一种应用于人耳识别的基于SVD和PCA的特征融合方法

Ear Recognition Based on Combination of SVD and PCA

采用一种融合奇异值主元投影特征与主元投影特征的特征提取算法对静态人耳图像进行识别。该算法一方面提高了人耳的识别性能,另一方面弥补了采用单一PCA和SVD算法提取人耳特征时的不足,减少了对噪声和光照条件的敏感性。在自建人耳库和CP人耳库中的实验表明算法的合理性和有效性,为实际中处理人耳识别问题提供了参考。

A method of ear recognition was brought forward that base the fusion of principal component analysis （PCA） and singular value decomposition （SVD）. The experiment results on CP and NCUEL ear database indicate that, the method not only can greatly improve the accuracy in recognition of image ear, but also has a perfect identification effect and extent. In addition, it also has a good robustness.