一种人脸表情分类的新方法——Manhattan距离  

New approach to facial expression classification——Manhattan distance

作　　者：李俊华[1] 彭力[1] 

机构地区：[1]江南大学通信与控制工程学院,江苏无锡214122

出　　处：《计算机工程与应用》2008年第2期74-75,79,共3页Computer Engineering and Applications

摘　　要：提出了一种利用Manhattan距离进行人脸表情分类的新方法。Manhattan距离计算出具有不同模式的两个对象的距离更大。在实验中,比较了Manhattan距离、欧氏距离、余弦距离在人脸表情分类中的性能,得出Manhattan距离比另外两类距离有着更好的识别效果。The paper is presented a new method of facial expression classification.It＇s called Manhattan distance.Manhattan distance yields a higher value for pairs of objects that are less similar to one another.The author compares Manhattan distance with Euclidean distance and COS distance in the experiment.The performance of Manhattan distance is better than other methods.

关 键 词：Manhattan距离 人脸表情分类 欧氏距离 余弦距离 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]