Comparative performance of a novel automated technique for identification of muck pile fragment boundaries

This paper evaluates the performance of a novel automated image analysis and size distribution estimation technique. The technique and its implementation are introduced. Experimental results of the technique are then presented, focusing on performance with respect to the algorithm’s effectiveness in dealing with size measurement. Next, results of a comparative study are presented: with respect to size distributions obtained by this method versus two other popular image-based methods reported in the literature. Our technique is shown to yield performance in good agreement with the physical (sieved) size distribution.