计算机工程与应用 ›› 2010, Vol. 46 ›› Issue (10): 147-149.DOI: 10.3778/j.issn.1002-8331.2010.10.047

• 图形、图像、模式识别 • 上一篇    下一篇

利用独立分量分析的运动模糊图像检索

相 林1,2,崔荣一1   

  1. 1.延边大学 计算机科学与技术学院 智能信息处理研究室,吉林 延吉 133002
    2.淮阴工学院 计算科学系,江苏 淮安 223001
  • 收稿日期:2008-09-23 修回日期:2008-12-08 出版日期:2010-04-01 发布日期:2010-04-01
  • 通讯作者: 相 林

Motion blurred image retrieval based on independent component analysis algorithm

XIANG Lin1,2,CUI Rong-yi1   

  1. 1.Intelligent Information Processing Lab,Department of Computer Science and Technology,Yanbian University,Yanji,Jilin 133002,China
    2.Department of Computing Science,Huaiyin Institute of Technology,Huai’an,Jiangsu 223001,China
  • Received:2008-09-23 Revised:2008-12-08 Online:2010-04-01 Published:2010-04-01
  • Contact: XIANG Lin

摘要: 研究了独立分量分析(ICA)算法在运动模糊图像检索中的应用。首先,对图片库中的图像进行ICA处理,构造由相互独立的基向量构成的子空间,将图片库中的图像及运动模糊图像分别向该空间投影,获得各自的特征。其次,利用特征向量间的余弦距离作为相似度度量标准,根据最近邻准则进行特征匹配与图像检索。最后,对人为加入高斯噪声、进行45°和90°旋转的运动模糊以及缺损图像进行了匹配检索实验。实验结果表明,利用ICA算法提取出的特征可以准确地检索出运动模糊图像的原图像,并且对噪声污染、旋转变换和图像缺损具有良好的鲁棒性。

关键词: 独立分量分析, 运动模糊图像, 图像检索, 特征提取

Abstract: The Independent Component Analysis(ICA) algorithm is applied to the motion blurred image retrieval.Firstly,ICA is performed on the images in picture database to construct a subspace spanned by independent basis vectors,and the respective features of both images in the database and motion blurred are obtained by projecting them to the basis space.Then,using the similarity criterion of cosine distance,feature matching and image retrieving are realized according to nearest neighbor rule.Finally,the matching experiments for image retrieval are carried out with motion blurred images added Gaussian noise and rotated 45°and 90°,and furthermore,partially defective image.The experimental results show that original image corresponding to motion blurred one can be found accurately according to the features extracted by ICA,and the proposed algorithm is reasonably robust in noise,rotation and defectiveness.

Key words: independent component analysis, motion blurred image, image retrieval, feature extraction

中图分类号: