Infrared Thermography and Nondestructive Evaluation

The domain of infrared thermography is recent and covers vast fields of applications. In the industrial context, infrared thermography, or TNDE (Thermographic Non Destructive Evaluation), is deployed either by the passive approach (by simple observation of the isotherms on the observed surface) or by the active approach (by stimulating the thermal response of the specimen). The interest of these thermal methods is obvious since they allow to probe a crucial thickness of material under the surface in order to reveal potential subsurface defects.

These research activities aim at the development of experimental methods and processing methods allowing to take decisions at a high level for the automatic interpretation of thermograms (thermal pictures): interpretation, temporal quantitative analysis, development of procedures to extract specific features.

It is worthwhile to mention that in most of the cases where TNDE is deployed the interpretation of the thermograms is still done by hand by the operator. A pressing need exists therefore in order to go on to the following stage and achieve the symbiosis: infrared thermography and automated vision. Classic image processing techniques (for images of the visible spectrum) are not directly applicable because of the very different formation process of infrared images (emitted rather than reflected radiation). It is thus necessary to adapt existing methods and especially to innovate in order to set up general high level decision making methods from explicit extracted data of thermograms.

More specifically, our fields of activities are:

• survey of the problems occurring in the case of non planar component inspection by a non uniform heat flux, Infrared vision and inspection
• simulation of infrared images by massively parallel processing,
• developments of the advanced concepts of pulse phase thermography (PPT) and its quantitative analysis,
• use of artificial intelligence concepts to solve the inverse problem,
• joint projects, such as the study of rottenness in electric power distribution poles and ageing in concrete structures, with various institutions
• set-up of a 'universal' JAVA-based software for infrared inspection,
• quantitative restoration of infrared images,
• investigation of tomographic infrared techniques to 'slice' a sample into thickness layers.

Some other projects are also considered, such as: the automated survey of phenomenons which release thermal energy (industries, road traffic), the detection of cracks by thermal lateral propagation, the application of transforms, thermal cyclic and chaotic stimulation, the increase of diagnostic reliability by data fusion.  It is also interesting to mention the publication of the following references which are now part of the Body of Knowledge of the American Society of Nondestructive Testing (ASNT) in infrared thermography:

• NDE of Materials by Infrared Thermography, X. Maldague, Springer-Verlag, London (UK), 1993. 224 p.+ 16 color.
• Infrared Methodology and Technology, X. Maldague ed., International Advances in Non Destructive Testing (IANDT), Gordon and Breach Publ., NY, 1994. 525 p.+ 17color. This monograph groups the original contributions of sixteen authors having a large experience in the field of infrared and coming from seven countries: Canada, France, United States, Italy, The Netherlands, Russia, Sweden.
• A new Handbook is under development.

We also collaborate with various international partners. For instance, a joint research project with the Catholic University of Louvain in Belgium is currently under way (1997-1999 period).

The team activities include also some aspects of automated inspection (visible spectrum). These activities are in close collaboration with the industry. Currently, the following projects are ongoing:

• automated decrypting of standard forms. Partner: CARDI Inc.
• automatic presence and position assessment of pins. Partner: MENDES Inc.
• fingerprint acquisition.

These aspects of automated inspection are very important for the involved partners. Indeed, the automated inspection allows to optimize production while reducing costs. All these activities allow to apply and develop, in a convenient and also useful way, some fundamental concepts of computer vision.