General automated flaw detection scheme for NDE X-ray images
This paper presents an approach to automated flaw detection (AFD) in an arbitrary X-ray image. The intensities in the digitized radiographic image are modeled as piecewise-smooth surface functions corrupted by noise and flaws. It has been observed that radiographs generated for NDE purposes containing flaws also have a combination of three unwanted features; background trends, geometrical structures, and noise. These features inhibit the performance of automated flaw detection algorithms. The proposed general processing scheme reduces the unwanted features in such a way that candidate flaws within the image can be identified. The proposed scheme is robust and is applicable to a wide variety of NDE imaging applications.