A Study of Efficiency, Accuracy, and Robustness in Intensity-Based Rigid Image Registration

Image registration;Image processing;GPU;Mutual information;Mathematical model;Optimization;Computer Science

Xu, Lin

University of Waterloo

关键词: Image registration;    Image processing;    GPU;    Mutual information;    Mathematical model;    Optimization;    Computer Science;   

Others  :  https://uwspace.uwaterloo.ca/bitstream/10012/4077/1/uw-ethesis.pdf

瑞士|英语

来源: UWSPACE Waterloo Institutional Repository

【 摘 要 】

Image registration is widely used in different areas nowadays. Usually, the efficiency, accuracy, and robustness inthe registration process are concerned in applications. This thesis studies these issues by presentingan efficient intensity-based mono-modality rigid 2D-3D image registration method and constructing a novel mathematicalmodel for intensity-based multi-modality rigid image registration.For mono-modality image registration,an algorithm is developed using RapidMind Multi-core Development Platform (RapidMind) to exploit the highlyparallel multi-core architecture of graphics processing units (GPUs). A parallel ray casting algorithm is usedto generate the digitally reconstructed radiographs (DRRs) to efficiently reduce the complexityof DRR construction. The optimization problem in the registration process is solved by the Gauss-Newton method.To fully exploit the multi-core parallelism, almost the entire registration process is implemented in parallelby RapidMind on GPUs. The implementation of the major computation steps is discussed. Numerical resultsare presented to demonstrate the efficiency of the new method.For multi-modality image registration,a new model for computing mutual information functions is devised in order to remove the artifacts in the functionsand in turn smooth the functions so that optimization methods can converge to the optimal solutions accurately and efficiently.With the motivation originating from the objective to harmonize the discrepancy betweenthe image presentation and the mutual information definition in previous models,the new model computes the mutual information function using both the continuous image functionrepresentation and the mutual information definitionfor continuous random variables. Its implementation and complexity are discussed and compared with other models.The mutual information computed using the new model appears quite smooth compared with the functions computed by others.Numerical experiments demonstrate the accuracy and efficiency of optimization methodsin the case that the new model is used. Furthermore, the robustness of the new model is also verified.