The Overlap Problem In On-Line Coarse Particle Size Measurement -Segmentation And Transformation

The most serious problem for on-line coarse particle size measurement from a conveyor belt by image analysis is the overlap phenomenon that occurs between adjacent particles. A methodology is developed and described in this paper to estimate the sieve size distribution from the overlapped projected particle images as might be obtained from a conveyor belt. Two aspects of the overlap problem have been taken into consideration. First, several image processing algorithms were studied to find the most suitable technique to separate the projected overlapped particle images for narrow sized particles. These image segmentation algorithms include two-dimensional thresholding, edge detection and Watershed discrimination techniques. The other aspect of the overlap problem is related to the determination of the general kernel function for the partial particle images obtained from the projection of the overlapped particle bed. In this regard, a computer simulation technique was applied to mimic the overlapped particle bed for narrow sized particles, then the general kernel functions of the partial particle images were determined.