A Study Applying Category-oriented Image Interpretation in Detecting High Enviromental Impact Landuse Categories in Slopeland- A Case Studyin Taipei County

• Main Authors: Chiou, Kuang-Shui, 邱桄水
• Other Authors: 朱子豪
• Format: Others
• Language: zh-TW
• Published: 1995
• Online Access: http://ndltd.ncl.edu.tw/handle/79278529670838024872

碩士 === 國立臺灣大學 === 地理學研究所 === 83 === 　　Improper development in slopeland will cause many serious disasters. Fast urban development in slopeland is the current trend in Taiwan, but now more and more development cases in slopeland, Remote Sensing tools should be heavily applied. But current Remote Sensing interpretation systems have not been sufficient in terms of degree of automatism, and accuracy or interpretation for the categories of slopeland. Therefore, detail of interpretation key features of categories should be studies in order to design a better automatic interpretation process. The major goal of this research is to extract the key features of golf courses and grave yards, to guide the design of interpretation process and system implementation. The slop-eland of Taipei county was chosed as the study area. 　　The method of this reserch is based on the concepts of artifical intelligence. The knowledge and procedure for automatic interpretation were extracted from human interpretation. The interpretation knowledge was inducted from ground, aero photographs and satellite images. The strength of feature and cost of automatic measuremennt were applied to design the procedure of automatic interpretation system. 　　The results of this reserch were found that the key features of golf courses is fairway, and the key features of grave yards is grass, small built-up, and barren. This system could detected one of two golf course. The error was caused by diff-erent grass condition of fairways. Grave yards could be detected by this system, but still some error occured which could be corrected by using more texture features to separate them from grave yards.