基于Nonsubsampled Contourlet变换的掌纹特征提取算法

A Palmprint Feature Extraction Algorithm Based on Nonsubsampled Contourlet Transform

为了研究Nonsubsampled Contourlet变换(NSCT)在掌纹识别应用方面的可行性和性能,本文采用MATLAB仿真手段,利用NSCT对纹理特有的各向异性和多尺度特点,以香港理工大学采集的掌纹图像为研究样本,建立掌纹识别的分析和研究平台,针对NSCT从掌纹图像中分解得到的多个系数矩阵,研究掌纹特征的提取算法。实验结果表明,NSCT在掌纹识别方面具有较好的性能和较高的识别率,从而验证了该方法在掌纹识别中的有效性和识别效果。

In order to study the feasibility and performance of Nonsubsampled Contourlet Transform in plam recognition applications, MATLAB simulation is adopted in this paper. With the set palmprint images in the database of the Polytechnic University of Hong Kong used as research samples, we also use the fully shift-invariant, multiseale, and muhidirection expansion of NSCT for texture, and then propose a palmprint feature extraction algrithom according to the coefficient matrices decomposed from palmprint images by NSCT. The experimental results show that NSCT not only yields better performance but also has a higher recognition rate. Furthermore, the results prove the reliability of palmprint recognition and the effect of recognition.