Research On Object-oriented Information Extraction From The First Satellite Of High Resolution

With the rapid development of remote sensing technology, access to information,how to improve the speed and accuracy of remote sensing images to extractinformation is an important research direction in the field of remote sensing dataprocessing. Traditional remote sensing image information extraction technology isbased on information extraction techniques like element involved in informationextraction factor pixel spectral information is often difficult to obtain satisfactoryextraction. In recent years, with extensive use of high-resolution remote sensingimages, object-oriented remote sensing image information extraction technology wideattention. The technology is similar to the spectral characteristics and texture"homogeneous objects" as the minimum information extracting unit, only the spectralinformation as a basis for extracting the information, and texture information ofremote sensing data will spatial configuration information, and even visual sensingworkers interpret the experience and knowledge together as information extractionbased remote sensing information extraction greatly improve the accuracy andefficiency, resulting in remote sensing image information extraction graduallyshowing its unique value and potential.In this paper, Selection of the latest high score-made high-resolution image, usefor the score imaging features, conducted radiometric calibration, geometriccorrection, image enhancement, image preprocessing research to Germany definienscompany developed the world’s first an object-oriented software eCognition classifiedas a tool to explore in depth the basis of object-oriented technology of remote sensingimage information extraction method, the object-oriented classification of ideas andtechniques used in remote sensing image information extraction. Through multi-scaleimage segmentation and parameter selection, class hierarchy and semantic structure creation, classification rules and other references, the use of different remote sensingdata, a lot of experiments and research.In this paper, in order to get the best segmentation parameters in the study area,the multiscale segmentation segmentation scale, color, shape, tightness andsmoothness set parameters such as a lot of tests to get the city’s best segmentationscale information extraction, establish classification rules for urban object-orientedinformation extraction. After extracting remote sensing information, evaluate theaccuracy based on the traditional maximum likelihood classification andobject-oriented method, the results show that the use of object-oriented image analysistechniques for high-resolution remote sensing image information extraction and highprecision.