基于主动形状模型的人脸脸型自动分类算法

Human face shape classification method based on active shape model

研究表明人脸脸型可以分为圆脸、椭圆脸、方脸、三角脸等。基于主动形状模型(ASM)提出了一种自动人脸脸型分类方法。首先利用各种脸形的样本进行训练以建立脸型形状模型库,然后运用ASM算法对测试样本自动定位正面人脸形状,比较其与各个脸型形状模型的距离,最后应用最近邻方法实现脸型的自动分类。仿真实验表明,该方法优于利用人脸轮廓曲率或下颌曲率的方法,能够充分挖掘人脸形状信息,分类结果稳定准确,可以有效提高大库人脸识别的速度和准确率。

To improve the recognition rate and speed on huge face database, automatic face shape classification has attracted more and more attention. Human face shape can be classified into several categories： round face, elliptic face, square face, triangle face, etc. This paper proposed a novel method to classify human face shape automatically based on Active Shape Model （ ASM）. First, face images of different shapes were trained by ASM to generate different shape models. Then ASM was used to detect the face shape of the test sample, and compared the distance with each face shape model. Finally, nearest-neighbor algorithm was employed to accomplish classification. Experimental results show that the proposed method outperforms the methods based on curvature feature of face contour or feature of chin contour. The method can detect full face shape information, and improve the recognition rate and speed on huge face database.