Defect identification based on parameter estimation of histogram in ultrasonic IR thermography

Abstract

Ultrasonic IR thermography is a powerful tool for the inspection of cracks, but the quality of raw image sequence varies greatly due to many factors. In order to facilitate an automatic identification of low contrast defects in ultrasonic thermography, a defect identification method was proposed based on parameter estimation of gray-level histogram. Taking an aluminum alloy beam as a sample, the crack identification was demonstrated. After selecting the suitable data from the raw image sequence, the mean value and standard deviation of the histogram for a kind of excess thermograms were estimated, and then the image enhancement and defect segmentation were performed according to the statistical parameters. An unsupervised process flow was concluded together with adaptive operation parameters. The results show that the method is fast, effective, robust and adaptive for various qualities of experimental data acquired in different test conditions. The method can improve the detectability of ultrasonic thermography and facilitate the automatic identification of defects.