Density feature clustering method for detection of internal defects in infrared non-destructive testing

Aiming at the detection of internal voids, internal expanded void defects and through-hole defects in aerospace composite materials, blind holes, conical through holes with different diameters and through holes with the same diameter were experimentally fabricated to simulate different diameters and types of defect existing in such materials. On this basis, the defects were detected using the pulsed thermal stimulation infrared non-destructive testing method. Since the original thermal wave image recorded by the infrared thermal imager cannot fully reflect the defect information, the defect recognition ability is enhanced and the contrast between the defect region and the non-defect region is improved by using the image sequence reconstruction algorithm based on feature density clustering. Finally, the proposed method was compared with thermographic signal reconstruction (TSR), principal component analysis (PCA) and pulsed phase thermography (PPT) and the superiority of detection in internal voids and internal expanded void defects was verified.