Fig. 1. GFRP sample. (a) Schematic of defect distribution; (b) the front and rear surfaces of sample

Fig. 2. Simulation results of phase-locked infrared thermography. (a) Variation of temperature of sample surface; (b) variation of surface phase with excitation frequency

Fig. 3. Flowchart of the multi-frequency phase-locking technique

Fig. 4. Selection of optimal excitation frequencies. (a) Theoretical relationship between defective phase difference and excitation frequency; (b) variations of L_f, P_f, and U_fwith defect depth

Fig. 5. Schematic of PCA algorithm

Fig. 6. Home-made infrared non-destructive testing system

Fig. 7. Raw phase images at different thermal excitation frequencies. (a) 0.126 Hz; (b) 0.080 Hz; (c) 0.054 Hz; (d) 0.038 Hz

Fig. 8. Enhanced phase images at different thermal excitation frequencies. (a) 0.126 Hz; (b) 0.080 Hz; (c) 0.054 Hz; (d) 0.038 Hz

Fig. 9. Detection results of GFRP sample. (a) Raw thermal image; (b) fused image

Fig. 10. Detected results for different excitations. (a) LPT; (b) DFMTWI (DOD); (c) DFMTWI (CC); (d) the proposed method

Fig. 11. Detected results with different image fusion algorithms. (a) MSD; (b) DWT; (c) LF; (d) PCA

Fig. 12. Optimal thermal wave excitation frequencies for different threshold values. (a) 60%; (b) 70%; (c) 80%; (d) 90%

Fig. 13. Variation of SNR with threshold value