基于OpenCL的图像灰度化并行算法研究

随着图像数据量的增加,传统单核处理器或多处理器结构的计算方式已无法满足图像灰度化实时处理需求.该文利用图像处理器(GPU)在异构并行计算的优势,提出了基于开放式计算语言(OpenCL)的图像灰度化并行算法.通过分析加权平均图像灰度化数据处理的并行性,对任务进行了层次化分解,设计了2级并行的并行算法并映射到"CPU+GPU"异构计算平台上.实验结果显示:图像灰度化并行算法在OpenCL架构下NVIDIA GPU计算平台上相比串行算法、多核CPU并行算法和CUDA并行算法的性能分别获得了27.04倍、4.96倍和1.21倍的加速比.该文提出的并行优化方法的有效性和性能可移植性得到了验证.

The Study on Image Gray-Scale Parallel Algorithm Based on OpenCL

With the increase of image data amount,the computing model of the traditional single-core processor or multi-processor structure can't meet the real-time processing requirements of image gray-scale.In this paper,the parallel algorithm of image gray-scale based on Open Computing Language(OpenCL)is proposed by using the advantages of Graphic Processing Unit(GPU)in heterogeneous parallel computing.By analyzing the parallelism of weighted average image gray-scale algorithm data processing,the task is decomposed hierarchically.Two levels parallel algorithm is designed and mapped onto the CPU+GPU heterogeneous computing platform.The experimental results show that compared with the performance of the serial algorithm,multi-core CPU parallel algorithm and parallel algorithm based on Compute Unified Device Architecture(CUDA),the image gray-scale parallel algorithm obtains 27.04 times,4.96 times and 1.21 times speedup in the NVIDIA GPU computing platform under the OpenCL architecture respectively.The validity and performance portability of the proposed parallel optimization method are verified.