摘要
针对单对象人脸识别的特点,提出了一种基于DWT-DCT平均脸和自适应阈值的单对象人脸识别算法,该算法首先利用肤色模型和人脸几何特征检测、定位、归一化内脸,然后对人脸进行DWT变换和DCT变换提取人脸特征,并计算DWT-DCT平均脸,最后利用自适应阈值进行人脸识别。实验证明,该方法具有识别有效性和认证可靠性。
A new human face recognition algorithm based on DWT- DCT average face and self-adapting threshold is proposed. Firstly, utilizing the face color and geometry feature,the inface image unifying are separated from the pending image. Secondly,some face recognition feature are obtained in DWT and DCT domain, and the average face are computed. Finally, self-adpating threshold is utilized in recognition. The experiments show that the algorithm takes on recognition validity and authentication security.
出处
《现代电子技术》
2008年第21期89-91,共3页
Modern Electronics Technique
基金
国家自然科学基金(40571128)
国家民委重点科研项目(Mzy02004)资助
关键词
人脸识别
单对象
平均脸
自适应阈值
face recognition
single object
average face
self - adapting threshold
