CAJ | 학술논문

数学形态学运算是一种高度并行的运算，其计算量大而又如此广泛地应用于对实时性要求较高的诸多重要领域。为了提高数学形态学运算的速度，提出了一种基于CUDA架构的GPU并行数学形态学运算。文章详细描述了GPU硬件架构和CUDA编程模型．并给出了GPU腐蚀并行运算的详细实现过程以及编程过程中为充分利用GPU资源所需要注意的具体问题。实验结果表明，GPU并行数学形态学运算速度可达到几个数量级的提高。
수학형태학운산시일충고도병행적운산，기계산량대이우여차엄범지응용우대실시성요구교고적제다중요영역。위료제고수학형태학운산적속도，제출료일충기우CUDA가구적GPU병행수학형태학운산。문장상세묘술료GPU경건가구화CUDA편정모형．병급출료GPU부식병행운산적상세실현과정이급편정과정중위충분이용GPU자원소수요주의적구체문제。실험결과표명，GPU병행수학형태학운산속도가체도궤개수량급적제고。
Mathematical morphology operation is a highly parallel processing, it involves a large amount of computation and is widely used for so many important filed that have high requirements on real-time. In order to improve the speed of mathematical morphological operations, a GPU Parallel mathematical morphology operations based on CUDA is proposed in this paper. We have made a description of the GPU hardware architecture and the CUDA programming model. The specific GPU parallel implementation process of erosion operation and some problems that involves make full use of GPU resources are given in this paper too. Experimental results show that, GPU parallel computing mathematical morphology is several orders of magnitude faster than normal morphology operation.