The high noise and low resolution of ultrasonic testing images of large-thickness honeycomb sandwich composite components bring challenges to quality evaluation and defect identification. In this study

image denoising and super-resolution reconstruction methods are investigated using ultrasonic C-scan images of the C919 inner door for the main landing gear. An efficient edge-preserving filter denoising scheme based on a gradient descent algorithm is proposed

effectively removing speckle noise while preserving image details. After denoising

the peak signal-to-noise ratio reaches 37.53 dB and the structural similarity is 0.92

outperforming the digital morphological filter by 12.99 dB and 0.04

respectively. The images reconstructed by the improved super-resolution residual network (ISRResNet) model have high resolution

rich content details

and clear edges. Furthermore

an image with Φ11 mm delamination defects is processed

the signal-to-noise ratio of defects is improved by 6.25 dB on average. Results show that the proposed denoising method and super-resolution model can effectively remove speckle noise

improve the image resolution

and enhance the accuracy of defect quantification. It can support high-quality ultrasonic testing of large-thickness honeycomb sandwich composite components.