一种融合小波变化和张量PCA的人耳识别的算法*

小波变换具有良好的时频分析特性，而且具有较快的算法特点，同时还能起到降维的作用。张量主成分分析方法用于人耳识别能获得比PCA方法更高的识别率。综合利用这两个算法的优点，提出了一种新的人耳识别方法，对人耳图像先采用小波变换做预处理得到4个子带图像，然后对每个子带图像用张量PCA进行特征提取，最后利用最近邻的方法实现人耳图像识别。实验结果表明，利用此方法与只用主成分分析识别相比，提高了识别率。

Recognition algorithm based on wavelet preprocessing and tensor PCA

Wavelet transform has good time-frequency analysis features, but also has characteristics of fast algorithms, while playing the role of dimension reduction. Tensor principal component analysis (PCA) method is used to identify the human ear than the PCA method can get higher recognition rate. Comprehensive utilization of the advantages of these two algorithms, proposed a new recognition algorithm of the human ear, to the human ear images using wavelet transform to do the first pre-processing to be four sub-band image, and then for each sub-band images using tensor PCA feature extraction to achieve efficient human ear image recognition. Simulation results show that using this method and only the principal component analysis identified tensor compared to improve the recognition rate.