Basic Element-oriented Urban Building Density Remote Sensing Information Extraction Research

With the development of remote sensing technology, there appear more and more data alternatives of remote sensing. The quick and precise information extraction from large numbers of remote sensing data, is in urgent need for remote sensing application. Research for information to be extracted from digital remote sensing image is one of important research orientations in remote sensing technology research. At the present the information extraction of remote sensing image is mainly based on the single pixel DN, while the inherent relation between pixels is discarded. Even the subsidiary image classification by texture character is just to improve the recognition precision and reliability of single pixel, without considering abundant spatial information in remote sensing image. In recent years, the basic element-oriented image analysis technology and multi-spatial scale level analysis provide a new resolution. From what's mentioned above the dissertation discusses the concerning idea and technology with TM data.TM data is broadly applied on resource and environment research for the easy access and the better quality with a low price. However in the urban research, TM data is mainly applied to urban expansion urban thermal island effect and urban greenbelt for the low spatial resolution. In recent years, the new high spatial resolution remote sensing data continuously come up and provide convenience for the urban research. But most of these data are at high price only about the recent five years information. So TM data has incomparable advantage in twenty years'historical scale research. Meanwhile its wave band character is adaptable to the urban target extraction. Besides, the data enjoy great value in use, if image information is fully studied.This dissertation with the analytical technology of the new basic-element oriented image, probes into building density distributing information extraction from urbanization area based on multi-spatial levels relative information extraction and TM data character. This dissertation firstly introduces the concerning concept and the present research and then focuses on particular address the information extraction by TM data:a, correct image radiation, precision geometry and image resize. Then b, segment the remote sensing image in the scale of three spatial levels and extract character space respectively. c, synthesis three levels of information by fuzzy mathematics theory, define the object information. d, produce the confusion matrix, and estimate the accuracy of the extracted information for...