计算机科学 ›› 2015, Vol. 42 ›› Issue (10): 297-300.

• 图形图像与模式识别 • 上一篇    下一篇

基于GPU的图像特征并行计算方法

张杰,柴志雷,喻津   

  1. 江南大学物联网工程学院 无锡 214122,江南大学物联网工程学院 无锡 214122,江南大学物联网工程学院 无锡 214122
  • 出版日期:2018-11-14 发布日期:2018-11-14
  • 基金资助:
    本文受国家自然科学基金资助

Parallel Computation Method of Image Features Based on GPU

ZHANG Jie, CHAI Zhi-lei and YU Jin   

  • Online:2018-11-14 Published:2018-11-14

摘要: 特征提取与描述是众多计算机视觉应用的基础。局部特征提取与描述因像素级处理产生的高维计算而导致其计算复杂、实时性差,影响了算法在实际系统中的应用。研究了局部特征提取与描述中的关键共性计算模块——图像金字塔机制及图像梯度计算。基于NVIDIA GPU/CUDA架构设计并实现了共性模块的并行计算,并通过优化全局存储、纹理存储及共享存储的访问方式进一步实现了其高效计算。实验结果表明,基于GPU的图像金字塔和图像梯度计算比CPU获得了30倍左右的加速,将实现的图像金字塔和图像梯度计算应用于HOG特征提取与描述算法,相比CPU获得了40倍左右的加速。该研究对于基于GPU实现局部特征的高速提取与描述具有现实意义。

关键词: 图像金字塔机制,图像梯度计算,GPU,CUDA

Abstract: Feature extraction and description are the foundation for many computer vision applications.Due to its high dimensional computation of pixel-wise processing,feature extraction and description are computationally intensive with poor real-time performance.Thus it is hard to be used in real-world applications.In this paper,the common computational modules used in feature extraction and description,pyramidal scheme and gradient computation were studied.The method used to compute these modules in parallel based on NVIDIA GPU/CUDA was introduced.Furthermore,computational efficiency was improved by optimizing memory accessing mechanism for global,texture and shared memory.Experimental results show that a 30x speed-up is obtained by GPU-based pyramidal scheme and gradient computation against that of CPU.By employing these GPU-based optimization techniques into HOG (Histogram of Gradient) implementation based on GPU,it obtains a 40x speed-up against that of CPU.The method proposed in this paper is of significance for implementing fast feature extraction and description based on GPU.

Key words: Image pyramidal scheme,Image gradient computation,GPU,CUDA

[1] Dalal N,Triggs B.Histogram of oriented gradients for human Detection[C]∥CVPR.2005:886-893
[2] Belongie S,Malik J,Puzicha J.Shape Matching and object recognition Using Shape Contexts[J].IEEE Trans.on Pattern Analysis and Machine Intelligence,2002,24(4):509-522
[3] Shi J,Thomasi C.Good feature to track[C]∥IEEE Conference on Computer Vision and pattern Recognition.1994:593-560
[4] Lucas B,Kanade T.An iterative image registration techniquewith an application to stereo vision[C]∥Proceedings of the International Joint Conference on Artificial Intelligence.1982:674-679
[5] Strengert M,Kraus M,Ertl T.Pyramid Methods in GPU-Based Image Processing[C]∥Proceeding of Vision,Modeling,and Visualization 2006.2006:169-176
[6] Dollár P,Appel R,Belongie S.Fast Feature pyramids for object detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2014,6(8):1532-1545
[7] Nvidia.NVIDIA CUDA A Programming Guide version 4.0[EB/OL].http://www.nvidia.com/object/cuda-cn
[8] 赵杰伊,唐敏,童若锋.基于CUDA的细分曲面阴影体算法[J].浙江大学学报(工学版),2012,46(7):1301-1306 Zhao Jie-yi,Tang Min,Tong Ruo-feng.CUDA based shadow volume algorithm for subdivision surfaces [J].Journal of Zhejiang University (Engineering Science),2012,46(7):1301-1306
[9] 唐家维,王晓峰.基于GPU的并行化Apriori算法的设计与实现[J].计算机科学,2014,1(10):238-243 Tang Jia-wei,Wang Xiao-feng.Design and Implementation of Apriori on GPU[J].Computer Science,2014,1(10):238-243
[10] Lindeberg T.Scale-space theory in computer vision[M].London:Kluwer Academic Publishers,1994
[11] 王森,杨克俭.基于双线性插值的图像缩放算法的研究与实现[J].动化技术与应用,2008,27(7):44-46 Wang Sen,Yang Ke-jian.An Image Scaling Algorithm Based on Bilnear Interpolation with VC++[J].Techniques of Automation and Applications,2008,7(7):44-46
[12] 冯煌.GPU图像处理的FFT和卷积算法及性能分析[J].计算机工程应用,2008,4(2):120-122 Feng Huang.Implementation and performance of FFT and convolution in image filtering on GPU[J].Computer Engineering and Applications,2008,4(2):120-122
[13] 吴恩华.图形处理器用于通用计算的技术、现状及其挑战[J].软件学报,2004,5(10):1493-1504 Wu En-hua.State of the Art and Future Challenge on General Purpose Computation by Graphics Processing Unit[J].Journal of Software,2004,5(10):1493-1504
[14] 马安国,成玉,唐遇星,等.GPU异构系统中的存储层次和负载均衡策略研究[J].国防科技大学学报,2009,1(5):38-43 Ma An-guo,Cheng Yu,Tang Yu-xing,et al.Research on Memory Hierarchy and Load Balance Strategy in Heterogeneous System Based on GPU[J].Journal of National University of Defence Technology,2009,1(5):38-43

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!