Comparative performance assessment of beam hardening correction algorithms applied on simulated data sets

Area of Science:

• Industrial Imaging
• Image Processing
• Metrology

Background:

• Industrial computed tomography (CT) is crucial for non-destructive testing but suffers from image quality degradation due to beam hardening artefacts.
• Beam hardening artefacts, appearing as cupping or streaks, impair image fidelity and dimensional accuracy in CT reconstructions.
• Existing correction algorithms, primarily developed for medical CT, require evaluation for their effectiveness in industrial CT applications.

Purpose of the Study:

• To compare the performance of four distinct beam hardening correction algorithms in industrial CT metrology.
• To assess algorithms based on image quality, artefact reduction, and dimensional measurement accuracy.
• To provide insights for selecting appropriate algorithms based on specific industrial needs.

Main Methods:

• Comparison of four algorithm types: single-material linearization, multi-material linearization, dual-energy, and iterative reconstruction.
• Application of algorithms to simulated dual-material phantom data under ideal and noisy conditions.
• Quantitative assessment using contrast-to-noise ratio, spatial resolution, artefact reduction, and dimensional measurement deviations.

Main Results:

• Single-material linearization partially reduces artefacts; multi-material linearization improves dimensional accuracy.
• Dual-energy methods eliminate artefacts and enhance low-contrast visibility and dimensional accuracy.
• Iterative algorithms reduce streaks but can cause aliasing, with performance dependent on computational power; noise degrades image quality.

Conclusions:

• The dual-energy method offers the most comprehensive solution for beam hardening correction in industrial CT metrology.
• Iterative methods are effective for streak reduction but require careful consideration of computational cost and potential aliasing.
• Algorithm selection should be tailored to specific industrial requirements for image quality, artefact control, and measurement precision.