Infrared Image Segmentation Algorithm Based on Multi Structure Morphology—Pulse Coupled Neural Network in Application to the Inspection of Aerospace Materials

Infrared images of aerospace materials were collected by infrared camera and analyzed to evaluate debonding defects. However, it is difficult to identify the defects effectively because of a considerable background noise, low contrast and fuzzy edge in the images. The MSM-PCNN algorithm is developed to segment infrared images and extract defect features. The infrared images of TBCs, CFRP and Al-HP samples with debonding defects were selected to be segmented by the PSO, RG and MSM-PCNN algorithm, and PSNR, SSIM, MSE and MAE were selected as image quality evaluation criteria. It is shown that the MSM-PCNN algorithm keeps more details about defects to be detected on a noisy background.