基于自适应包围盒划分的体绘制加速方法研究  

Study on accelerated volume rendering method based on adaptive bounding box division

作　　者：王睿[1] 陈春晓[1] 刘高[1] 阚星星 

机构地区：[1]南京航空航天大学生物医学工程系,南京210016

出　　处：《仪器仪表学报》2014年第11期2560-2566,共7页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(61171059)资助项目

摘　　要：针对医学影像体绘制计算量大,绘制速度慢,实时性差的问题,利用GPU可编程图形流水线完成了对算法的硬件加速,并采用空域跳过技术对基于GPU的光线投射算法进行了改进。提出了一种基于图像分割和窗技术的自适应包围盒划分策略以提高体绘制速度;对体绘制过程中产生的阶梯状伪影,通过频繁在体数据和分割掩模纹理之间切换采样或在获得真正的投射光线采样起点后便不再对分割掩模进行采样两种方法可有效去除阶梯状伪影。实验结果表明,本文提出的加速算法绘制速度可提高3倍以上。Aiming at the problems of heavy calculation burden, slow rendering speed and poor real-time capability when computing medical image large-scale volume data with ray-casting algorithm, this study utilizes GPU programmable graphics pipeline to implement the hardware acceleration of GPU-based ray-casting volume rendering. The empty-skipping technique is adopted to improve the GPU-based ray-casting algorithm; adaptive bounding box division technique based on image segmentation and window level is proposed to further improve the rendering speed by dividing the volume data bounding box into several blocks. The step-like artifact produced in volume rendering process can be effectively removed with two methods. One is frequently sampling between the volume data and segmented mask texture, and the other is no longer sampling the segmented mask when the sampling start point of ray casting is obtained. Experiment results show that the rendering speed of the proposed algorithm can be increased by more than three times.

关 键 词：医学图像 体绘制 光线投射算法 GPU 空域跳过 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TH789[自动化与计算机技术—计算机科学与技术]