大尺度图像编辑的泊松方程并行多重网格求解算法.pdfVIP

大尺度图像编辑的泊松方程并行多重网格求解算法 杜振龙1,2 1 2 1 , 李晓丽 , 郭延文 , 沈钢纲 (南京工业大学 电子与信息工程学院，南京 210009) (南京大学 软件新技术国家重点实验室，南京 210000) 摘 要 随着获取设备的发展，大尺度、高分辨率数字图像已逐步步入现代人们的生活，大尺度图像的梯度域编辑显得更为 重要。求解大规模未知数的泊松方程是大尺度图像梯度域编辑的关键，传统多重网格算法的迭代、约束和插值操作单独进行， 内存和外存间通讯量大，算法效率低，为此本文提出了一种面向大尺度图像梯度域编辑的并行多重网格求解泊松方程的算法， 利用多重网格的迭代、约束和插值过程的内存数据访问局部性和更新相关性，构造滑动工作窗口，使迭代、约束和插值操作 并行运行，提高了多重网格算法求解泊松方程的计算效率。全景图拼接实验表明本文所提算法的运行效率高于超松弛迭代、 高斯塞德尔迭代和传统多重网格算法。 关键字 泊松方程, 并行多重网格, 大尺度图像编辑 Parallel Multigrid Approach for Solving Poisson PDE in Gigapixel Image Editing 1,2 1 2 1 Zhen-Long DU , Xiao-Li LI , Yan-Wen GUO , Kangkang SHEN 1 (College of Electronics and Information Engineering, Nanjing University of Technology, Nanjing 210009) 2(State Key Laboratory of Novel Software Technology, Nanjing University, Nanjing 210000) Abstract With the development of image acquisition technology, gigapixel images are being produced and emerged into the modern life, how to efficiently compile these gigapixel images within gradient domain is the research hot of image processing and computer graphics. To solve the Poisson PDE with large-scale unknowns is crucial to gigapixel image editing in gradient domain, traditional multigrid approach separately performs iteration, restriction and prolongation and bears heavy communication load between RAM and external memory. In the paper, a parallel multigrid approach for solving Poisson PDE is proposed, which exploits the locality and relevance of memory accessing and updating among the different stages for parallel performing the iteration, restriction and prolongation in the sweeping window. Experiments of image stitching show that the presented method has the higher efficiency than the algorithms of s