Landslide occurrence prediction using optical flow

Abstract

There are many natural disasters and many people suffer from them every year. One of them, the landslide, is difficult to predict, since the predictive phenomenon has not been clarified, although the topography and stratum that are likely to cause landslides are roughly expected. In this study, we propose a method of predicting the occurrence of landslide by taking the video of the slope and using the optical flow measurement to obtain the sliding velocity of the slope. We used spatio-temporal differentiation method as an optical flow measurement method, which is suitable to measure small displacements. However, various weather conditions such as wind can affect accurate velocity vectors. In this research, we reduce the influence of weather conditions and explore the possibility of predicting the occurrence of landslide. It is expected to monitor a wide range of mountain slopes to predict the occurrence regardless of the weather.