Real-Time Froth Velocity Extraction of Zinc Flotation Based on Improved SURF

Abstract

The zinc flotation efficiency and its production index are closely related to the froth velocity which involves significant information about the concentrate grade and mineral recovery. Given the situation that the froth surface is easy to rotate and collapse, of which large deformation cannot be calculated easily, it is difficult to extract the froth velocity accurately and in time. The traditional static froth velocity extraction method based on image template matching detection is far from perfect in matching accuracy and speed both. Therefore, a realtime froth velocity extraction method of zinc flotation based on improved SURF is presented in this paper. Firstly, the SURF with rotation invariance, scale invariance and illumination invariance is introduced. Due to the unsatisfying accuracy of the descriptor, the improved descriptor along with 96-dimensional features is proposed. Then, in order to overcome the deficiency of the extra time it costs, an iterative update descriptor matching method is proposed to improve the matching speed and accuracy by analyzing the froth motion track. Finally, taking advantage of the improved SURF descriptor and new matching method, the dynamic characteristic of froth velocity is extracted from froth monitoring video. The comparative experimental results show the superiority and robustness of the proposed method, which can dynamically and accurately reflect the working condition in the actual froth video sequence, laying the foundation for real-time optimization of zinc flotation based on machine vision.