The application of shape descriptor analysis to the study of aggregate wear

Abstract

Two identical samples of fresh rock aggregate have been artificially abraded in two rotating mills. Two subsamples consisting of 16 pieces from each mill were traced throughout 90 h of gentle abrasion and were examined at various stages. The examination used a video camera to capture data from outlines of silhouettes of the 16 pieces. Maximum projections were analysed using Fourier and Fractal methods to produce shape descriptor values which monitored shape, texture and roughness at different stages of the abrasion process. Full accounts of the methods employed are reported. Random projections were analysed in less detail. Reproducibility of the textural descriptors between the two subsamples is good when examined by maximum and by random projections, the latter method showing more scatter. The objectivity and speed of these methods of shape quantification have already been exploited in other disciplines, but this study appears to be the first time they have been used to assess the performance of aggregates and more specifically, armourstone used in coastal protection works. The results are discussed in this context.