Diffusion Based Thermal Tomography

Thermal imaging is one of the fastest growing areas of nondestructive>testing. The basic idea is to apply heat to a material and study the way the temperature changes within the material to learn about its composition. The technique is rapid, inexpensive and most importantly has a wide coverage area with a single experimental measurement. One of
the main goals in thermal imaging has been to improved flaw definition through advanced image processing. Tomograpic imaging is a very
attractive way to achieve this goal. Although there have been some attempts to implement tomographic principles for thermal imaging, they
have been only marginally successful Similar results have been obtained with other imaging approaches such as those used in magnetic resonance imaging . The principal reason for this is that conventional tomography algorithms rely upon wave propagation (either electromagnetic or acoustic ) and are inherently unsuitable for thermal diffusion.

Thermal Imaging System

Thermal Imaging

As economics becomes an increasingly important consideration in materials evaluation, there has a major emphasis placed on thermal imaging. While it has not supplanted traditional methods such as ultrasonics or radiography as the method of choice for critical components, it has become widely utilized when cost and coverage area take precedence over resolution capability. The major shortcoming of thermal imaging to date is that it is principally a qualitative technique. While providing a rapid, low cost method for gross defect detection, thermal images provide little quantitative information about the size, shape, precise location ( especially depth information ) of potential flaws. Nor does conventional thermal imaging provide any insight about the nature or composition of a material as does ultrasound ( via stiffness / micromechanics analysis ).

It is important to note that, just as for sound and electromagnetism, conductive heat propagation in a solid is governed by well understood mathematical relationships involving fundamental material parameters ( thermal conductivity, heat capacity and density ). Hence thermal studies can, in principle, provide similarly quantitative materials characterization information as other methods. When coupled with an appropriate reconstruction algorithm, as for radiography ( CAT scan ), magnetic resonance imaging, or acoustics, similar three dimensional images of property distributions can, in principle, be generated with full dimensional accuracy . Since this type of information could be generated relatively simply and inexpensively using an IR camera with large coverage area instead of other dangerous ( CAT ), expensive ( MRI ) and/or cumbersome ( acoustic ) techniques with significantly limited coverage areas its attractiveness is clear.

Tomographic Image Reconstruction

The objective in the proposed research program is the development of a new diffusion based, fully tomographic approach to thermal imaging. The method should be fully three dimensional and take advantage of the wide coverage capabilities of thermal imaging wherever possible. Here the key is the development of a suitable model for the propagation of heat through a material so that the effects of each individual pixel properties on heat propagation can be determined. This information is then used as the basis of an iterative, diffusion based reconstruction scheme using a modified ART ( algebraic reconstruction technique ) approach to find the best match between local material properties and the measured time temperature profiles. This method is expected to be rapid, inexpensive and provide three dimensional material property information unavailable with any other nondestructive evaluation approach.