Research On Road Extraction Method Based On Remote Sensing Image

Due to extract roads from remote sensing images is not only scientifically challenging,but also has main realistic meanings for applications such as automatic mapping,pattern recognition,and geographic information system(GIS)data updating.Since the traditional road extraction method is obtained through ground survey,which is expensive.Therefore,how to automatically and accurately extract road information from remote sensing images has become one of the important exploration directions in the field of remote sensing.Because the content of remote sensing images is complex and diverse,the extraction of roads will be subject to various disturbances such as light intensity and building shadow occlusion,which will lead to mistaken and missed extractions.Therefore,based on the theory and method of mathematical morphology,this paper uses adaptive morphology method and path morphology method to extract roads in remote sensing images.This paper will expand the main content of the paper from the following aspects :(1)This paper expounds the research background and significance,methods of extracting roads from remote sensing images at home and abroad,and research status of mathematical morphology.The basic theory and operation of mathematical morphology are briefly introduced,and experiments and analysis are carried out.(2)In view of the complex background of remote sensing images,when using traditional morphology to extract road targets,it is easy to change the road structure due to the use of fixed structural elements,which affects the accuracy of image segmentation.Therefore,a road extraction method from remote sensing images based on adaptive morphology is proposed.Firstly,the nonlinear structure tensor of the image is calculated and decomposed to obtain two feature attributes,namely the eigenvalue and the eigenvector.According to these two feature attributes,an adaptive ellipse structure element is constructed,and the corresponding adaptive morphological operation is defined.Secondly,construct the morphological high and low hat transformation according to the road features to enhance the road target,and realize the preliminary extraction of the road by the maximum inter-class variance method.Then set the shape parameters to identify whether the target in the image is a road area.Finally,the nonroad target is removed by the adaptive morphological filtering method,so as to extract the road area.The final experimental results prove that the method can accurately extract road regions from remote sensing images with complex backgrounds.(3)Due to the continuous improvement of the spatial resolution of remote sensing images,there are many background interferences such as vehicles,buildings,shadows and other artificial objects in the process of road extraction.Even if the road extraction based on traditional mathematical morphology can basically remove these interferences,but at the same time,there are also small areas of adhesion between roads and non-roads,as well as road breakage caused by shadow occlusion,and it cannot be completely extracted for roads with large curvature.Therefore,a road extraction method using path morphology is proposed.First ly,the image is preprocessed by grayscale and adaptive median filtering.Secondly,the road image is enhanced by gray-scale linear transformation,and on this basis,the gray-level window slicing method is used to segment the image to obtain the initial road.Then the adhering road is reduced by the path opening operation.Finally,after shape filtering and hole filling of the image,the road centerline is accurately extracted.Based on a large number of experimental datas,the results of this method and other methods are compared and analyzed,which verify the effectiveness of the proposed method in road extraction from remote sensing images.