有效用于人脸识别的光照不变特征表示算法

发布时间：2018-06-15 16:52

  本文选题：光照不变特征表示 + 人脸识别　； 参考：《计算机工程与应用》2017年01期

【摘要】：在光照变化环境下,人脸识别的鲁棒性是人脸识别系统中一大挑战。针对光照变化对人脸识别的影响,对经典光照不变特征表示算法进行了研究,提出一种基于局部标准差光照不变的人脸特征表示算法及其加权形式。结合完备线性鉴别分析(Complete-Linear Discriminant Analysis,C-LDA)算法提取特征,在Extended Yale-B与YALE人脸库中,与其他处理光照变化的经典方法相比,如多尺度Retinex(Multi Scale Retinex,MSR)、韦伯脸(Weber-Face,WF)和局部归一化(Local Normalization,LN),提出的算法能获得更高识别率。
[Abstract]:The robustness of face recognition is a major challenge in face recognition systems under varying illumination conditions. Aiming at the influence of illumination variation on face recognition, the classical illumination invariant feature representation algorithm is studied, and a face feature representation algorithm based on local standard deviation illumination invariance and its weighted form is proposed. Combined with the complete linear discriminant analysis (Complete-Linear discriminant Analysis) algorithm, the features are extracted. In extended Yale-B and Yale face databases, compared with other classical methods to deal with illumination changes, For example, multiscale Retinexy Multi scale Retinexan MSRs, Weber-Weber-FFs and Local NormalizationLNs, the proposed algorithm can obtain higher recognition rate.
【作者单位】： 暨南大学信息科学技术学院;暨南大学电气信息学院;
【基金】：广东省学科建设专项资金项目-科技创新(No.2013KJCX0023) 珠海市公共技术服务平台科技项目(No.2013D0501990013)
【分类号】：TP391.41
，

本文编号：2022743