Analysis And Research Of Road Extraction From High-Resolution Remote Sensing Images

With the development of remote sensing technology, the resolution of remote sensing images becomes higher and higher. Nowadays, high-resolution remote sensing images have been applied in various fields such as national land resource investigation, update of foundational geographic data, monitoring for changes of land utilization and so on. In remote sensing images, road is not only a kind of important basic geographic information, but also regarded as clues and reference for extraction of other ground objects. Correct road extraction is crucial to further application of high-resolution remote sensing images. However, high-resolution remote sensing images have huge information with more and more road objects. There are some interrupts caused by trees or shadows of buildings. So, road extraction from high-resolution remote sensing images has its scientific meaning and practical values.Road extraction from remote sensing images has been carried out since 1970s. Some algorithms have good results from remote sensing images with relative simple scenes. For example, it is good at extracting roads from rural areas and main avenues from urban regions. But there is not a perfect algorithm which can extract the whole roads from high-resolution remote sensing images with very sophisticated scenes so far, because of a plenty of interferences in images such as trees, shadows of buildings, vehicles on roads, road centerlines and etc. These interferences not only make road network in remote sensing images disordered, but also make road edges blurred. Beside these, some features similar to those of roads caused by parking lots, pool, rivers and roofs of buildings also make road extraction more difficult.On the basis of analysis and research of some existing road extraction methods, considering the characteristics of high-resolution remote sensing images, this thesis offers a method of road extraction in rural and suburban areas from high-resolution remote sensing images. In the framework of this method, road samples from images are extracted manually at first. Then, spectral features and texture features are calculated. A road extraction model from high-resolution remote sensing images based on these features is given. A feature fusion method based on Dempster-Shafer evidence theory is proposed for the road extraction in global scope. After fusion, a binary image of roads is obtained. Then an iterative direction filter is used to remove the noises in this binary image. Finally, morphological transformations are used to reduce interior spots in roads and link some broken roads to obtain final object road images.The experiments show, comparing with some methods based on single feature for road extraction, this method has obvious advantages especially in the extraction of road detail information, for its full use of multi-features of road and the fusion of multi-features.