摘要
针对单样本人脸识别率低的问题,提出基于人脸单样本图像生成相应的多姿态虚拟人脸样本图像的方法,并利用DCT和正交投影相结合的方法进行更快速的识别。实验表明,应用线性物体类理论和改进的特征脸方法相结合生成多姿态人脸图像,并在此基础上进行人脸识别比使用单样本和一次DCT进行识别,识别率得到很大的提高。
Because the face recognition rate based on single sample is very low＇, a method generating corresponding virtual multi-view face images based on single face sample is proposed, and quicker face recognition is achieved with the method combining DCT and orthogonal pro- jection. Experiment shows that, to combine linear object class with improved eigenfaces to generate multi-view face images, and to carry out face recognition based on this, it has much higher recognition rate compared with the methods using single sample and using DCT once only.
出处
《计算机应用与软件》
CSCD
北大核心
2013年第1期67-70,87,共5页
Computer Applications and Software
基金
国家自然科学基金项目(61070245)
