论文标题
分布式并行图像信号外推框架使用消息传递接口
Distributed Parallel Image Signal Extrapolation Framework using Message Passing Interface
论文作者
论文摘要
本文引入了分布式并行图像信号外推的框架。由于高质量的图像信号处理通常会带来高计算复杂性，因此需要并行执行。所提出的框架允许应用现有的图像信号外推算法，而无需修改它们以进行并行处理。现有算法的不变应用是通过将输入映像分为重叠图块来实现的，这些图像通过消息传递接口分布到计算节点。为了保持计算开销低点，提出了一种新型图像瓷砖算法。使用此算法，在很小的处理时间可能会有几乎最佳的平铺。为了显示框架的功效，它用于平行化高复杂性外推算法。仿真结果表明，所提出的框架对外推质质量没有负面影响，同时在计算簇上提供良好的缩放行为。
This paper introduces a framework for distributed parallel image signal extrapolation. Since high-quality image signal processing often comes along with a high computational complexity, a parallel execution is desirable. The proposed framework allows for the application of existing image signal extrapolation algorithms without the need to modify them for a parallel processing. The unaltered application of existing algorithms is achieved by dividing input images into overlapping tiles which are distributed to compute nodes via Message Passing Interface. In order to keep the computational overhead low, a novel image tiling algorithm is proposed. Using this algorithm, a nearly optimum tiling is possible at a very small processing time. For showing the efficacy of the framework, it is used for parallelizing a high-complexity extrapolation algorithm. Simulation results show that the proposed framework has no negative impact on extrapolation quality while at the same time offering good scaling behavior on compute clusters.
