一种基于双子空间的人脸美感分析方法

Dual Subspace Algorithm for Facial Attractiveness Analysis

子空间技术是一种有效的人脸美感本征描述方法。为了克服单一子空间在人脸图像美感描述方面的不足,提出了一种基于主成分分析(PCA)与广义矩阵低秩逼近(Generalized low rank approximation matrix,GLRAM)双子空间的自动人脸美感分析方法。通过组合PCA和GLRAM子空间获取人脸美感特性的全局及局部本征描述,并利用高斯场模型(Gaussian field model,GF)构造组合子空间的内在几何结构关系。实验选用了一个光照、背景、表情、年龄和种族等变化比较显著的数据库,结果表明,提出的基于双子空间算法优于基于单一子空间的人脸美感分析方法。

Subspace technique is an efficient method for automatic facial attractiveness analysis. To enhance the intrinsic description for facial attractiveness, a dual subspace method on the subspaces of principal component analysis （PCA） and generalized low rank approximation ma- trix （GLRAM） is proposed. Thus, their individual characteristics in characterizing the global and local intrinsic description of facial attractiveness can be collaboratively boosted. Besides, the Gaussian field （GF） model is applied to reflect the geometric structure in sample space. The experiment is performed on a challenging database. It takes on significant variations in the aspects of illumination, background, facial expression, age, race, and so on. Experimental results show the advantages of the proposed dual suhspace method for facial attractiveness analy- sis over the individual subspace one.