Technical Note: Correction of geometric x-ray image intensifier distortion based on digital image correlation

Purpose X-ray image intensifiers (XRIIs) inevitably produce images suffering from geometric distortion. Presently, various local and global methods exist to correct for these distortions. However, the performance of global methods is limited for dominant local distortions, and local methods tend to suffer from patch discontinuity and are generally sensitive to noise. In this paper, a novel local method is presented based on digital image correlation (DIC), which does not suffer from patch discontinuity and noise. Methods As DIC is a very accurate and robust technique to analyze deformations, it is our candidate of choice to outperform the existing correction methods. The performance of our technique was first validated through distortion simulations. Next, it was validated experimentally for four different orientations of the XRII. Results A theoretical study on images suffering from a simulated distortion (including noise and blurring) yielded corrections with an average accuracy of (0.20 +/- 0.04) pixels. We obtained experimental data with our 14" XRII (292 mm field of view), suffering from a maximum distortion between 9.6 and 12.9 mm, and an average distortion between (4.4 +/- 1.3) mm and (6.1 +/- 2.5) mm over the image field for the different orientations. For an adequate choice of the facet size in the DIC analysis (greater than 40 pixels), the weighted mean residual error of our method varied between (0.037 +/- 0.003) mm and (0.054 +/- 0.003) mm, regardless of the XRII orientation. The maximum residual error varied between 0.081 and 0.185 mm. Conclusions From the simulations, we concluded that the proposed technique is affected by neither Gaussian noise nor blurring. Furthermore, it is shown that our method can reach an accuracy that is on par with or better than the current standard tools. The novel method is fast, requires minimal operator intervention and can be fully automated.

Publication year:2020

Keywords:A1 Journal article

BOF-keylabel:yes

BOF-publication weight:1

CSS-citation score:1

Authors from:Higher Education

Accessibility:Open