fourier-based characterization of roughness of irregular particles in sem images

Surface roughness plays a vital role in macro and micro-scale granular behaviors. this paper presents a novel technique for the objective assessment of particle roughness. the technique uses fourier descriptors and image analysis of scanning electron microscopy photographs. low pass filtering of fourier descriptors is used to produce several smooth versions of the shape boundary at different smoothing scales. to quantify particle irregularity and roughness, the perimeter of these smoothed boundaries compared with the perimeter of the original particle boundary. the performance of the proposed criterion evaluated on several random subsets of the particles in an sem image dataset. these subsets ranked according to the computed roughness values, and the automatic rankings compared to the manual rankings generated by a human expert. after rank-based analyzes of the proposed roughness metric, it is found that the rankings generated by the proposed 2d metric are well correlated with manual rankings.