摘要： 提出一种基于皮肤纹理特征的高分辨人脸图像识别算法。通过人脸轮廓提取人脸的皮肤纹理特征点，利用Gabor滤波器提取人脸关键区域的皮肤纹理特征，根据纹理区域相关性进行特征匹配。在FRGC v2.0数据库上的实验结果表明，当认假率为0.1%时，该算法的认证率可达97.8%，且使用高分辨率人脸图像可提高人脸识别的性能。

关键词: 皮肤纹理, 高分辨率, Gabor特征, 区域相关性, 人脸识别

Abstract: This paper proposes a novel high resolution face recognition method based on skin texture feature. In this method, texture features are extracted based on facial contour, and Gabor wavelets are exploited to extractive skin texture features. It introduces a feature matching algorithm based on texture regional relevance. The proposed method is evaluated on the experiment of FRGC v2.0 and obtains 97.8% verification rate at False Accept Rate(FAR) is 0.1%, which is comparable to the best known results. It shows the face recognition performance can be significantly increased with the use of high resolution image.

Key words: skin texture, high resolution, Gabor feature, regional relativity, face recognition

中图分类号: 

• TP18