Urban road information extraction from high resolution remotely sensed image based on semantic model

Abstract

The road is an important fundamental geographic information. Acquiring the road information quickly and accurately has a great significance for GIS data updating, image matching, target detection, and automated digital mapping. Automatic/semi-automatic extraction of road information of remote sensing images is the problem of visual interpretation computer research, RS, and GIS. Application of high resolution satellite images and development of semantic model theory provides more possibilities and a higher degree of accuracy for object extraction of remotely sensed image. The OAR model of human cognition has been introduced, experimental study has been carried out on extracting road information from Quick Bird multi-spectral Imaging with semantic model, the result shows that the length accuracy of extracted road was 89.19%, the width accuracy is 71.54%, and the intact rate 50.32%. The extracted result is better than that of object-oriented extracted. As a whole, that the road information extraction semantic model of highresolution satellite remotely sensed image is efficiency.