迭代重构算法是当前解决冷冻电子断层三维重构问题的有效算法,但其海量计算需求限制了其广泛应用.提出了一种冷冻电子断层三维重构的并行迭代算法,在曙光4000H高性能计算机上的实验结果表明并行迭代算法不仅能够获得较好的加速性能,而且能够有效提高重构结果的精度.
In Cryo-electron tomography (Cryo-ET), iterative algorithms are the most effective methods to obtain 3D reconstruction models. However, their huge computational costs limit their applications in practical. This paper presents a parallel iterative 3D reconstruction algorithm to solve the problem. Experimental results on Dawning 4000H cluster reveal that our algorithm can not only achieve the high speedups, but also improve the resolution of 3D reconstruction images.