小波分解与PCA方法的掌纹特征提取方法

Palmprint recognition based on wavelet decomposition and PCA

提出了一种新的掌纹特征提取方法,其目的在于在不降低识别率的情况下,提高掌纹特征提取速度。首先将原始掌纹图像进行小波分解,获得低分辨率的掌纹图像;其次通过主成分分析(PCA)方法获得一个低维子空间,即"特征掌";最后通过将训练、测试样本在该"特征掌"上投影来提取掌纹特征。实验结果表明,所提出方法与单一PCA方法比较,在同样识别率情况下,特征提取速度明显提高。

This paper proposed a new method for feature extraction of palmprint. This method improved upon feature extraction speed of palmprint under without reducing recognition rate. First, the original palmprint images became the lower resolution images using wavelet decomposing. Second, used principal components analysis, reduced the lower resolution palmprint images dimensionality. This low dimensional feature subspace was called ＂Eigenpalms＂. At last, those samples of a training set and a testing set were projected this ＂Eigenpalms＂. The experiment result shows that much training time has been saved by using this algorithm in features extracting recognition.