Using Different Classified Methods to Discuss the Classified Accuracy of SPOT and IKONOS Satellite Images
以不同分类法探讨SPOT与IKONOS两种卫星影像之分类准确度

Remote sensing technology can be applied in research, planning, and management on a large-scale environment. It is a powerful tool for natural resource investigation. However there are various sources of satellite images and classified methods, so the researchers have to find the best combination to meet their requirements. In this study, the SPOT and IKONOS satellite images were used as material, and collocated with different classified methods to examine the spatial distribution of vegetation types in Kenting National Park. It is feasible for discussing the above issue by using satellite images, and different classified methods to get different classified accuracy. Although the spatial resolution of IKONOS is higher than SPOT, it is contrary to get lower classified accuracy by omission and commission. Moreover, there are discrepancies between different vegetation types in spectrum characterlstics, so it needs to improve the accuracy of classification further. Good training area and proper methods can improve the accuracy of classification. But owing to the atmospheric effects, the topographic effects, and the overlap of spectrum characteristics in different categories, the classified accuracy of satellite images were always to be influenced. Therefore, if this problem can be solved effectively, the results of images classification will be usefulness.