摘要
探讨低分辨人脸识别中如何保持高低分辨率人脸图像特征在公共特征子空间的一致性问题。基于耦合映射方法,联合利用高低分辨率人脸图像的局部流形几何结构信息与标签信息,增强耦合映射关系矩阵的判别能力和可分性,使得相同类别的高低分辨率人脸图像在公共特征子空间中的距离应尽可能接近,而不同类别的高低分辨率人脸图像之间的距离应尽可能疏远。在3个标准人脸库中对提出方法的有效性进行了验证。实验结果表明,该方法在不同的特征维度、Rank级别和分辨率下较同类方法均有明显地提升,具有较好地应用潜能。
Aiming at the key problem of how to maintain the consistency of high-and low-resolution face image features in the common feature subspace based on coupled mapping learning for low-resolution face recognition,the proposed method combines local manifold geometry structure information and label information of high-and low-resolution face images to enhance discriminant ability and separability of the coupled mapping relationship matrix such that the distances between the high-and low-resolution face images from the same class are as close as possible in the common feature subspace,while distances between high-and low-resolution face images from different classes are as large as possible.The effectiveness of the proposed method is verified on three standard face databases,and the experimental results indicate that the proposed method has significant improvement than the similar methods under different feature dimensions,rank levels,and resolutions,showing better application potential.
作者
郑冬冬
张凯兵
ZHENG Dongdong;ZHANG Kaibing(School of Electronics and Information, Xi′an Polytechnic University, Xi′an 710048, China)
出处
《西安工程大学学报》
CAS
2019年第6期666-672,共7页
Journal of Xi’an Polytechnic University
基金
国家自然科学基金面上项目(61971339)
陕西省科技厅自然科学基础研究重点项目(2018JZ6002)
西安工程大学博士科研启动基金(BS1616)
关键词
低分辨人脸识别
耦合映射
流形学习
公共特征子空间
标签信息
low-resolution face recognition
coupled mapping
manifold learning
common feature subspace
label information
