Monocular Vision-Based Surface Parameters Estimation for the Tailings Pond

Abstract

Safety monitoring of the tailings pond is one of the most important actual issues needs to be solved in the mining development. In this paper, facing the problems that large tailings pond covers a great span and there are no stability features on the tailings pond surface, a novel surface parameter estimation method based on some typical marking poles and the monocular vision-based measurement techniques has been proposed. In this method, after internal and external parameters calibration of the camera system, image segmentation and pole detection are carried out firstly using a linear combination of the suitable color components. Secondly, precise positioning of the contact point between the pole and pond surface will be done based on the priori knowledge about the pole position and the model about the shape of contact area, and then the final plane parameters of the pond surface will be estimated accurately using the contact points of 4 marking poles. Experimental and real case application results show that this proposed method can well meet the accuracy requirements of large tailings pond's surface measurement and safety monitoring.