摘要
人脸图像的庞大信息量使其不适合于直接识别。提出用离散余弦变换处理人脸原始图像,得到降维的特征矢量,并结合前馈神经网络对人脸进行分类识别。通过对ORL人脸库多幅人脸图像的仿真实验表明,系统的识别率较高,且训练时间大大降低,是一种高效的识别方法。
A great quantity of information in the face image makes it improper to be used to direct recognition. Discrete cosine transformation is used to process raw face image, and dimension reduced feature vectors are obtained. Feedforward neural netwok is also used to classify and recognize face. Simulation experiments are conducted based on face images in ORL face database. The results show that the recognition rate is quite high and the train time is notably shortened, so the method is efficient for face recognition.
出处
《红外与激光工程》
EI
CSCD
北大核心
2003年第3期294-298,共5页
Infrared and Laser Engineering
关键词
人脸识别
离散余弦变换
多层前馈神经网络
降维
Face recognition
Discrete cosine transformtion
Multilayer feedforward neural netwok
Reduction of dimensions