摘要
为解决人脸识别过程中出现的无法有效区分多姿态人脸的问题,进一步提高人脸表情识别率。本文在分析现有人脸表情识别方法的基础上,提出新的识别技术,即采用基于径向基函数(RBF)神经网络的方法,首先对图像皮肤和非皮肤像素进行分离,把人脸区域从检测到的皮肤区域中提取出来,然后以人脸表情数据库JAFFE为测试数据库,对人脸图像进行Gabor小波变换(GWT)和离散余弦变换(DCT),最后将该算法用于径向基函数神经网络的训练过程,建立相应优化模型,并将其应用到人脸表情的识别中,研究结果表明,具有收敛速度快、识别率高等优点,比文献中的方法提高了3%和8%的识别率。
During face recognition,there exists a serious problem that it is very difficult to distinguish faces of various gestures. On the basis of the analysis of the existing facial expression recognition methods, this paper proposes a new recognition technology,skin and non-skin pixels were separated.RBF neural network is enrolled to recognize various faces.Face regions were extracted from the detected skin regions.Facial expressions are analyzed from facial mages by applying Gabor wavelet transform （GWT） and Discrete Cosine Transform （DCT） on face images.Finally applied the new algorithm to the problem of expression recognition.The results show that,this method has fast convergence speed,high recognition rate,than the methods in literatures to improve the recognition rate of 3% and 8%.
出处
《石河子大学学报（自然科学版）》
CAS
2014年第2期255-259,共5页
Journal of Shihezi University(Natural Science)
基金
国家自然科学基金项目(61262021)
教育部社科研究基金项目(11XJJAZH001)
石河子大学科学技术研究发展计划项目(2012ZRKXYQ18)
关键词
人脸识别
径向基函数
GABOR小波变换
离散余弦变换
face recognition
radial basis function
gabor wavelet transform
discrete cosine transform
