MACHINE VISION-BASED MOTION ESTIMATION OF FLOTATION FROTH USING MUTUAL INFORMATION

摘要

The motion estimation of froths in the flotation ofrnminerals is difficult due to the effects of bubblerndeformation, bursting and merging making it difficult forrnthe traditional machine vision methods to estimate thernfroth velocity. In this paper, we propose a new method forrnthe motion estimation of flotation froth using mutualrninformation with a bin size of two (MI2) as the blockrnmatching metric. Experimental results show that thernproposed motion estimation technique improves thernmotion estimation accuracy in terms of peak signal-tonoisernratio of the reconstructed frame. The computationalrncost of the proposed method is almost the same as thernstandard machine vision methods used for the motionrnestimation of flotation froth.