欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
锥束CT成像算法的正投影计算量大，消耗时间长．为此，本文提出了一种基于GPU实现的正投影加速算法．该算法在GPU的可编程管线中，通过纹理映射方法实现了圆轨迹锥束扫描模式下正投影计算．由于投影逐片并行计算，因此该算法具有较高的计算速率，且支持全浮点运算精度．在该正投影算法的基础上，本文还给出了三种算法优化方法．这些优化方法进一步提高了算法的执行效率．通过对Shepp-logan模型的正投影计算以及应用于迭代法进行三维图像重建等实验验证了本文算法的优点．
英文摘要：
The forward projection of cone-beam CT image reconstruction algorithms requires both excessive computation and time consuming. A novel CT forward projection method based on graphics processing unit is proposed for accelerating the forward projection calculation. The method achieves the forward projection calculation in circular cone-beam scanning mode by using texture mapping techniques in GPU programmable pipeline. Because projections are calculated slice by slice in parallel, this method has the advantages of higher computational efficiency and calculation at full floating point precision. Three optimization methods are also presented for further improving the execution efficiency of the method. The advantages of this method are verified by the experiments of the projection calculation of Shepp-logan phantom and 3D image reconstruction with iterative image reconstruction algorithm.
位置：
同期刊论文项目
