基于GPU的图像处理算法研究

On Algorithms of Image Processing Based on GPU

对基于统一计算设备架构(CUDA)的图形处理器(GPU)在图形处理方面的算法进行了研究和实现.针对目前图像处理算法日益复杂,性能要求越来越高,而传统的基于CPU的图像处理算法无法满足需求的情况,充分利用GPU突出的并行处理能力,采用CUDA技术,利用C++语言实现了图像处理算法.研究并设计了高斯模糊处理算法、彩色负片处理算法、透明合并处理算法的GPU并行运算流程,与CPU的性能对比表明基于GPU图像处理算法的效率更高.

This paper is related to the study and implementation of Compute Unified Device Architecture （CUDA）-based graphics processor （GPU） graphics processing algorithm. For image processing algorithms are increasingly complex, increasingly high performance requirements, while traditional CPU-based image processing methods cannot meet demand, full use of GPU parallel processing capability prominent, with CUDA, using C＋＋ language to implement image processing algorithm. To study and design of Gaussian blur processing algorithms, color negative processing algorithms, combined processing algorithms GPU parallel computing process, and through the CPU to achieve the same performance comparison of the treat- ment effect, Description the efficiency of the GPU-based image processing algorithms.