1996 Volume 14 Pages 109-129
This paper reviews techniques for obtaining and processing images of particles and objects (irradiation, response, surveying, storage and analysis). It presents the methodologies by which images can be classified and recognised (detection of domains, boundaries, shapes and textures) with a view to developing correlations between the information provided by the images and the physical and chemical properties of the examined particles. It discusses procedures for recognising the boundaries of images of particles, for analysing their properties (geometrical, Fourier series, fractals, etc.) and for recognising the structures and textures of multi-component particles. The instruments used to acquire and store the images and the hardware needed to process the digitized information are described and assessed. The software needed to analyse the data (loading, colour-level correction, enhancement, filtering, thresholding and labelling) are presented. Procedures for classifying vector structures that may be used to characterise images of individual particles (pattern vectors) or classes of particles (feature vectors) are discussed, and those which are used to recognise particles are analysed. Applications of these methodologies and several case studies concerning mineral grains (free and associated), polished sections of minerals, macerals, inorganic and organic materials are described.