Non-uniformity correction and sound zone detection in pulse thermographic nondestructive evaluation

- A new segmentation technique for removing non-uniformity is introduced.
- A suitable bi-variate function is chosen to represent non-uniformity.
- A method to automatically identifying sound zone pixels is introduced.
- These procedures are validated using metallic and composite specimens.

An important requirement in the thermographic nondestructive evaluation is the identification of actual sound zone or the base line with which the defective areas are compared to determine the actual temperature contrast and the corresponding defect severity. In a part under inspection, the actual sound zone is not known a priori and various approximations have been used in the past to serve as this unknown base line. Determination of actual sound zone in a defective test object is still a challenge. A related issue before the identification of this base line pixel is in the elimination of non-uniformity in the temperature distribution across the specimen surface which is unrelated to the actual defects. This spurious contrast is often introduced by the limitations of the instrumentation or the test procedure and it has to be eliminated before pixels in the sound zone can be located. This paper presents an automated procedure for simultaneously eliminating spurious contrast and locating sound zone pixels, directly from experimental data in a thermographic nondestructive evaluation. Location of actual sound zone pixels facilitates accurate thermal contrast computations, extraction of thermal properties such as break time and thermal diffusivity. In addition, based on the actual sound zone temperature profile it is possible to normalize experimental thermographic results in such a way that they can be directly correlated with results from numerical simulations.