Super-Resolution in Ultrasonic NDE

Abstract

This paper discusses the use of an iterative back projection (IBP) super-resolution (SR) image reconstruction technique on the carbon epoxy laminates with simulated porosity defects. In order to first validate and evaluate the application of the proposed method, three artificially simulated delamination defects in carbon epoxy laminates were considered. Based on the preliminary results, it was verified that the contrast signal-to-noise ratio (CNR) of the SR image was higher than the bi-cubic interpolation image. Further, the peak signal-to-noise ratio (PSNR) value of SR result had an average increase of 5.7088 dB compared to the bi-cubic interpolation method. This validates the proposed approach used to generate the reconstructed SR images with image quality similar to the original simulated UT images. After the validation, the UT image reconstruction algorithm was applied to the ultrasonic C-scan amplitude images of a porosity sample. Based on the results, it was demonstrated that the SR image achieved better visual quality with an improved image resolution. It was also demonstrated that this method was capable of detecting the defects with more confidence by recovering the defect outline compared to the LR C-scan image. The defect outline in SR images is more distinct to recognize, allowing post-processing work such as measurement of defect size, shape, and location to be much easier.