Research On Geographic Conditions Survey Based On Remote Sensing Technology

The remote sensing technology has been widely used in the national geographic condition survey. With rich information of spectrum, shape and texture, the high resolution remote sensing images have already become the main source of data. Since they have comprehensive and abundant information, how to fast and automatically get accurate information from mass of high resolution remote sensing images has become the most urgent problem needed to be solved.The fundamental information of the national geographic condition survey and three examples are briefly analyzed in this paper. Three kinds of remote sensing image interpretation methods and the analysis of remote sensing image features as well as 17 kinds of typical features including the confusing features in Tibet autonomous Region are discussed in detail. Furthermore, interpretation system of geographic condition survey in Tibet Autonomous Region is built up. The important part in this paper is three experiments of three interpretation methods in use of images from the same satellite with similar topography and equal difficulty. Thus, we get scientific conclusions as follows:(1) In different regions we should carry out the geographic condition survey in different ways according to local conditions.The interpretation method is the basic way to get information in the national geographic condition survey and the establishment of interpretation is vital to remote sensing image interpretation.(2) By using visual interpretation method, features in A08 experiment are divided into 14 kinds, with 94.01% as the overall accuracy and 0.93 as Kappa coefficient. By means of the automatic computer interpretation method, features in B05 experiment are divided into 16 kinds, with 90.68% as the overall accuracy and 0.89 as Kappa coefficient.Through computer–human interpretation method,features in C07 experiment contain 19 kinds, with 93.13% as the overall accuracy and 0.89 as Kappa coefficient. Comparing the same features in these three experiments, results come out that the overall classification accuracy with visual interpretation method is higher than that with computer–human interpretation method and the overall classification accuracy with computer–human interpretation method is higher than that with automatic computer interpretation method.(3) The use frequency of visual interpretation method, automatic computer interpretation method and computer–human interpretation method are 36%, 6% and 58%;in addition,automatic computer interpretation method and visual interpretation method are applicable to regions with simple topography and uncomplicated features while computer–human interpretation method is appropriate for areas with complex topography and complicated features; furthermore, visual interpretation method and computer–human interpretation method are used more frequently than automatic computer interpretation method in the national geographic condition survey.