面向对象的高分辨率遥感影像城区建筑物分级提取方法精选.pdfVIP

39 1 Vol.39, No. 1 2010 2 Acta Geodaetica et Cartographica Sinica Feb.,2010 :1001-1595(2010) 01-0039-07 1, 2 1, 2 1, 2 1, 2 1, 2 陶 超 , 谭毅华 , 蔡华杰 , 杜 博 , 田金文 1. , 430074; 2. , 430074 Object-oriented Method of ierarchical Urban Building Extraction from igh-resolu- tion Remote-Sensing Imagery 1, 2 1, 2 1, 2 1, 2 1, 2 TAO Chao , TANYihua , CAI uajie , DUBo , TIANJinwen 1. Institute of Pattern Recognitionand Artificial Intelligence, uazhong University ofScienceand Technology, Wuhan430074, Chi- na; 2. The State Key Laboratory for Multi-spectra Information Processing Technology, uazhong University of Science and Tech- nology, Wuhan 430074, China Abstract :An automatic urban building extraction method for high-resolution remote-sensing imagery, which com- bines building segmentation based on neighbor total variations withobject-oriented analysis, is presented in this paper. Aimed at different extraction complexity from various buildings in the segmented image, a hierarchical building extractionstrategy with mult-i featurefusion is adopted. Firstly, we extract some rectangle buildingswhich remain intact after segmentationthrough shape analysis. Secondly, in order to ensure eachcandidate building tar- get to be independent, multidirectional morphological road-filtering algorithm is designed which can separate buildings fromthe neighboring roadswithsimilarspectrum. Finally, wetake the extracted buildings and the excluded non-buildings as samples to establishprobability model respectively, and Bayesiandiscriminating classifier is used for making judgment of the other candidate building objects to get the ultimate extraction result. The experimental results have shownthat the approach is ableto detect buildings withdifferent structure and spectral fea