FEATURE EXTRACTION AND ANALYSIS OF RIVER CHANGE DETECTION BASED ON REMOTE SENSING IMAGES

摘要

The characteristics of rivers in Taiwan are the steep slope with high sediment concentration. The distribution of the precipitation is non-uniform due to the mountain region and the typhoon. The dynamics of channel meandering, therefore, become complicate and not easy to simulate with numerical modelling because of the geographic environment and extreme events. To understand the morphology of the river channel, the field measurements and remote sensing methods for channel detection are necessary. However, there is no efficient method to delineate the channel from remote sensing images. This study conducted edge detection concepts to automatically capture the river bank based on remote sensing images. The images of the river channel were obtained from different seasons at the same location to test the capability of our concepts. This study shows that the river channel can be extracted automatically and the morph-dynamics of the channel can be detected from multi-temporal remote sensing images for change detection.