Many industrial manufacturing processes depend on the rapid and accurate measurement of circular objects, for example, pipe, tube and ball and roller bearings, to ensure that the components satisfy the engineering tolerance standards. Despite the importance of these measurements, they are often carried out using time-consuming, manual and randomly sampled procedures. Cross-sectional imaging techniques offer the advantage of simultaneous dense measurement of dimensions and the detection of defects. Levels of precision achievable using x-ray computed tomography (CT) to measure circular objects are determined here. Edge detection procedures have achieved subpixel measurement to a precision of 3.5 x 10−2 pixels (17 microns) using a medical x-ray CT scanner. Even when the image function is properly sampled and the image is a perfect isotropic disc, it is possible for significant anisotropic effects to be introduced when sampled data is processed. Methods of overcoming such effects are...
Russell Horney, Imants D. Svalbe, Peter Wells