Method for Optimizing Sample Orientation to Maximize Defect Visibility in X-ray Projections

Rapid detection of internal defects is a key requirement in industrial non-destructive testing (NDT), where conventional X-ray computed tomography (CT) often remains impractical due to long acquisition times and the need for full 3D reconstruction. In this work, we present a simulation-based method for optimizing sample orientation to maximize defect visibility in individual X-ray projections. The approach leverages a prior CT scan to generate geometric models of the inspected object and its defects, which are then used to simulate projections from multiple viewing directions. A defect-based visibility score is computed for each projection by evaluating local contrast and defect size relative to the surrounding material. The proposed metric is compared with simpler global projection descriptors based on average and minimum attenuation and is evaluated on two samples with similar geometry but different defect distributions. The results show that the defect-based score consistently identified orientations with enhanced defect visibility and exhibits strong correlation across samples, while global intensity-based metrics fail to reliably predict defect detectability. The method enables systematic selection of optimal projection views and supports rapid, task-specific inspection workflows without requiring full 3D reconstruction.