Study On The Goal-Oriented Information Extraction Method

Remote sensing images information extraction is a major content in the remote sensing images research field, and it is also the subject of the study in the pattern classification. Remote sensing images, and the accuracy of information obtained directly affect the application of remote sensing data and its practical value. High-resolution remote sensing images possess a wealth of information, which are the rich resources of needed information. This paper use the most recently applied quickbird (" QuickBird") remote sensing images as the basic data. With the help of those data, according to the information extracted technical methods and processes, a series of researchs on the information extraction have been carried out. Therefore, the author put forward information extraction techniques and specific methods and processes.This study uses the graphics workstation as a research platform, and takes various remote sensing images processors to extract the targeted information (the YellowRiver in the area of Merdang Wood land information, the water system and the information of traffic elements ) . Meanwhile, the precise verification and evaluation are based on the spot investigated and surveyed data, which are the basis of the precise extraction of information. The major researching contents and methods of this paper include :(1) Done on the integration of the space data of multiple sources and gave the data support to the GIS policy analysis. Aimed at solving the problem of mediamatrix of space data in GIS, used the data format integration to accomplish the space data integration. By using this method to achieve the amalgamation and integration of various space data in the researching area, and provided a data basis of the space policy-making and analysis in GIS.(2) Based on the basic methods of remote sensing images from domestic and overseas, therefore, extracted the useful images according to a series of processing processduers, such as enhancing the images, registering the images, amalgamating the data, and superving classfication (u nsupervising classfication) .(3)Accomplished the images'remoting and processing technique based on the ERDAS, PCI and ERDAS Imagine. Based on the remoting images data of the Yellow River in the area of Merdang Wood land as the data source, explored the unsupervising classfication of the water system and the information of traffic elements,and so on.In addition, compared with the classfied results to work out the conclusion of the best integration method.(4)In the ERDAS Imagine, ENVI and PCI softwares, used the different extracting informations as target subjects to extract informations according to the different algorithms.It turned out that different extracting subjects, should ues the suitable softwares and extracting methods.It evaluated the research results form the accuracy of the time and the space.(5)Other categories of study and implementation.The multiple sources remote sensing information extraction of the Yellow River in the area of Merdang Wood land has been already applied to the practice.It turned out that the various methods put forward in this paper have realistic directing functions, can complete plenty of unusual works which can't be done by the regular methods, save the man power, financial investments, and materials, improve the rate working efficiency, and give a strong technique support to the project decisions.The process of extracting the informations has been improved from certain degrees.According to the comparison and analysis on those methods in this paper, the author has found out that different information extraction methods are applied to differet subjects, and we should choose the suitable methods to carry out according to the practical types of the needing informations.