PSI - Issue 77

Andrzej Katunin et al. / Procedia Structural Integrity 77 (2026) 18–25 Author name / Structural Integrity Procedia 00 (2026) 000–000

23

6

resulting visualization, the obtained fused thermograms for each scenario were raised to the power of 0.01. The resulting images are presented in Fig. 4.

(a)

(b)

(c)

Fig. 4. The fused enhanced thermograms for damage scenarios: (a) 10%, (b) 25%, (c) 50% of thickness reduction in the damaged area.

The results demonstrate increasing of edge sharpness with respect to the best raw thermograms presented in Fig. 2, and simultaneously significant reduction in measurement noise. 3.3. Damage quantification To quantify detected damage in the acquired fused enhanced thermograms and automate the quantification process, the Niblack thresholding algorithm (Niblack, 1985) was applied with the window size of 50 pixels and the multiplication factor of 0.2. This algorithm was selected due to its local character, allowing extraction of a damage signature from the non-uniform background by calculating the local threshold according to the following formula: = � + ( ) , (4) where � and ( ) are the mean value and standard deviation from the local area defined by window centered around the considered pixel. To quantify the identified damage, the resulting binary image was used. Based on the achieved contrast, the boundary between damage signature and healthy region is clearly visible, and the damage signatures were manually circumscribed using the polygonal line. Finally, the determined envelope of a damage signature (red line) along with the true damage contour (black line) were superimposed on the fused enhanced thermograms for each scenario, which was presented in Fig. 5.

(a)

(b)

(c)

Fig. 5. The results of damage quantification for: (a) 10%, (b) 25%, (c) 50% of thickness reduction in the damaged area.