Means of eliminating background effects for defect detection and visualization in infrared thermography

Pulse infrared thermography is renowned as an effective and rapid method of nondestructive inspection. The primary difficulty in making useful interpr etations of a Thermal image is the presence of extraneous effects such as t hermal noise, optical distortion through the thermal imaging system, and of most concern, non-uniform heating through the uneven excitation of the sur face. A means for removal of background trends caused by uneven illuminatio n is presented. It is based on a polynomial fitting technique applied to a rectangular infrared image on a line-by-line basis in either horizontal or vertical directions. The polynomial coefficients are found by choosing thre e key pixels through a conditional iterative technique. The proposed algori thm enables improvements in flaw visibility and assists the effective appli cation of additional image processing techniques. The procedure is found to remove trends without distorting real flaw indications as long as the indi cation spans less than 67% of the interval under examination. (C) 2000 Soci ety of Photo-Optical Instrumentation Engineers. [S0091-3286(00)01904-8].