基于几何特征及C 4.5的人脸美丽分类方法

从机器学习的角度来探索人脸美,提出与中国女性美丽程度相关的17维特征提取方法,然后运用C4。5分类树对不同美丽评分的人脸图像进行训练和测试。对510幅中国女性人脸图像的实验结果表明,文中提出的人脸美丽评价方法简单可行。对于美丽与否的两类别,平均分类精度达到94。1%。而对于4种美丽等级的分类,可达到71。6%的精度。研究表明通过合适的特征及C4。5机器学习来进行人脸美丽的智能感知是可行的。

Facial Beauty Classification Based on Geometric Features and C 4.5

An automated Chinese female facial beauty classification approach is presented through the application of machine learning algorithm of C4.5. Seventeen geometric features are designed to abstractly represent each facial image. With large set of 510 Chinese female facial images, high average accuracy of 94.1% is obtained for two-level classification－beautiful or not, and the average accuracy of 4-level classification is 71.6%. The results show that the notion of beauty perceived by human can also be learned by machine through using machine learning techniques.