TVR-DART: A More Robust Algorithm for Discrete Tomography From Limited Projection Data With Automated Gray Value...

Area of Science:

• Image reconstruction
• Discrete tomography
• Computational imaging

Background:

• Discrete tomography (DT) is crucial for analyzing objects with limited material compositions.
• Existing algorithms like DART can be sensitive to noise and require extensive parameter tuning.
• Automated gray value estimation is needed for robust DT reconstructions.

Purpose of the Study:

• To introduce a novel, automated, and robust iterative reconstruction algorithm for discrete tomography.
• To enhance image reconstruction accuracy and reduce parameter tuning effort compared to existing methods.
• To provide an easy-to-use tool for the tomography community.

Main Methods:

• Developed the total variation regularized discrete algebraic reconstruction technique (TVR-DART).
• Incorporated simultaneous exploitation of two prior knowledge types within an optimization framework.
• Utilized compressive sensing principles and automated estimation of gray values and thresholds during iterations.

Main Results:

• TVR-DART demonstrated more accurate reconstructions than existing algorithms under noisy conditions.
• The algorithm performed well with limited projection images and/or small angular ranges.
• Achieved improved robustness and reduced parameter tuning requirements compared to the original DART algorithm.
• Validated through extensive experiments on simulated, μCT, and electron tomography data.

Conclusions:

• TVR-DART offers a significant advancement in discrete tomography reconstruction.
• The algorithm provides accurate and robust imaging, particularly in challenging data scenarios.
• It presents a user-friendly and effective solution for discrete tomography applications.