| Summary: | 碩士 === 國立成功大學 === 測量工程學系 === 87 === Traditional methods of image classification use only tone , texture , and geometric characteristics . Classification depends on the spectral property of the pixels . By spectral sample interpretation , we can perform image classification . When we interpretate an image , we need to consider knowledge more than image interpretation , in order to get satisfying result . And the information in GIS is a good resource of ancilly data .
By using expert system , we expect to integrate spectral information in remote sensing and data in GIS , and get better result of the classification in image interpretation .The aim of using expert system is how to use indriect data in GIS to classify , and inference from knowledge in special domain range to make decision in classification .
In this research , our goal is the classification result of forest type (deciduous and coniferous) . We integrate the information in remote sensing and GIS to establish the knowledge base of forest type . It presents a method of classification for special thematic classification . In this research , with the assistance of expert system , we not only identify the area of forest , but also the area of forest type . The result is much better than spectral classification only . The role of expert system is to integrate the knowledge and data in remote sensing and GIS , and it may let unprofessional users do more exact image interpretation .
|
|---|
