作者: Hammad Dilpazir , Hasan Mahmood , Zia Muhammad , Hafiz Malik
DOI:
关键词:
摘要: A method based on multivariate mutual informa tion (MMI) is proposed for face recognition. Unlike the existing frameworks, not hindered by rigorous computation feature extraction and learning spaces. The uses information-theo retic framework training set used to estimate underlying joint marginal densities, which are utilized calculate information. information each pixel value highlight regions, that correspond maximum in formation recognition process. Performance of evaluated two image datasets. performance also compared with principal component analysis (peA) algorithms. Keywords-Multiva riate information, Face recognition, Information theory, Image processing.