PSI - Issue 37

Kafayat Eniola Hazzan et al. / Procedia Structural Integrity 37 (2022) 274–281 Hazzan and Pacella/ Structural Integrity Procedia 00 (2019) 000 – 000

277

4

These regions were analysed using the boundary function (bwboundaries), to trace regions in the filtered image. The boundary function was modified in this study to exploit the information it returns at identified crack regions (Fig 2. d). The output of the function was analysed for pixel continuity. There were some instances where very close regions were computed as separate cracks. In these cases, the regions were combined to class the regions as one crack region (Fig. 3 a). The output of this method generates an array of identified crack regions, the length of which, provides the quantity of cracks on the surface. Each element of the array is an individual array of 2 columns containing the X and Y co ordinates (Fig 3. b). The size of the arrays is dependent on the number of cracks and physical size of the crack. The area of each crack was calculated by plotting the co-ordinates and using the convhull function. Fig 3. c compares the cracks in the identification method (IM) to the original SEM file.

Fig. 3. a) Pixel analysis for continuity, regions in proximity combined for example top: 3 regions identified classed as one crack; bottom: 4 regions identified classed as one crack. b) Example of n x 1 array of identified cracks each element containing an m x 2 array. c) Cracks identified plotted using information retrieved and generated from the identification method.