快速不变矩算法基于CUDA的并行实现  

CUDA-based parallel implementation of fast moment invariants algorithm

作　　者：韩斌[1,2] 孙文赟[1] 周飞[1] 王士同[2] 

机构地区：[1]江苏科技大学(张家港校区)信息工程学院,江苏张家港215600 [2]江南大学信息工程学院,江苏无锡214036

出　　处：《计算机应用》2010年第7期1983-1986,共4页journal of Computer Applications

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.

关 键 词：不变矩 并行计算 统一计算架构 协同计算 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]