Particle Shape Characterization by Image Analysis
Résumé
Some behaviour traits of systems of particles depend on the shape of particles involved. Here we show how developments in digital image capture and subsequent automated analysis, favoured by the progress in computer science (hardware and software), can be used for detailed quantification of particle shape. Methods are described for the acquisition and treatment of digital images leading to binary images. It is then shown how these treated images can be used to give basic particle size descriptors which can be combined in different ways to give shape descriptors such as circularity and aspect ratios. We further show that the procedures involving operations such as erosion and dilation and the determination of convex bounding hulls, etc. can be used to give more detailed shape descriptors based on the concepts of mathematical morphology. Examples of these methods are presented to characterize the shape of twinned crystals and particle agglomerates. Finally, some cases can demand specific algorithms: this is the case for fractal-like structures or for biological particles, such as filamentous bacteria. In conclusion, the methods described here can be generalized to develop very specific shape descriptors for rapid characterization of any system, and are applicable to any system where valid images can be obtained.