| Summary: | 碩士 === 國立中央大學 === 土木工程研究所 === 87 === Recently，change detection from satellite images has been applied in various fields wildly. For example: the exploitation of the golf courses, the demarcations of the damageable zone after typhoons and the surveying of the crop growth. Satellite images sustain great help for monitoring the land change because of its characteristics of multispectral , period and digital.
In this study, we will use the method of direct multidate classification for change detection. It combines the properties of multi-spectral and temporal variability of the satellite images. We classified all kinds of changed types from supervised fuzzy classification and compounded all training areas information by the method of perm. To avoid the errors from the atmospheric effect, we got the training areas from the first image and the second image severally. At last, we used the conception of the knowledge bases to shorten the computer time and increase the efficiency and the accuracy of change detection.
This method has been tested by the simulated images and SPOT images. The results showed that it can detect the change of the areas with very high accuracy and automatic operation.
|
|---|
