Multi-Temporal Aerial Images in Riverbed Change Detection

• Main Authors: Ju-Jung Ke, 柯如榕
• Other Authors: Jung-Der Tsai
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/27747598978719836881

碩士 === 國立中興大學 === 土木工程學系所 === 102 === Due to climate and environmental changes in recent years, heavy rains or typhoons may produce changes in river environment, which is responsible for water supply and drainage tasks. However, a wide range of large-scale surveys of river changes often demand time-consuming and laborious manual measurements on riverbed terrain for river management purposes. This study focuses on how to use the existing aerial photographs to provide information in a more convenient way to those who request. In this study, multi-temporal aerial images purchased from the Aerial Survey Office, Forestry Bureau, Council of Agriculture were used in the experiment. After geometric registration and preprocessing, the principal component analysis (PCA) was performed for transforming aerial images for use in unsupervised classification. The result was compared with the result from supervised classification of the original image. The Kappa test was applied to assess the accuracy of the two classification approaches. It was shown that the supervised classification method is better than unsupervised classification method for subsequent riverbed changes analysis. Riverbed change analysis using multi-temporal aerial images is capable to provide immediate large-scale information for those who need to master changes of rivers in some circumstances.