PSI - Issue 57

Haelie Egbert et al. / Procedia Structural Integrity 57 (2024) 179–190 Haelie Egbert et al. / Structural Integrity Procedia 00 (2019) 000 – 000

185

7

Figure 5. Binary reduction of high-speed gear tooth image (a) before adaptive local thresholding and (b) after thresholding

Figure 6. Processed binary gear tooth image (a) without a visible crack and (b) with visible crack The binary image reduction was applied to all frames, such that the crack surface area was defined by contrasting pixels. The crack is often not optically detectable through the raw or processed camera images at the beginning of the 6,207 cycles recorded. Figure 6(a) shows an image of the tooth when a crack is not visible to the camera and Figure 6(b) shows an image of a tooth from the same test when a sizeable crack is visible to the camera. As the crack is the region of interest for the measurement and the image processing algorithm, all other portions of the image could be discarded to result in an image as shown in Figure 7, which shows the cropped portion of the crack surface area defined by the binary reduction of the image. For each frame, a distance formula was employed to calculate crack length within the image. The algorithm began each section by scanning all rows and columns of the frame for a black pixel, correlating to a zero value. All black pixels were then recorded and sorted from least to greatest, to find the first row and column with a black pixel in it. This first row and column corresponded to the top left most black pixel within the frame, as indicated in Figure 7(b) by Point A. Rows and columns were then sorted greatest to least to find the last row with a black pixel, shown in