利用独立分量分析的运动模糊图像检索  

Motion blurred image retrieval based on independent component analysis algorithm

作　　者：相林[1,2] 崔荣一[1] 

机构地区：[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

出　　处：《计算机工程与应用》2010年第10期147-149,共3页Computer Engineering and Applications

基　　金：吉林省科技发展计划国际合作项目(No.20050703-1)

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。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.

关 键 词：独立分量分析 运动模糊图像 图像检索 特征提取 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]