The Research Of The Multi-scale Urban Road Extraction Method On The High Resolution Remote Sensing Image

High-resolution remote sensing has become an important meanstechnology of access to geographic information. The quickly and accuratelyextracting urban road is of great significance and role on geographicalconditions monitoring, basic surveying and mapping, geographic informationupdates and traffic management. Paper pin road features high resolutionremote sensing image extraction problem, based on the analysis of the researchstatus, characteristics of remote sensing images scale expression of urbanroads and described the establishment of high-resolution remote sensing roadextraction rule sets, and experimental analysis.Analyze the road extraction in high resolution images at home and abroad.Generalize the extraction method and the noise factor of the road extraction onhigh resolution images. Analyzes the scale features of fast-track, main roadsand secondary roads of three levels of urban roads from the characteristic spectral imaging, geometry and topology. Research the expression methodsand content of the scale features of three levels of urban roads.Based on three kinds of roads scale characteristics expression anddescription of urban roads to build rule set of city remote sensing. Cityfast-track considered three features: spectral characteristics, geometric features,context features. Through designating the first and end target of way,associated with features to build a circular rule set to connect the end and firstto extract the main road. Using morphological dilation and erosion operationson the initial extraction of roads and lack of deburring process or filling tomake road boundary smooth. Based on algorithm of regional growth, presetthe right initial seed point, growth criterion and termination conditions toextract the line feature of secondary road.Remote Sensing Application Taiyuan0.5m resolution, Kang Yi in therules were set to build the actual verification, extraction of the city fast lanesand roads, by the expansion erosion operation realized in MatLab environment,roads and post-processing based on the region secondary roads featureextraction algorithm increased recognition. Randomly selected sample points,evaluate the accuracy of the extracted result, the evaluation results indicate thatthe method is basically feasible.Finally, the paper summarizes, and issue recommendations for furtherresearch.