Assessing visual and system flaw detectability in nondestructive evaluation

NDE methods assume capability to detect certain flaw size. Sometimes the capability is validated using probability of detection (POD) demonstration. Although, NDE imagers may highlight indications that meet certain indication detection criteria, the inspection results are evaluated by a certified operator and the operator takes responsibility of the results of his assessment. The operator assesses both i.e. the NDE data of the indication and the visual indication itself. Normally, assessment of NDE data also has a visual component and may also have visual inspection of the part. Visual assessment of image data and visual inspection of part are two kinds of visual flaw detection. Every visual flaw detection has either implicit or explicit flaw detectability size associated with it. Explicit flaw detectability size has POD/confidence associated with it. The paper provides methodology to assess flaw detectability in visual detection. Indication contrast, indication size, background illumination, and noise are important factors in flaw detectability. Visual perception is affected by these factors. Visual acuity of operators is measured using Jaeger chart. Visual resolution is measured by another visual acuity test using line-pair targets. Contrast sensitivity as a function of line-pair frequency can be measured using a line-pair target with sinusoidally varying luminance across the line-pair widths with logarithmically decreasing contrast along length of the line-pairs. Operator contrast sensitivity function, indication contrast-to-noise ratio (CNR) indication size are necessary for assessment of flaw detectability in visual inspection. Alternately, a function called CNR sensitivity function, CNR and new parameter called resolution ratio have been proposed to relate to reliability of detecting indication. In visual assessment of data, important factors to consider are system resolution, resolution of data display monitor, contrast sensitivity, indication contrast, indication size, background illumination, and noise.