Relative Magnitude of Gaussian Curvature Using Neural Network and Object Rotation of Two Degrees of Freedom
Relative Magnitude of Gaussian Curvature Using Neural Network and Object Rotation of Two Degrees of Freedom
复制标题
DOI：
--
复制
发表时间：
2007
期刊：
影响因子：
--
通讯作者：
Yi Ding;Y. Iwahori;Tsuyoshi Nakamura;Lifeng He;R. Woodham;H. Itoh
中科院分区：
文献类型：
--
作者：
Yi Ding;Y. Iwahori;Tsuyoshi Nakamura;Lifeng He;R. Woodham;H. Itoh
介绍了人脸识别系统的颜色处理方法，并提出了新的研究方向。我们证明了颜色信息有助于人脸识别的性能，并发现特别是YCbCr和YCg‘ cr ’颜色空间最适合人脸识别。本文对基于主成分分析（PCA）的人脸识别算法在RGB、HSV、YCbCr、YCg‘ cr ’等不同颜色空间下的性能进行了测试。利用彩色FERET数据库对识别率进行性能评价。在我们的实验中，研究了不同颜色域的独立pca算法对不同面部表情和年龄的鲁棒性。
This paper presents color processing for face recognition systems and proposes new directions for them. We show that color information helps performance of face recognition and found that specifically YCbCr and YCg'Cr' color spaces are the most appropriate for face recognition. In this paper, the performance of the principal component analysis (PCA)-based face recognition algorithm is performed in various color spaces including RGB, HSV, YCbCr, and YCg'Cr'. The performance evaluation was conducted with the color FERET database in terms of the recognition rate. In our experimentation, robustness of the independent PCA-based algorithms with different color domains is investigated for different facial expressions and aging.