Research On Road Extraction Algorithm Of High Resolution Remote Sensing Image

In recent years,with the development of satellite remote sensing technology,we can obtain a large number of high-resolution remote sensing image data.How to obtain effective information efficiently and accurately from a large amount of data has become a problem to be solved.High resolution remote sensing image is widely used in urban construction,traffic planning,map data update,natural disaster detection,military target strike,driverless and other fields.As an important infrastructure in human society,road plays an essential role in social development and progress.Therefore,the extraction of road network information has become an important part of remote sensing image information extraction.Affected by the complex environment around the road(such as vehicles,buildings,etc.),there are various difficulties in accurately extracting the road network information.After the research of many scholars,various road extraction methods have been proposed.The research on road extraction of highscoring remote sensing images has made great progress,but its extraction accuracy needs to be further improved.Aiming at the problems in road extraction of remote sensing images,The main contents of this paper are as follows:1.Image smoothing based on Gaussian filtering.For the smoother road contour map,after the skeleton refinement algorithm is processed,the road centerline obtained is smoother,so the initial road network image is smoothed and then the centerline is extracted,which can remove part of the “burr” phenomenon.To smooth the image,you need to choose a suitable filter.The filters considered in this paper include: mean filtering,median filtering,and Gaussian filtering.After several sets of experiments,Gaussian filtering can produce a good smoothing effect on the initial road extraction map.2.Aiming at the problem of "burr" in the extracted road center line,this paper uses "intersection method" to remove the "burr" phenomenon.The idea of "intersection method" algorithm is: first search the "intersection" of each line segment in the road centerline extraction drawing,then search the length of each line segment passing through the "intersection" starting from the "intersection",and then delete the line segment whose length is less than the threshold value,so as to eliminate the "burr" phenomenon.The innovation of this paper is to propose a new "intersection point" search rule.3.In view of the problem that the extracted road network information produces the fracture,this paper proposes an adaptive threshold fracture road connection method.The first step of the algorithm is to find the "point to be connected",that is,the "end point" of the segment where the fracture phenomenon occurs.The direction of the "points to be connected" in the image is calculated after the search is completed.Within the quadrant to which each "points to be connected" direction belongs,the adaptive threshold R is the value of each "point to be connected" and the target closest to the point.The distance between the points.Determine the search range of a sector according to the direction of the point and the value of R.When there is another "point to be connected" in the search range of "point to be connected",two "points to be connected" are connected.Otherwise,connect the "point to be connected" with the target point closest to the point,traverse all the "points to be connected",and complete the connection of the broken road to obtain a complete road network information map.