Object-Based Automatic Extraction Of Change Information Based On High-Resolution Remote Sensing Image Research

With more details, clearer geometry feature and space information, the operation of high-resolution remote sensing image (HR) change fundamentally. Traditional remote sensing image processing will no longer apply to HR. Therefore, in order to extract change information automatically from the different time remote sensing images and ensure the efficiency, with the QuickBird (HR)images as the main source, though a comprehensive analysis, concluded on the basis of research results at home and abroad, this thesis develops several results which focus on automatic change information extraction in object-based image analysis as follows:(1) The dissertation put up an improved algorithm to pre-processing, which uses the binary mask based on CVA adaptive threshold. Based on the pre-processing of image, this method through the adaptive threshold divided to determine the changes in the scope of the image accurately first, generating record changes and no-changes binary mask of the image in order to carry out the first filtration.(2) Face to the problem of how to automatic matching feature rules for the most effective classification from so many features is always a key issue in the field of feature selection and extraction. Traditional methods often require mass of human intervention, by repeatedly comparing experiment to determine the differences characteristics property values of the image objects, which result in greatly reduced efficiency. This paper break through the traditional pure artificial ways, put up a automatic feature optimize selection approach based on J-M (Jeffries-Matusita) and Bayesian threshold.(3) The dissertation put up an object-based change information extraction strategy based on multi-information fusion. This method resolves the problem of the OBIA which is mostly used in the field of technical limitations of classification. It provides a new way to automatic extract change information from HR. Besides, that makes the OBIA extended to this field.(4) With the technical methods and process flow of the OBIA technology based on multi-information fusion change information extraction software integration, this paper developed and designed a OBIA intelligent System of HR, besides incorporated the entire job process stream.