Development of an iterative Otsu method for vision-based structural displacement measurement under low-light conditions

Vision-based structural displacement measurement method presented great potential for structural health monitoring application due to its distinguished advantages, and LED targets were always employed to deal with the issue of poor visibility under low-light conditions. However, the ambient light and unfavorable weather seriously influence the identification performance even with LED target. In this paper, an iterative Otsu method was developed to help to realize the correct segmentation of LED spot image by iteratively calculating the image threshold, so as to obtain the structural displacement under low-light conditions even with environmental interference. Furthermore, a simply supported beam model and a cable-stayed bridge model were experimentally investigated under low-light conditions with strong-light and fog interference, and the final results demonstrated that comparing to existing Otsu method and template matching method, the developed method presents an excellent anti-disturbance ability against the environmental interference to realize the correct identification and guarantee the computational efficiency.