摘要
为进一步提高ASD-POCS算法的图像重建速度,对ART部分的收敛因子进行改进,提高该算法的收敛速度,同时将改进的算法在GPU上进行加速以减少算法的迭代时间,并对Shepp-Logan模型进行重建,实验表明:在数据量相同且保证成像质量的前提下,基于GPU加速的改进ASD-POCS算法较基于CPU的ASD-POCS算法运行速度提高了近2倍,达到了加速的目的。
In order to improve the speed of image reconstruction of ASD - POCS algorithm, the convergence fac- tor of the ART part is improved, which can lead to the ASD - POCS algorithm converges faster. For reducing the iteration time, the improved algorithm is accelerated on GPU and reconstructing Shepp - Logan model. Under the premise of dealing with the same amount of data and ensuring the quality of image, the results indicate that the speed of improved ASD - POCS algorithm based on GPU is nearly 2 times faster than that of the ASD - POCS algorithm based on CPU, achieving the purpose of acceleration.
出处
《核电子学与探测技术》
CAS
北大核心
2016年第8期807-810,共4页
Nuclear Electronics & Detection Technology
基金
山西省回国留学人员科研项目(2014-052)资助
