Quantitative Abel Tomography Robust to Noisy, Corrupted and Missing Data

A mixed-variable optimization (MVO) approach to quantitative tomography was applied to experimental x-ray data. The results were found to be comparable to previous tests on synthetic data. The MVO method was tested for robustness to realistic data problems: actual radiographic occlusions, simulated amplified noise, and random pixel rejection. Significant levels of data corruption, which easily render inverse methods ineffective or unreliable, do not noticeably impact the MVO method reliability. The success of the MVO method lies in its use of a reduced dimension object description designed to capture prior knowledge about the class of potential imaged objects.