The Method Of Urban Greenland Information Extraction Based On ZY-3Remote Sensing Imagery

Urban green space is the land covered with natural or man vegetation in the city, and its forms are complex and diverse. It plays a role that improving the ecological environment, appearance, tourism and recreation conditions of our city. In addition, it is one of the important symbols that measuring the level of urban modernization. national census geography consists of three parts: topography, land cover and geographic units. Land cover refers to census of category, location, scope and size of the following categories which are vegetation farmland, orchard, woodland and grassland. Studying the extraction method of urban green space information has a great significance in national census geography. In comparison with traditional ground-based investigations, extraction of urban green space conducted on remote sensing images is faster and more efficient. At present, the extraction methods of urban green space using remote sensing images are summarized as the following two types:using traditional methods, such as supervised classification, vegetation indices or visual interpretation; automatic extraction of urban green space based on image segmentation and object-oriented technology, establishing training zones by sample member function or creating rule sets upon the features of target objects.In order to improve the accuracy of automatic classification of high resolution remote sensing images, we used fuzzy rule-based object classification combined with vector and rule sets. The fuzzy discriminant function used to identify the target objects is built up with various information of target objects that presented on remote sensing images, such as spectral features, texture features, pological relations and so on.First, it is summarized the research status in the field of remote sensing information extraction, focusing on vegetation extraction methods and research processes. Research pixel based and object-oriented extraction method, focusing on researching object-oriented multi-scale segmentation and fuzzy classification method systematic. For laying the extracted foundation of high-resolution remote sensing images of urban green space, using fuzzy classification method to build object-based rule sets do two experiments to extract urban green space and green vegetation cover type classification based on other high-resolution image data and TM data, and summarizing the shortage of the low spectral and low spatial resolution image data in the extraction processWith built-up area of Taiyuan as study area, using ZY-3remote sensing data combined with vector data can systematically analysis and extract urban green space.Urban green space were analyzed from spectrum, geometry and texture characteristics of these three aspects about urban green space (Take content parks and attached green in built-up area) into consideration on ZY-3; vegetation cover types were analyzed, by spectral characteristics, texture characteristics of green grass and green woodland. Were established remote sensing interpretation signs about urban green space and vegetation cover types. The built-up area of Taiyuan urban green space information is extracted through rule sets supported by fuzzy classification method with spectrum, geometry, texture feature and spatial correlation with vector, and the extraction results of effectiveness is evaluated. Evaluation results show that the method is basically feasible and has a certain reference to National.Geographical General Investigation which is ongoing.