RESUMEN

An automated image analysis procedure for the segmentation of anhydrous fly ash from backscattered electron images of hydrated, fly ash blended Portland cement paste is presented. A total of six hundred backscattered electron images per sample are acquired at a magnification of 2000. Characteristic features of fly ash particles concerning grey level, shape and texture were used to segment anhydrous fly ash by a combination of grey level filtering, grey level segmentation and morphological filtering techniques. The thresholds for the grey level segmentation are determined for each sample by semiautomatic histogram analysis of the full image stack of each sample. The analysis of the presented dataset reveals a standard deviation of the reaction degree of fly ash of up to 4.3%. The results agree with a selective dissolution method to quantify the reaction degree of fly ash showing the potential of the presented image analysis procedure.