,road is an important part of transportation system,which plays an extremely important role in traffic planning,vehicle navigation,mapping,disaster relief and other fields.And how to accurately and reliably extract the complete road information from high-resolution remote sensing images is a hot direction of current research.In this paper,we propose a remote sensing image road extraction method based on objectoriented technology to address the problems of road spectral similarity,road complexity and feature occlusion in the existing road extraction research.The main research contents and results of this paper are as follows:(1)An improved balanced contrast enhancement technique is used in image contrast enhancement to enhance the image.The algorithm can effectively enhance the contrast,brightness and color information of the image,and reduce the interference of other ground object information to the road information.The hyperbolic sine function is used to enhance the image contrast and color information,and the cumulative distribution function of type II generalized logistic distribution is introduced to improve the image brightness.The results of comparative experiments show that the method can effectively enhance the road information,increase the contrast between the road and other features,and facilitate the extraction of road areas.(2)A high-resolution remote sensing image segmentation method based on regional multifeature fusion is proposed.When performing region merging,the region adjacency graph and nearest neighbor graph(RAG&NNG)are used to maintain the regional adjacency network to improve the speed of region merging;the spectral,edge,shape and other characteristics of the image are used to establish the region merging criterion,and the optimal segmentation scale is obtained by automatic threshold selection and optimal scale selection to make the boundary of each feature in the image clear.Finally,by comparing with other algorithms,it is proved that the method can effectively reduce the "over-segmentation" and "under-segmentation" area and obtain clear feature boundaries.(3)A road centerline extraction algorithm for high-resolution remote sensing images based on multi-scale segmentation is proposed.First,based on the optimal scale segmentation results,the initial road area is obtained by integrating road object features;the multi-directional linear structural elements are used to perform morphological filtering to eliminate the area that is connected to the road,and the refined road area is obtained by filling holes;the mathematical morphology thinning algorithm is used to refine the road,and the "intersection method" is used to find road intersections to remove burrs;finally,the fractured centerline connection method is used to connect the broken parts of the road network to obtain a complete road network.The effectiveness and feasibility of the method in this paper are proved by quantitative analysis of the comparative experimental results.