摘要
针对人脸识别中小波变换后信息利用不充分和权系数选择困难的缺点,提出一种基于小波子带能量加权的人脸识别算法.首先应用二维离散小波变换(2D-DWT)对图像进行二层小波分解,然后将得到的第2层4个子带进行加权组合,并给出一种基于小波子带能量权系数求解法,无需进行人工实验选取.在此基础上,采用PCA+LDA方法进行特征提取,并在ORL人脸库中进行实验,与传统的算法相比有较快的识别速度和较高的识别率.
To tackle the problems with insufficient use of information and hardship of selecting the weights posterior to Wavelet decomposition, a new face recognition method is proposed based on wavelet weighted sub-band energy. The work starts with face images decomposed using the two-level wavelet decomposition, followed by combining the four sub-bands of the second level with a right weight, which is solved by sub-bands energy based on wavelet coefficients independent of artificial selection experiments. The feature extracting method of PCA＋LDA is adopted on the basis of the stated mechanism. Experiment using ORL face databases is conducted with results showing that the proposed approach is superior to the traditional methods in terms of the speed and recognition rate.
出处
《宁波大学学报（理工版）》
CAS
2009年第2期207-211,共5页
Journal of Ningbo University：Natural Science and Engineering Edition
基金
国家自然科学基金(10571095)
关键词
人脸识别
小波变换
线性判别分析
主成分分析
face recognition
wavelet transform
linear discriminant analysis
principal component analysis
