Research And Implementation On Acceleration Algorithm Of Ray Casting Volume Rendering Based On GPU

Visualization in scientific computing has important application value in medical image visualization,geological exploration visualization,computational fluid dynamics and other fields.It is of great significance to study the related technology of visualization of scientific computing and put forward the efficient volume rendering scheme.It is very important to perfect the visualization theory and the application field of rich volume rendering.Ray casting algorithm is one of the classical direct volume rendering algorithms.The quality of image rendering is high and has been widely applied in practice.Because of the large volume of 3D volume data,the rendering algorithm based on voxel rendering has high computation complexity,slow rendering speed and poor real-time performance.In this paper,the research of ray casting algorithm is mainly focused on improving rendering effect and achieving real-time rendering.The main contents of the work include:(1)Using octree encoded compact bounding box algorithm to optimize rendering effect.The algorithm generates sub blocks by virtual segmentation of body data and selects effective sub blocks when traversing octree,which can skip the invalid voxel(null body element),thus reduce the amount of computation and increase the speed of drawing.(2)The Blinn-Phong illumination model is improved to get better volume rendering effect.The gradient mode is used as a coefficient in the illumination model to enhance the illumination intensity at the edge of the organization,which can effectively improve the rendering effect.(3)Based on the parallel computing framework of GPU,the volume rendering process is reconstructed.The process of light entry point calculation,departure point calculation and ray traversal sampling in the CPU is moved to GPU for parallel processing,and properly balance the computing performance of vertex processors and fragment processors.The purpose of real-time rendering is achieved by using the high speed floating point calculation ability of GPU.