Defect detectability based on square wave lock-in thermography
Pengfei Zhu, Dan Wu, Yifan Wang, Zhifei Miao
Abstract
A lock-in thermography technique based on a periodical square wave is used to detect stainless steel plates with defects. Combining a neural network with lock-in thermography, an image processing technique is proposed, and the results are compared with traditional image processing methods. A full-field defect reconstruction technology is proposed that combines pulsed phase thermography, threshold segmentation technology, and lock-in thermography technology to reconstruct the full-field depth image. This method has fast processing speed and high detection accuracy. Finally, the effects of excitation frequency and duty cycle on thermal image quality, defect detection range, and defect detection accuracy are investigated through extensive experiments to arrive at the optimal excitation frequency and duty cycle.