Wavelet aided multivariate outlier analysis to enhance defect contrast in thermal images

Abstract

A novel two-stage signal reconstruction approach is proposed to analyze raw thermal image sequences for damage detection purposes by Infrared Thermographic NDE. The first stage involves low-pass filtering using Wavelets. In the second stage, a Multivariate Outlier Analysis is performed on filtered data using a set of signal features. The proposed approach significantly enhances the defective area contrast against the background in infrared thermography NDE. The two-stage approach has some advantages in comparison to the traditionally used methods, including automation in the defect detection process and better defective area isolation through increased contrast. The method does not require a reference area to function. The results are presented for the case of a composite plate with simulated delaminations, and a composite sandwich plate with skin-core disbonds.