现代电子技术
現代電子技術
현대전자기술
MODERN ELECTRONICS TECHNIQUE
2014年
20期
103-106
,共4页
OpenCL%异构框架%边缘检测%Prewitt算法%并行实现
OpenCL%異構框架%邊緣檢測%Prewitt算法%併行實現
OpenCL%이구광가%변연검측%Prewitt산법%병행실현
OpenCL%heterogeneous framework%edge detecetion%Prewitt algorithm%parallel implementation
Prewitt算法是数字图像分割中最常用的边缘检测算法。采用传统CPU上的串行方法实现该算法需要较大的计算量、耗时较长，因此，通过GPU对其进行性能加速有着重要的意义。然而由于GPU硬件体系结构的差异性，跨平台移植是一件非常困难的工作。针对上述问题，提出了一种基于OpenCL异构框架的Prewitt图像边缘检测并行算法。实验结果表明，该并行算法比CPU上的串行算法运行速度快，加速比可达30倍，有效地提高了大规模数据处理的效率，可移植性好，具有较高的应用价值。
Prewitt算法是數字圖像分割中最常用的邊緣檢測算法。採用傳統CPU上的串行方法實現該算法需要較大的計算量、耗時較長，因此，通過GPU對其進行性能加速有著重要的意義。然而由于GPU硬件體繫結構的差異性，跨平檯移植是一件非常睏難的工作。針對上述問題，提齣瞭一種基于OpenCL異構框架的Prewitt圖像邊緣檢測併行算法。實驗結果錶明，該併行算法比CPU上的串行算法運行速度快，加速比可達30倍，有效地提高瞭大規模數據處理的效率，可移植性好，具有較高的應用價值。
Prewitt산법시수자도상분할중최상용적변연검측산법。채용전통CPU상적천행방법실현해산법수요교대적계산량、모시교장，인차，통과GPU대기진행성능가속유착중요적의의。연이유우GPU경건체계결구적차이성，과평태이식시일건비상곤난적공작。침대상술문제，제출료일충기우OpenCL이구광가적Prewitt도상변연검측병행산법。실험결과표명，해병행산법비CPU상적천행산법운행속도쾌，가속비가체30배，유효지제고료대규모수거처리적효솔，가이식성호，구유교고적응용개치。
Prewitt algorithm is the most commonly used edge detection algorithm in digital image segmentation,but large amount of calculations and great time consumption are needed to be suffered if traditional CPU serial method is used to imple-ment the algorithm. Therefore,it is important to accelerate its performance by GPU. However,the cross platform transplantation is very difficult because of the difference of GPU hardware system structure. In view of the above questions,a parallel algorithm of Prewitt image edge detection based on OpenCL heterogeneous framework is proposed. The experimental results show that the running speed of the parallel algorithm is faster than that of the serial algorithm in CPU,and its speedup ratio is 30 times as the serial algorithm. It improved the efficiency of large-scale data processing effectively. It has good portability and high application value.
