高分辨率遥感图像中道路提取方法研究-电路与系统专业论文.docxVIP

AbstractAbstract Abstract Abstract As foundational geographical Information，the identification and accurate positioning of road network is significant for image understanding，GIS data obtaining， cartography and acting as the reference of other objects．How to make better use of remote sensing images to extract road information has been an important research topic in recent years．Road extraction Can be classified into full automatic and semi．automatic patterns according to the automatic degree of extracting．At present，researchers both at home and broad have conducted plentiful studies on也ese two different ways．However, except for some semi·automatic systems，there is still not any automatic method with good applicability．Moreover,lots of applications focus on the road networks in medium or low resolution．As the resolution gets higher,the fact that roads in high resolution and in medium or low resolution have quite different image features makes the road extraction in high resolution more complicate and challenging． In this paper,we have been studied the extraction of mountain roads and urban roads iIl lligh resolution respectively．It mainly consists of the following three parts： (1)Grounding on the traditional morphological open and close operators，the area operators call better adjust to the changes of image shape and eliminate the block effect． First applying area operators to preprocess the images，and then through watershed transform call ke印the road information better．This part deals mainly埘th the implementation procedure of area closing and the application of area operators in the automatic extraction of mountain roads．And the operators’validity is verified through experimental results as well． (2)For the urban roads with centerlines，to adopt the semi．automatic method which is on the basis of template matching Can extract the centerlines of the city’S main roads in real—time and accumtely．First we centre around a point on the centerline to build a corresponding t