Ultrasonic C-scan image restoration method using the Richardson-Lucy algorithm and a flaw measurement model

Abstract

In ultrasonic imaging, blurred edges have a heavily negative impact on the accuracy of the flaw images. To improve ultrasonic C-scan image quality, this paper proposes a deconvolution-based ultrasonic C-scan image restoration method combining the Richardson-Lucy (RL) algorithm and a flaw measurement model. In order to implement an optimal restoration process, an ultrasonic C-scan imaging model is presented and employed to simulate the C-scan images. By restoring the simulated images, a relationship between the optimal iteration number and the ratio of the flaw size to the sound wavelength is established. Furthermore, the equivalent sizes of the flaws are extracted from a flaw measurement model-based non-destructive approach and are used to calculate the optimal iteration number of the RL deconvolution algorithm. Finally, the experimental ultrasonic C-scan images are restored by using the RL iterative algorithm with the optimal iteration numbers, 1, 2, 6, and 14, respectively. The experimental results show that the flaw sizes, obtained by the 6 dB drop method in the original C-scan images, have errors larger than 26 %. After restoration, the maximum error of the flaw size in the C-scan image is only 10 % and the minimum is 2 %, indicating that the proposed restoration method effectively improves the accuracy of the ultrasonic C-scan imaging.