摘要
将三维重建中捆集调整算法用于优化重建结果,是非常关键的步骤,然而传统单核串行算法耗时量大不太适合大场景重建.对此,首先对捆集调整算法本身进行了改进;然后在此基础上提出了多核并行捆集调整算法并采用图像处理器(GPU)实现该算法.实验表明,所提出的多核并行捆集调整算法提高了算法优化参数的精度和处理速度.
The bundle adjustment algorithm is used to optimize the final reconstruction results of 3D scene reconstruction,which is a very critical step in 3D scene reconstruction.However,the serial bundle adjustment algorithm is large timeconsuming and not suitable for large-scale scene reconstruction.Therefore,the bundle adjustment algorithm is improved.Then the multi-core parallel bundle adjustment algorithm is proposed on the basis of the improved algorithm,and the graphic processing unit（GPU） is used to realize the algorithm.The experiment shows that the accuracy and the processing speed of the algorithm are greatly improved by using the proposed method.
出处
《控制与决策》
EI
CSCD
北大核心
2013年第9期1403-1408,共6页
Control and Decision
基金
国家自然科学基金项目(61175031)
教育部基础科研基金项目(N110204004)
关键词
三维重建
捆集调整
并行运算
图像处理器
3D scene reconstruction
bundle adjustment
parallel computation
graphic processing unit（GPU）
