CHARACTERIZATION OF POWDERS BY VIDEO IMAGE-ANALYSIS

The application of video image analysis for the automatic characterisation of powders was investigated. As model material, pea samples were ground both in a hammer and a roller mill. Eleven different granulometries of powders were obtained by varying the screen apertures of the hammer mill and the gap between rolls. One hundred and ten video images of the powders were collected and the texture was characterized by the grey level run lengths method (Galloway, 1975). According to their granulometries measured by analytical sieving, powders were classified into five qualitative groups. To identify the group of each powder, discriminant analysis was applied to the data assessed by the grey level run lengths method. About 75% of the samples were assigned to the correct group, both in the "learning set" and in the "verification set".