Adaptive fused visualization for large-scale blood flow dataset with particle-based rendering

Abstract

The visualization for the large-scale volume dataset obtained from blood flow simulations needs to be performed to help researchers to confirm the detailed positional information of the red blood cells, platelets and vascular walls. However, such a large-scale dataset is difficult to be visualized with an interactive frame rate on a normal computer. Nevertheless, the fusion of different objects, such as red blood cell volumes, platelet volumes and vascular wall surfaces, is also needed in the visualization, which makes it even more challenging. To solve this problem, we propose a system for visualizing such data by applying the particle-based rendering (PBR) (Sakamoto et al. Comput Graph 34(1):34–42, 2010); (Sakamoto and Koyamada Ultra Vis 176–185, 2012) method. This rendering method does not require any visibility sorting; thus, it can handle large-scale volume dataset, and the fusion of different objects is also easy to be implemented. To achieve an interactive frame rate analysis while maintaining good image quality, we combine two types of PBR in the system: object-space PBR (O-PBR) (Sakamoto et al. Comput Graph 34(1):34–42, 2010) and image-space PBR (I-PBR) (Sakamoto and Koyamada Ultra Vis 176–185, 2012). O-PBR method can handle large-scale data at high rendering speed but the image quality is not adequate when the view is enlarged. I-PBR method can provide high-quality rendering but have a limitation for the renderable data size. To obtain a best performance for blood flow visualization, our system adaptively switches O-PBR and I-PBR based on the data size within the view frustum and the computer resources. With this proposed adaptive visualization approach, the large-scale blood flow dataset can be efficiently visualized with the fusion of different objects while maintaining an interactive frame rate and good image quality.