摘要:
由于对光照、姿态变化的不敏感, 三维人脸识别算法已经受到人们的极大关注, 其中三维人脸特征的表示、获取以及多种表示特征的有效融合仍然是三维人脸识别的核心问题. 本文提出一种三维人脸识别方法, 该方法针对归一化的三维人脸数据, 选取人脸的曲面特征和描述人脸特征相互关系矩阵的主分量特征作为人脸表示特征, 给出了各特征的提取方法及同类特征的相似性度量, 进而提出了一种对各类特征进行加权融合的方法, 即通过分析不同特征的分类识别能力, 根据Fisher的线性判别准则, 以类内和类间特征相似度的均值差与类内和类间的散度平方和之比的大小作为该类特征权重, 在决策层为不同的特征赋予不同的权重. 最后, 基于公开发布的BJUT-3D三维人脸数据库进行了识别性能实验. 实验结果证明, 本文的特征融合方法比一般的加权策略有更好的识别性能．
Abstract:
The 3D face recognition attracts more and more attention because of its insensitivity to the variance of illumination and pose. There are many crucial problems to be solved in this topic, such as 3D face representation and effective multi-feature fusion. In this paper, a novel 3D face recognition algorithm is proposed and its performance is demonstrated on BJUT-3D face database. This algorithm chooses face surface property and the principle component of relative relation matrix as the face representation features. The similarity metric measure for each feature is defined. A feature fusion strategy is proposed. It is a linear weighted strategy based on Fisher linear discriminant analysis. Finally, the presented algorithm is tested on the BJUT-3D face database. It is concluded that the performance of the algorithm and fusion strategy is satisfying.
点击查看大图
下载:
