A new image analysis technique for the quantitative assessment of microcracks in cement-based materials

The development of a new image analysis technique for the automatic detection and quantification of microcracking in cement-based materials is described. An impregnation technique is used to highlight microcracks and other microdefects (porosity, air bubbles) present in the concrete matrix. Microcrack observations of the polished samples are carried out using an optical microscope. The first step of the image analysis process consists of extracting the various defects from the observed image. The extraction process is essentially an automated thresholding operation carried out on the grey-level histogram. The second step consists of automatically sorting the various extracted defects on the basis of a certain number of shape factors. For the third step, an innovative technique is used to reconnect several neighboring microscopic fields. Finally, the characteristics of the cracking pattern are quantitatively determined using classical stereological techniques. The approximation of microcracks by straight lines enables researchers to draw maps of the sample crack network over an area of a few tens of square centimeters. Various tests were carried out to verify the accuracy and the reproducibility of the results yielded by the method. Results obtained for undamaged and mechanically loaded samples are also presented.