THREE-DIMENSIONAL RECONSTRUCTION BASED ON IMPROVED MARCHING CUBES ALGORITHM
THREE-DIMENSIONAL RECONSTRUCTION BASED ON IMPROVED MARCHING CUBES ALGORITHM
复制标题
DOI：
10.1142/s0219519420400023
复制
发表时间：
2020-11-01
影响因子：
0.8
通讯作者：
Cui, Qi
中科院分区：
文献类型：
--
作者：
Wang, Monan;Luo, Haiyang;Cui, Qi
在标准移动立方体(MC)算法的基础上，提出了一种改进的MC算法。首先，将MC算法中原有的15种拓扑构型增加到24种，有效地避免了空洞现象的产生。为了进一步提高三维重建的速度，本文采用中点选择法代替线性插值法，将24种构型分为三种类型。每个类对应一个线程。采用多线程并行处理，提高了计算速度。利用关键区域实现多线程同步，并根据消息映射表的思想设计了协议映射表。函数指针由宏触发。处理函数由函数指针调用，完成对协议映射表的封装，保持了类的开闭原则，保证了类的可扩展性。通过对改进MC算法的精度验证和重建速度验证，得出改进MC算法能够弥补空洞问题的结论。通过在Windows和Linux两个平台上的计算时间比较，改进的MC算法的重建速度比标准MC算法快约30%，比Masala算法快40%。最后，将该算法应用于医学图像三维重建系统中，通过两组实例验证了该算法的准确性和适用性。
Based on the standard Marching Cubes (MC) algorithm, this paper proposes an improved MC algorithm. First, the original 15 topological configurations in the MC algorithm are increased to 24, which effectively avoid the generation of voids phenomenon. To further improve the speed of three-dimensional (3D) reconstruction, in this paper, the midpoint selection method is used instead of the linear interpolation method, and the 24 configurations are divided into three types. Each class corresponds to a thread. The multi-thread parallel processing is used to improve the calculation speed. The critical region is used to realize multi-thread synchronization, and then we designed a protocol mapping table according to the idea of the message mapping table. The function pointer is triggered by macro. Processing function is called by function pointer and completes the encapsulation of the protocol mapping table, which maintains the opening and closing principle of the class and ensures the scalability of the class. Through the improved MC algorithm accuracy verification and reconstruction speed verification, it is concluded that the improved MC algorithm can make up for the voids problem. By comparing the calculation time under the two platforms of Windows and Linux, the reconstruction speed of the improved MC algorithm is approximately 30% faster than the standard MC algorithm and 40% faster than the Masala algorithm. Finally, the algorithm is applied to the medical image 3D reconstruction system, and the accuracy and applicability of the algorithm are demonstrated by two sets of examples.