Real-Time Parallel Volume Rendering Based On Hardware/Software Co-Design

Volume rendering is a significant part of 3D data visualization, is widely used in medicine, geological survey, hydrodynamics, and so on. Researchers propose several volume rendering algorithms, in which Ray Casting is more efficient, but enormous rays and sampling lead to longer time.From now on, there are mainly two implementations of volume rendering, one is software implementation, the other is hardware implementation. Software implementation is widely used in volume rendering, but enormous volume data results in worse performance. Thereafter researchers tend to graphic hardware to improve rendering performance. These special hardware rendering systems can achieve real-time requirement, and algorithm pipelines can be optimized, yet algorithm flow is fixed, the flexibility is bad, requirement of hardware configuration is great, and the price is expensive.This paper uses configurable FPGA resolution, and proposes an efficient parallel Ray Casting based on hardware/software co-design, which implements complicated but less computational jobs and Human-Computer interaction by software, and implements simple but computation-intensive jobs parallel by hardware. Thus rendering speed is improved, and it has good HC interaction and practicability. Proposal of histogram guided adaptive local transfer function can improve the quality of rendering image and enhance details of image. The proposal of algorithm framework makes implementation of portable mini diagnose system possible.