Vision-based distortion removal in mirror scanning for laser ultrasonic systems

Abstract

Laser mirror scanner (LMS), which is a two dimensional scanner, have found wide spread use in various optical systems to steer the laser beam because of its high speed, compact size and low cost. Due to these benefits, laser mirror scanner is an effective laser steering tool that can be used in nondestructive testing (NDT) for on-site inspection. However, LMS inherits pincushion distortions in its scan-field due to its mirror arrangement. These pincushion distortions can considerably reduce the positioning accuracy of LMS. Therefore, in this dissertation a vision-based distortion removal method is proposed to remove the pincushion distortions from the LMS scan-field. In this distortion removal method, a RGB camera is utilized to generate correction table for the LMS by using a compensation technique. This correction table is able to remove the pincushion distortions of the LMS with an accuracy of 1~2 mm and it takes less than 140 seconds for 13 x 13 scan-field (i.e. 169 points) measurement process. During the implementation phase, the major challenge that was experienced and resolved is the distortion of the RGB camera. In addition, this dissertation also presents the implementation of distortion removal method in three different laser ultrasonic systems, which includes angular-scan pulse-echo ultrasonic propagation imaging (A-PE-UPI) system, guided-wave ultrasonic propagation imaging (G-UPI) system and robot pulse-echo ultrasonic propagation imaging (Robot PE-UPI) system. To demonstrate the performance of vision-based distortion removal method, various composite structures were inspected with the aforementioned laser ultrasonic systems. In the results, it was confirmed that the vision-based distortion removal improved the performance of all the LMS based laser ultrasonic NDT systems.