Computational analysis for thermal NDE of composites

Abstract

Nondestructive Evaluation (NDE) is an important tool for ensuring the inspectability of a structural design and assessing the integrity of the structure during fabrication and service. NDE test results are typically examined by an inspector to determine the location and size of damage. There is significant potential for reducing the human effort involved in this procedure by digitally processing this data to enhance the signatures of flaws and to perform automated identification of suspected flaws. Computational NDE focuses on the development of methods for the simulation of NDE techniques and reduction of NDE data for an assessment of the integrity of the structure. This paper examines a technique that enhances the contrast between damaged and undamaged regions to improve the quality and reliability of flaw identification. An anisotropic diffusion algorithm is applied to the data. Anisotropic diffusion techniques are shown to significantly reduce image noise while maintaining defect contrast and preserving the important features of a flaw. The use of this algorithm is shown to improve detectability levels for thermal NDE data for both standard array imaging infrared cameras as well as the cheaper, more portable microbolometers of interest today. By increasing and automating detectability, significant advances can be made in the use of thermal NDE tools.