快速不变矩算法基于CUDA的并行实现

英文篇名：CUDA-based parallel implementation of fast moment invariants algorithm
中文刊名：计算机应用
英文刊名：Journal of Computer Applications
作者：韩斌 ; 孙文赟 ; 周飞 ; 王士同
英文作者：HAN Bin1 ; 2 ; SUN Wen-yun1 ; ZHOU Fei1 ; WANG Shi-tong2(1.School of Information Engineering ; Jiangsu University of Science and Technology(Zhangjiagang Campus) ; Zhangjiagang Jiangsu 215600 ; China ; 2.School of Information Engineering ; Jiangnan University ; Wuxi Jiangsu 214036 ; China)
中文关键词：不变矩 ; 并行计算 ; 统一计算架构 ; 协同计算
英文关键词：moment invariant ; parallel computing ; Compute Unified Device Architecture(CUDA) ; cooperative computing
出版日期：2010-07-01
机构：江苏科技大学(张家港校区)信息工程学院;江南大学信息工程学院;
年：2010
期：07
出版单位：计算机应用

摘要

不变矩自提出以来被广泛应用于目标识别系统中进行特征描述,这需要能够实时计算不变矩值。虽然已经提出了许多不变矩的快速算法,但仍无法在单台PC机上实现不变矩的实时计算。分析了基于差分矩因子的不变矩快速算法的并行性,提出了一种基于统一计算架构(CUDA)的快速不变矩并行实现方法,并在NVIDIA Tesla C1060 GPU上实现。对所提出算法的计算性能与普通串行算法进行了对比分析。实验结果表明,所提出的并行计算方法极大地提高了不变矩的计算速度,可有效地用来进行实时特征提取。
Moment invariants have been used as feature descriptors in a variety of object recognition applications since it was proposed.It is necessary to compute geometric moment values in real-time rate.Despite the existence of many algorithms of fast computation of moments,it cannot be implemented for real-time computation to be run on a PC.After analyzing the parallelism of fast moment invariants algorithm based on differential of moments factor,a new parallel computing method based on CUDA(Compute Unified Device Architecture) technology was presented and implemented on NVIDIA Tesla C1060 GPU(Graphic Processing Unit) in this paper.The computation performance of the proposed method and the traditional serial algorithm was contrasted and analyzed.The experiments show that the parallel algorithm presented in the paper greatly improves the speed of the computation of moments.The new method can be effectively used in real-time feature extraction.

引文

[1]HUMK.Visual pattern recognition by moment invariants[J].IRETransactions on Information Theory,1962,8(1):179-187.
    [2]WEE C Y,PARAMESRAN R,MUKUNDAN R.Fast computationof geometric moments using a symmetric kernel[J].Pattern Recogni-tion,2008,41(7):2369-2380.
    [3]PHILIPS W.A newfast algorithm for moment computation[J].Pat-tern Recognition,1993,26(11):1619-1621.
    [4]ZAKARIA MF,VROOMENLJ,P ZSOMBOR LA,et al.Fast al-gorithm for the computation of moment invariants[J].Pattern Recog-nition,1987,20(6):639-643.
    [5]STRACHANL N J C,NESVADBA P,ALLEN A R.A method forworking out the moment of a polygon using an integration technique[J].Pattern Recognition Letters,1990,11(3):351-354.
    [6]王冰.基于差分矩因子的灰度图像矩快速算法[J].计算机学报,2005,28(8):1367-1375.
    [7]闫晓东,郑林涛,孙燮华.一种新的二维图像几何矩快速计算方法[J].计算机应用与软件,2005,22(7):100-101.
    [8]RIEGEL E,INDINGER T,ADAMS N A.Implementation of a Lat-tice-Boltzmann method for numerical fluid mechanics using thenVIDIA CUDA technology[J].Computer Science Research and De-velopment,2009,23(3):241-247.
    [9]SPURZEM R,BERCZIK P,MARCUS G.Accelerating astrophysi-cal particle simulations with programmable hardware(FPGA andGPU)[J].Computer Science Research and Development,2009,23(3/4):231-239.
    [10]陈飞国,葛蔚,李静海.复杂多相流动分子动力学模拟在GPU上的实现[J].中国科学B辑:化学,2008,38(12),1120-1128.
    [11]章浩,庞振山,姚长利,等.基于CUDA技术的矿产储量计算[J].地质通报,2009,28(2/3):216-223.
    [12]张兵,赵改善,黄骏,等.地震叠前深度偏移在CUDA平台上的实现[J].勘探地球物理进展,2008,31(6):427-432.
    [13]高智勇,万双,舒振宇,等.基于CUDA的H.264/AVC视频编码的设计与实现[J].中南民族大学学报:自然科学版,2009,28(1):67-71.