Research And Implementation, Based On Remote Sensing Images Of Surface Features Automatic Extraction Of Information Technology

Natural resource is the mankind denies, is a important condition of national economic development and the people's raising of living standard with the material foundation of survival and development, it is the important source of wealth of society. Along with the unceasing increase of population and the development of the modernization cause of our country, economic, social development and the contradictory more and more sharp, natural calamity of population and resource and environmental problem also become serious day by day , are serious to have restricted the coordinative development of economy, society, resource and environment. Therefore , with modern high-tech means - the technology of remote sensing accurate find out resource and environmental resources, for the social development and construction of national economy, it is already urgent to offer science decision.Because of excellent technical predominance , remote sensing has been used in investigation of territorial resources early in 1970s . With rapid development of technique system of remote sensing , increase of remote sensing platforms , descending cost of imagery , remote sensing has become a rapid and effective , method in the field of investigation of territorial resources . Especially , high-resolution satellite remote sensing image appears which can provide space basic information and Various natural object , landforms information and greatly increase practicability of in investigation of territorial resources.Rapid recognition various natural resources and urban building from high resolution remote sensing image is an arduoud task . It's more practical to vectorize that had recognition information . The paper discusses two contents about image recognition of high resolution satellite image and information extraction technique . One is rapid recognition of nature object information ; the other is vectorization of the had information.With tome resolution and space resolution increasing , objects less than 1 meter have been recognized on high resolution remote sensing image . Degree of extracting ground information is from affirmance phase to description in detail . Thus rapid deyecting unit natural object from image has data base . In despite of having high resolution image as fine data base ,corresponding image segmentation technique is still slower than satellite imaging technique ,it made a lot valuable information can not in first time untie to translate come out , have wasted data .The paper discusses image segmentation theory , and then establishes the technique route that is colligating kinds if segmentation technique to rapidly recognize ground information on high resolution remote sensing image , main points are Water department and road . The new method is based upon analysis to object feature character. This is a beneficial try for recognizing others basics data.To force practicability, resource investigation must have the function which is able to automatically get vector data . In the paper we primarily research and discuss vectorization mentod of objects boundary . Aming at rectangle object character ,this paper describes a new vectorization approach based on corners position. The arithmetic adopt piecewise liner fitting of least-square algorithm to vectorizen boundary . The corner is the point of intersection which is computed through fitting beeline . Becase the corner doesn't always exist on boundary, the method is robust.