PSI - Issue 2_A

G. Meneghetti et al. / Procedia Structural Integrity 2 (2016) 3185–3193 G.Meneghetti/ Structural Integrity Procedia 00 (2016) 000–000

3188

4

is removed from the surface by compressed air. The microscope and the specimens area are obscured from ambient light by means of a cover; a number of leds provides repeatable light conditions. However, the surface geometry originate undesired shadows on the image, which may cause malfunctions in the algorithm for pits recognition. In order to minimize the effect of different light conditions, the images are processed as follows: 1) A region of interest (ROI) is defined in the image and the gray level of each pixel is evaluated; 2) For each pixels row in the ROI, a moving average over the gray levels of 100 pixels is calculated; 3) The averaged frame is subtracted to the original image; 4) All pixels having a grey level greater than a predefined threshold are considered to belong to a damaged area; 5) The damages may be filtered by shape and dimensions in order to discard all candidate defects which can not be attributed to pitting damage; Finally the total extension of damaged area is compared with a target defined by the user and adopted as failure criterion to stop the test.

Figure 2: example of image acquisition and processing on a pair of cylindrical specimens with chamfered edges.

Figure 3 – Slide-to-Roll Ratio along the tooth profile.