Spatial Temporal Denoised Thermal Source Separation in Images of Compact Pulsed Thermography System for Qualification of Additively Manufactured Metals

We introduce a Spatial Temporal Denoised Thermal Source Separation (STDTSS) unsupervised machine learning (ML) algorithm for detection of material flaws in additively manufactured (AM) metals using pulsed thermography (PT) images obtained with compact infrared (IR) camera. Quality control of actual AM structures is necessary before their deployment, because AM metals can contain defects, such as pores, due intrinsic features of AM process. The PT nondestructive evaluation (NDE) method is based on recording material surface temperature transients with IR camera following thermal pulse delivered on material surface with flashlight. The PT method has advantages for NDE of arbitrary size AM structures because the method involves one-sided non-contact measurements and fast processing of large sample areas captured in one image. To reduce the cost and enable in-service NDE in spatially constrained environment, it is highly desirable to develop PT with compact and inexpensive IR camera. However, data cube obtained with PT based on compact IR camera suffers from strong thermal noises and loss of features due to relatively low sampling rate. The purpose of STDTSS algorithm is to compensate for uncertainties detection due to reduced resolution of low-cost hardware. The STDTSS algorithm consists of spatial and temporal denoising using Gaussian and Savitzky–Golay filtering, followed by the matrix decomposition using Principal Component Analysis (PCA), and Independent Component Analysis (ICA) to automatically detect flaws. We construct a compact PT system based on FLIR A65 camera, and demonstrate performance of STDTSS method in detection of imprinted calibrated defects in AM stainless steel 316L specimens.