Research On Extraction And Application Of Urban Spatial Information Based On GF-2 Remote Sensing Image

Remote sensing technology is one of the main technical means for earth observation,and for the acquisition of spatial information on the earth.With the development of remote sensing technology,the research and application of high resolution remote sensing image has attracted more and more attention.The research and application of object-oriented image classification methods has risen around with the appearance of high resolution remote sensing image,which spread from point to surface,and be used in urban mapping,urban land dynamic monitoring,public safety management and emergency warning et al.What can be certain is that object-oriented image analysis(OBIA)has blossomed in many fields.With the comparison of traditional pixel-based remote sensing image classification,object-oriented image classification can overcome the "salt & pepper noise" and " mixed pixel" and some other issues,by means of the image segmentation algorithm,the image is divided into numerous "image objects",with the help of common geometric feature,spectral feature,texture feature and context between these image objects,thus can carry out accurate extraction for each type of land features,which has a huge advantage.But the current researches of object-oriented classification mainly in the forest,agriculture,and other suburban areas,which lack direct researches on the extraction of inner urban area with complex land features,and the study area of the current researches these are also very small,failed to make full use of the features of "image objects" to extract complex spatial information,thus not reflect the advantage of object-oriented classification sufficiently.Allowing for these research situations,this research starts from overall,for the purpose of extracting urban spatial information,and takes Beijing as the study area,with the 1m spatial resolution of GF-2 multi-spectral and panchromatic fusion image for data support,utilizing the object-oriented classification method and taking the V-IW model proposed in this paper as theoretical guidance to organize the main urban spatial information,which be divided into 3 categories and 14 types,then through multi-resolution segmentation,spectral difference segmentation for the acquisition of "image objects" of GF-2 image,based on spectral features,band values,geometry features,texture features and the context to establish rule-sets,so as to extract the main urban spatial information.To build a bridge between RS and GIS,and furthermore,for urban spatial analysis application.The results show that the overall accuracy is 83.24% and the Kappa coefficient is about 0.8069.In the extraction of single type of spatial information,vegetation,river extraction accuracy is higher,higher than 80%,the accuracy of the building such as roof,roads has a little lower accuracy,and the overall results of urban spatial extraction are satisfactory.