Application of image quality assessment for rockfall investigation
D Kim, AS Balasubramaniam, Ivan Gratchev, Kim, SH Chang · Griffith Research Online · 2019
This paper presents an automatic image sensing method to detect rockfall events using time-lapse photographs and digital image processing. Based on the change detection concept, an image quality assessment procedure is proposed to detect rockfall events in this study. Sets of temporal images taken from a rockfall field test are selected to assess similarity between the images. Two representative image quality assessment algorithms have been employed to find the applicability of the algorithms for rockfall event detection. In this study, the error map which highlights rockfall movements by differences between two images is visualized using OpenCV and Python scripts. The results show that time-lapse photographs obtained from a fixed photographed position can be effectively used to detect rockfall event initiated from the captured area. Also, the well-known SSIM approach gives excellent results from its efficiency of similarity detection.