Research On Road Extraction From Remote Sensing Images By Integrating Edge Information And Regional Information

With the development of space remote sensing technology,China has launched multiple series of high-resolution remote sensing satellites in the past decade,and high-resolution image data has greatly improved traditional surveying and mapping techniques and methods.As an important artificial feature in high-resolution images,road is one of the important components of cities and an important indicator to measure urban development,which not only has a wide range of application requirements in urban planning and construction,change detection,natural disaster analysis and traffic management,but also an important part of most maps and geographic information systems.Traditional surveying and mapping methods to obtain road information are time-consuming,laborious and inefficient,while the use of highresolution remote sensing images for road information extraction has the advantages of high degree of automation,fast acquisition speed and real-time update.With the continuous development of remote sensing technology,the resolution of remote sensing images has been improved,and the description of ground feature information has become richer and richer,which has brought convenience to road extraction but also brought new challenges,and the influence of homophromy,isotropic and heterogeneous spectrum,complexity and occlusion has also increased the difficulty of identifying and extracting road information.Although the research on road extraction of high-resolution remote sensing images has made some progress in recent years,its performance has not yet met the application requirements,based on the traditional edge detection algorithm has the problem of edge omission and serious misjudgment,and the regional growth has the defect of excessive growth leading to the confusion of roads and buildings,so this paper attempts to embed the road extraction method of high-resolution remote sensing images that integrates edge information and regional information,and the main research content is as follows:Combining bilateral filters and double-threshold maximum between-class variance(OTSU)to optimize the traditional Canny operator edge detection method: Bilateral Filters Canny algorithm(BFCanny).The algorithm fully combines the advantages of bilateral filtering,which is nonlinear filtering,which can complete noise reduction smoothing on the basis of maintaining the edge,overcoming the defect of Gaussian filters that only considers the relationship between pixels in space and loses edge information,and at the same time combines the genetic algorithm to improve the optimization of OTSU,and optimizes the double Threhold step of Canny’s algorithm with automatically obtained thresholds.Experiments show that the algorithm extracts road edge information better than the traditional Canny detection algorithm.Combining Adaptive Median Filter and Double-threshold Maximum Between-Class Variance(OTSU)to optimize the traditional Canny operator edge detection method: Adaptive Median Filter Canny algorithm(AMFCanny).Based on the fact that Median Filter causes image discontinuity when there is more noise,this paper introduces an adaptive median filter according to the window of dynamic adjustment of median filtering under preset conditions,and adjusts the Canny operator kernel combined with the optimized OTSU algorithm,and completes the road edge detection comparison experiment between the traditional Canny algorithm,BFCanny algorithm and AMFCanny algorithm.The edge detection of roads will lead to unclosed and discontinuous edges,and the regional growth of roads will cause excessive growth and blurred boundaries.Firstly,the edge information of the road is obtained by edge detection,and then the edge information is used to guide the growth of the area to complete the extraction of the road,and the accuracy of the traditional algorithm and Xie Qainli method is compared,and the experimental results show that the proposed algorithm effectively improves the completeness and smoothness of the road extraction,and the robustness is better.Finally,the road extraction results are optimized by the basic operation of morphology,morphological reconstruction and RSL algorithm.