基于最大尺寸子空间的模糊人脸图像识别  

Fuzzy Face Recognition Based on Subspace of Maximum Size

作　　者：黄丽韶[1] 文海英[1] 顾思思[1] 

机构地区：[1]湖南科技学院电子与信息工程学院,湖南永州425199

出　　处：《控制工程》2016年第11期1790-1795,共6页Control Engineering of China

基　　金：湖南科技学院计算机应用技术重点学科资助;湖南科技学院一般科学研究项目(16XKY052);永州市科技局指导性科技计划项目(永科发[2015]9号;项目编号16)

摘　　要：针对传统人脸识别方法识别率不高的问题,提出一种改进的拉斯曼流形的模糊人脸图像识别算法。该方法将图像成像模型和微分几何融合在一起来研究模糊图像的识别问题,首先建立了一个由图像与预先设定的最大尺寸的完备正交基卷积得到的子空间,并证明了干净图像和模糊图像得到的子空间在零噪声和模糊基具有一定特征的假设下是相同的;然后通过将子空间当作格拉斯曼流型中的一个点分析了这种子空间表示方法来进行模糊图像的识别,同时还给出了在模糊量具有同质性和空变性条件下的模糊图像识别问题。最后通过仿真实验结果表明,提出的方法能够有效提高人脸识别的精确度,比常用的人脸识别算法具有更高的效率。For the problem that the face recognition recognition rate is not high in traditional recognition methods, the paper presents an improved flow pattern Rathmann blurred face image recognition algorithm. The method considers the image forming model together with differential geometry to study the blurred image recognition problem. Firstly, this paper establishes a complete orthogonal basis obtained by convolving the image with the presetting maximum size of the subspace, and proves that the subspace obtained by clean images and blurred images are the same under the assumption that the zero noise and fuzzy groups have certain characteristics; Secondly, the method analyzes the blurred image recognition through considering the subspace as a sub-point of Grassmannian, at the same time, the fuzzy image recognition problem is also given under the condition of homogeneity and spatial variability in the fuzzy weight. Simulation results show that this method can effectively improve the face recognition accuracy and has a higher recognition rate than the common face recognition algorithm.

关 键 词：模糊基 子空间 人脸识别 格拉斯曼流型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]