Abstract:
This master thesis addresses the need for effective validation and optimization of computed tomography (CT) scan trajectories, crucial for industrial applications. The research focuses on designing and automating the creation of 3D scanning objects that can systematically test and verify the performance of trajectory optimization algorithms. The central research question explores how to design such objects using tools like Blender, while ensuring that these test scenarios are both efficient and scalable. A key goal is to generate a comprehensive dataset of these scanning objects, enabling the evaluation and comparison of various trajectory optimization methods.
Research Objectives:
1. Designing Scanning Objects: Establish a method for creating 3D objects in Blender that specifically target challenges faced in CT trajectory optimization, such as irregular geometries, material contrasts, and complex edge structures. These objects will serve as benchmarks for evaluating trajectory algorithms.
2. Dataset Creation for Trajectory Evaluation: One of the core deliverables of this thesis is to generate a standardized dataset of 3D objects. This dataset will enable comprehensive evaluation and comparison of different CT trajectory optimization algorithms, using metrics such as scan efficiency, image quality (measured by SSIM, PSNR), and artifact reduction.
3. Trajectory Optimization Validation: Evaluate CT trajectory optimization methods using the generated dataset. Simulate scan trajectories and validate algorithm performance based on the reconstructed image quality and optimization of scan time. Metrics such as structural similarity, noise reduction, and coverage of scan angles will be analyzed.