Flood flow analysis on aerial photos by image correlation method with support of stereo visualization

Abstract

Stereo images of flood flow in river channels taken by airplanes contain a lot of useful information for river engineering works. The image correlation analysis (ICA) on a computer, which is a standard technique for stereo image analysis, sometimes produces anomalous results because the color patterns to be traced on the flood water surface are not very clear. On the other hand, a classical manual analysis (CMA) through a stereoscope usually works well even in such condition by the excellent perspective sensitivity of human eyes. But the CMA needs much more labor and time than ICA. In this study, a new technique was developed for analyzing the stereo images of river freshets by combining the above mentioned two methods: A small amount of suggestions made by the CMA are sent to a computer to improve the ICA results. The proposed method was applied to the aerial photos of a flood of the Tone River in Japan. Comparison of the analytical results from two sets of stereo images carried out independently showed that the proposed method gives high accurate result with less labor cost.