In this thesis we describe and validate a statistical respiratory motion model for motion compensation during image-guided cardiac interventions. In a preparatory training phase, a preoperative 3-D segmentation of the coronary arteries is automatically registered with a cardiac-gated biplane cineangiogram at different breathing phases to build a subject-specific motion model. This motion model is used as a prior within the intraoperative registration process for motion compensation by restricting the search space to the most dominant modes of the motion profile. In this way, the interventionalist is provided visual guiding assistance during cardiac catheterization procedures under live fluoroscopy. The model-constrained registration increases the robustness and accuracy of the dynamic re-registrations, especially for weak data constraints such as low signal-to-noise ratio, the lack of contrast information in fluoroscopic images, or an intraoperative monoplane setting. This allows for reducing radiation exposure without compromising on registration accuracy. Synthetic data as well as phantom and clinical datasets have been used to validate the model-based registration in terms of accuracy, robustness and speed. We were able to significantly accelerate the intraoperative registration with an average 3-D error of less than 2mm even for monoplane settings and tracked guidewire data in the absence of contrast agent, which makes respiratory motion correction feasible during clinical procedures. Moreover, we propose a new methodology for global vessel segmentation in 2-D X-ray images combining a per-pixel probability map from a local vessel enhancement filter with local directional information. The global segmentation provides additional connectivity and geometric shape information, which allows for sophisticated postprocessing techniques in order to refine the segmentation with respect to the requirements of the particular application.