计算机工程

• 图形图像处理 • 上一篇    下一篇

基于GPU的可见光与红外图像融合快速实现

(南京航空航天大学航天学院,南京 210016)   

  1. (南京航空航天大学航天学院,南京 210016)
  • 收稿日期:2012-08-13 出版日期:2013-11-15 发布日期:2013-11-13
  • 作者简介:闫钧华(1972-),女,副教授、博士,主研方向:多源信息融合,目标检测与识别;杭谊青、孙思佳,硕士研究生
  • 基金项目:
    国家自然科学基金资助项目(41101441);南京航空航天大学基本科研业务费专项科研基金资助项目(NN2012083, NS2010214, NP2011048)

Image Fusion Fast Realization of Visible Light and Infrared Image Based on GPU

(College of Astronautics, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China)   

  1. (College of Astronautics, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China)
  • Received:2012-08-13 Online:2013-11-15 Published:2013-11-13

摘要: 为利用统一计算设备架构(CUDA)强大的并行处理能力实现快速图像融合,提出一种适用于并行运算的图像融合算法,包括高斯滤波、直方图均衡、基于小波变换的图像融合。通过CUDA编程对以上算法进行实现,并将其与对应的CPU程序相比较,实验结果表明,图形处理单元(GPU)执行效率比CPU高出一个数量级,并且随着数据量的增加,GPU的加速比还会增大。

关键词: 图像融合, 图形处理单元, 统一计算设备架构, 可见光图像, 红外图像, 并行处理

Abstract: Using powerful parallel processing capability of Compute Unified Device Architecture(CUDA) to realize quick image fusion is mainly studied. The image fusion algorithms which have good effect and suit for parallel computing are researched, which are consist of Gaussian filtering, histogram equalization, image fusion algorithm based on wavelet transform, etc. The above algorithms are realized by CUDA, and are compared with corresponding CPU programs. Experimental result shows that using CUDA to realize image fusion is quicker than CPU by an order of magnitude. With the increasing of data, GPU’s speed-up ratio will also be increased.

Key words: image fusion, Graphics Processing Unit(GPU), Compute Unified Device Architecture(CUDA), visible light image, infrared image, parallel processing

中图分类号: