SAND GRAIN ANALYSIS - IMAGE-PROCESSING, TEXTURAL ALGORITHMS AND NEURAL NETS

A quantitative method for characterising and classifying quartz grain form by mathematical analysis of surface texture is presented. Scannin g electron images of quartz grains were 'frame grabbed' and converted to a digitised grey level image. Image enhancement, segmentation and h istogram equalisation were applied to produce standardised images. Two major textural approaches were then applied. The Roberts gradient ope rator determined the degree of surface edgeness while calculation of s patial grey level dependence matrices allowed production of distributi on maps of surface homogeneity, entropy and correlation. Textural para meters were obtained from samples of 0.5 mm quartz grains from three d istinct populations: Desert quartz; Fire Island, New York, beach grain s; and Brazilian crushed quartz. A comparative analysis using discrimi nant analysis and neural networks was undertaken to quantify the succe ss in classifying the different populations. Both methods achieved exc ellent degree of quartz grain classification. However the use of neura l networks provided a more robust method of analysis particularly when presented with incomplete data sets. (C) 1998 Published by Elsevier S cience Ltd. All rights reserved.