Evaluation of Perspective Error in Image Processing Method for Size Distribution of Blasted Muck-pile

Achieving the optimal size distribution and fragmentation of blasted muck-pile, has an important role on the controlling of mining processes and mineral processing as well as production costs reduction. In order to optimize the production process and to have rapid and reliable evaluation of blasting results, one of the common modern methods in size distribution of rock fragmentation is image processing method. Despite of numerous advantages of this method, the associated inherent limitations and errors affect the precision, accuracy, and reproducibility of measured results. Understanding such limitations and errors and decrease their effectiveness will improve the results of image processing. In this study, while introducing different types of image processing errors, errors of inappropriate imaging angle (perspective distortion) analyzed and reduction procedures have been suggested. The results of studies on 240 digital images of blasted fragmented rocks, show an average of 97% and standard deviation of 4.5% for elimination of perspective errors. Also, when the distance factor is 0.8 to 1. 5, the results are more reliable. Moreover, with a smaller particle size, optimal results can be achieved when the distance factor is reduced to 0.2 to 0.4. Furthermore, the frequency of distance factor in a state of lognormal distribution of data, is in the range of 0.5 to 1.5 with the average of 1.48, has the range of frequency error with an average of 3%. Also, the comparison of sieve size distribution curve of muck-piles with the size distribution curve of images taken from it, before and after perspective error elimination using auto mode of Split Desktop software, showed that after perspective error elimination, mean of size distribution error, depend on the amount of perspective distortion, were decreased 5 to 25 %.