MATLAB与OpenACC结合实现中值滤波算法并行优化

针对当前基于CUDA的中值滤波并行加速算法存在的编程复杂,图像数据提取和显示环节实现繁琐等问题提出了利用MATLAB和OpenACC结合的并行算法。该算法采用MATLAB实现图像的提取并转换成数组文件,之后利用OpenACC实现串行代码到并行代码的转换并处理MATLAB转换的数组文件,最后通过MATLAB将滤波后的数据显示成图像。通过仿真实验进行中值滤波算法的加速实验,结果表明:提出的算法在大量减少代码转换工作量的情况下加速效果较串行程序提升明显,且与CUDA加速效果相当。

Parallel optimization of median fltering algorithm based on the combination of MATLAB and OpenACC

In this paper, a parallel algorithm combining MATLAB and OpenACC is proposed for the complex programming problems of current CUDA-based median filtering parallel acceleration algorithm and the cumbersome implementation of image data extraction and display. The algorithm uses MATLAB to extract the image and converts it into an array file. Then it uses the OpenACC language to convert the serial code to the parallel code and processes the array file converted by MATLAB. Finally, the filtered data is displayed as an image through MATLAB. The acceleration experiment of the median filtering algorithm is carried out by simulation experiments. The experimental results show that the proposed algorithm accelerates the algorithm faster than the serial program in the case of greatly reducing the amount of code conversion workload. And it is equivalent to the effect of CUDA acceleration.