摘要
在现有的人脸表情识别系统中,速度和识别率是最重要的两个衡量标准,为提高人脸表情判别速度和识别率,采用了一种改进了的ASM和分类树表情识别的新方法。首先对传统的ASM的特征点定位过程进行改进,主要用条带法进行局部特征点定位和使用选择性特征点提取算法来提高特征点定位的速度和准确性。用分类树识别算法来改进经典的模板匹配分类法。实验结果表明,在JAFFE人脸表情数据库中进行实验可以获得更好的识别效果。
In the existing facial expression recognition system,the speed and the recognition rate are the two most important measurements,and in order to improve the facial expression identification speed and recognition rate,a modified ASM and classification tree expression recognition method is proposed pincipally.Firstly,the traditional ASM feature point location process is improved,with strip method for local feature point location,selective feature point extraction algorithm for improving the speed and accuracy of feature point location,and classification tree algorithm for improving the classic template matching classification.The experiment in the JAFFE facial expression database indicates a much better recognition effect.
出处
《通信技术》
2012年第4期77-79,共3页
Communications Technology
基金
福建省自然科学基金(No.2010J01340)
关键词
人脸表情识别
条带
选择性特征提取
分类树识别
facial expression recognition
strip
selective feature extraction
classification tree
