基于CUDA加速的SIFT特征提取

doi:-

东北大学学报(自然科学版) ›› 2013, Vol. 34 ›› Issue (2): 200-204.DOI: -

基于CUDA加速的SIFT特征提取

王蓓蕾¹，朱志良¹，孟琭²

(1.东北大学软件学院，辽宁沈阳110819；2.东北大学信息科学与工程学院，辽宁沈阳110819)

收稿日期:2012-08-21 修回日期:2012-08-21 出版日期:2013-02-15 发布日期:2013-04-04
通讯作者: 王蓓蕾
作者简介:王蓓蕾(1979-)，女，辽宁沈阳人，东北大学博士研究生，讲师；朱志良(1962-)，男，辽宁沈阳人，东北大学教授，博士生导师.
基金资助:
国家自然科学基金资助项目(61101057).

CUDAbased Acceleration Algorithm of SIFT Feature Extraction

WANG Beilei¹， ZHU Zhiliang¹， MENG Lu²

1. School of Software， Northeastern University， Shenyang 110819， China; 2. School of Information Science ＆ Engineering， Northeastern University， Shenyang 110819， China.

Received:2012-08-21 Revised:2012-08-21 Online:2013-02-15 Published:2013-04-04
Contact: WANG Beilei
About author:-
Supported by:
-

摘要/Abstract

摘要： 提出一种基于统一计算设备架构(CUDA)加速的尺度不变特征变换(SIFT)快速计算方法，用以解决SIFT特征提取计算过程耗时过长的问题.该方法充分利用图像处理单元(GPU)在并行计算、浮点计算、内存管理等方面的优势，合理分配主机端和设备端的资源及其在SIFT特征计算中所承担的角色.实验表明，与CPU架构下的SIFT特征提取算法相比，本文算法可以大幅度加快SIFT特征提取的计算速度，其加速比随着SIFT特征点数目的增加而增加，在本文实验中最大加速比可达1954.

关键词: CUDA加速, 尺度不变特征变换, 图像特征, 特征描述符, 图像处理单元

Abstract: A novel algorithm for accelerating scaleinvariant feature transform (SIFT) was presented on the basis of compute unified device architecture (CUDA)， which could solve the timeconsuming problem in SIFT feature extracting. This algorithm took the advantages of graphic processing unit (GPU) in parallel computation， float point computation and memory management， and reasonably allocated the computational resources and the corresponding computational tasks to the host and device in the SIFT feature extracting. Experimental results show that， compared with the CPUbased acceleration algorithm， the proposed CUDAbased algorithm greatly speeds up the extracting of SIFT features. The acceleration ratio increases with the number of SIFT feature points. The maximum acceleration ratio in the experiments was 1954.

Key words: CUDA acceleration, scaleinvariant feature transform, image feature, feature descriptor, GPU

中图分类号:

TP39141

王蓓蕾，朱志良，孟琭. 基于CUDA加速的SIFT特征提取[J]. 东北大学学报(自然科学版), 2013, 34(2): 200-204.

WANG Beilei， ZHU Zhiliang， MENG Lu. CUDAbased Acceleration Algorithm of SIFT Feature Extraction[J]. Journal of Northeastern University, 2013, 34(2): 200-204.

参考文献

[1] Lowe D G.Object recognition from local scaleinvariant features［C］//The Proceedings of the seventh IEEE International Conference on Computer Vision(ICCV).Kerkyra，Greece，1999:1150/1157.
[2] Lowe D G.Distictive image features from scaleinvariant keypoints［J］.International Journal of Computer Vision，2004，60(2):91/110.
[3] 王瑞，梁华，蔡宣平.基于GPU的SIFT特征提取算法研究［J］.现代电子技术，2010 (15):41/43.(Wang Rui，Liang Hua，Cai Xuanping.Study of SIFT feature extraction algorithm based on GPU［J］.Modern Electronic Technique，2010 (15):41/43.)
[4] 年华.GPU通用计算与基于SIFT特征的图像匹配并行算法研究［D］.西安:西安电子科技大学，2010.(Nian Hua.GPU general computation and study of parallel computing of image registration based on SIFT feature［D］.Xi’an:Xidian University，2010.)
[5] 左颢睿，张启衡，徐勇，等.基于GPU的快速Sobel边缘检测算法［J］.光电工程，2009，36(1):41/43.(Zuo Haorui，Zhang Qiheng，Xu Yong，et al.Fast Sobel edge detection algorithm based on GPU［J］.OptoElectronic Engineering，2009，36(1):41/43.)
[6] Sinha S，Frahm J M，Pollefeys M，et al.Feature tracking and matching in video using programmable graphics hardware［EB/OL］.［2012/07/05］.http://www.springerlink.com/content/5rv615p24360/.
[7] Zhang Q，Chen Y R，Zhang Y M.SIFT implementation and optimization for multicore systems［C］//IEEE International Symposium on Parallel and Distributed Processing.Miami:IEEE Press，2008:1/8.
[8] 田文，徐帆，王宏远，等.基于CUDA的尺度不变特征变换快速算法［J］.计算机工程，2010，36(8):219/221.(Tian Wen，Xu Fan，Wang Hongyuan，et al.Fast scale invariant feature transform algorithm based on CUDA［J］.Computer Engineering，2010，36(8):219/221.)
[9] Mikolajczyk K，Schmid C.Scale ＆ affine invariant interest point detectors［J］.International Journal on Computer Vision，2004，60(1):63/86.

基于CUDA加速的SIFT特征提取

CUDAbased Acceleration Algorithm of SIFT Feature Extraction

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 1

编辑推荐

Metrics

本文评价