基于四阶偏微分方程的并行图像去噪算法

Parallel Image Denoising Algorithm Based on Fourth-order Partial Differential Equations

四阶偏微分方程(PDE)图像去噪方法具有良好的去噪性能,但该类方法计算量大,耗时长.为提高算法的快速性和有效性,提出一种高效并行的四阶PDE图像去噪算法.该方法基于MPI并行环境,通过分析四阶PDE离散化后差分方程求解的并行性,对噪声图像进行条状重叠的数据划分,采用并行方式对图像去噪,极大地降低了运行时间.

Image noise removal methods based on the fourth-order Partial Differential Equations（PDE） show their good denoising performance,but they also have some problems such as the great amount of calculation and long consuming time.In order to improve the speed and efficiency of the algorithm,an efficient parallel algorithm is proposed in this paper,which can reduce the running time greatly by analyzing the parallelism of the discrete fourth-order PDE and dividing the noise image into overlapping strips in the parallel environment of MPI.