Research On Establishment And Application Of Regional Dynamic Target Feature Database For Multi-source Remote Sensing Imagery

As a crucial means for earth observation,remote sensing imagery interpretation has always been a research direction which attracts attention from scholars at home and abroad.Meanwhile,as a large number of high-resolution remote sensing satellites are launched,researchers from various countries have devoted to high-resolution image interpretation.The improvement of the resolution for remote sensing imagery makes more detailed feature information acquired,and it increases the observation institutes' demands on the research direction image interpretation based on targets at the same time.Besides,due to the rapid development of the remote sensing technologies,a large number o f remote sensing data need be processed,it can improve the efficiency and accuracy for interpretation to deal with the data high-efficiently,normatively and intelligently.Therefore,the paper revolves around the regional dynamic target feature database for multi-source remote sensing imagery and introduce its composition,establishment method and application.The specific content shows as follows:First,for the regional target feature database for multi-source remote sensing imagery,the paper introduces some target features and its extraction method.For optical images,it mainly introduces three kinds of classical target features(such as geometric features,moment invariant features and texture features),and for high-resolution SAR images,aforementioned target features only consider from the image and ignore the nature of SAR image.So the paper focuses the research on the scattering characteristics which are proved to reconstruct targets successfully into the target feature vectors and replaces texture feature with it in target multi-dimension feature vector.Finally the paper makes up a multi-dimensional feature with the features from optical and high-resolution SAR images.Then,the paper introduces the establishing process of the regional target feature database for multi-source remote sensing imagery.Target extraction is a key step in the complete establishment,and there are a lot of researches about target extraction for optical images,so the paper focuses on the researches on the target extraction for high-resolution SAR images.First,it introduces some traditional target detection method for high-resolution SAR images and analyze some problems when the interferences of the background is more obvious in the images.Then,the paper puts forward a high-resolution SAR images with the spatial constraints of the optical images.Besides the paper compares the proposed method and the traditional methods.Finally,the paper designed a software system to accomplish a complete establishment of the regional target feature database for multi-source remote sensing imagery.Finally,the paper lists some applications on image interpretations for regional target feature database for multi-source remote sensing imagery.First,the paper applies it into the regional identification,and its interpretation results can help observers accomplish some tasks which include regional monitoring and early warning and so on.Then the paper adds the regional target feature database into the application on regional target change detection and puts forward a change detection method based on regional target feature database,which can determine the kind of the target on the basis of the traditional method that judge the existence for targets,and it can enrich the resul ts of image interpretation.Finally,to make target feature database more intelligent,the paper makes a dynamic analysis on it,including dynamic update and combination of regional target feature database,which can offer help for other image interpretation,such as target recognition and so on.