Evaluating a calibration method for the estimation of fragmented rocks 3D-size-distribution out of 2D images

Image analysis is a tool having great potential for the quantification of fragmented rocks size-distribution. This recent technique should be validated according to sieving measurements. Current acquisi-tion allows the obtaining of only 2D information from images. In addition, the 3D passage is extremely prob-lematic. Firstly, the usual stereological methods do not work because the reasoning is carried out on non-random projections. Secondly, estimated sizes differ from those measured by sieving. Last, a quantity of mat-terial measured during sieving remains inaccessible by image analysis, due to masking and segregation. This paper evaluates a method directly connecting 2D raw data to 3D sieving measurements. The cited problems are reduced as much as possible thanks to the adoption of an isotropic 2D sorting criterion, and thanks to samples without masking and segregation. Important results have been obtained. Firstly, the characteristic size is correctly estimated in 2D, giving access to computing the uniformity index through known models. Secondly, the stereological approach, bringing into play proportions, is not sufficiently robust to reconstruct easily volume distributions. Limits of this kind of methods, which are, unfortunately, currently used in several granulometric applications, are analyzed.