基于log-Gabor与FT数据融合的掌纹识别算法

Palmprint recognition based on fusion of Log-Gabor and fourier transform

本文提出一种基于数据融合的掌纹识别算法。该算法首先利用log-Gabor小波提取掌纹纹理特征,同时用傅立叶变换(Fourier Transform,FT)提取掌纹频域统计特征,然后分别与库中模板进行匹配,最后对两种匹配结果进行数据融合,得到最终识别结果。该算法中的基于手掌中基准点的手掌定位方法,使得定位中的方向校正更为准确。实验结果达到较高的识别率,验证了该算法的有效性。

This paper proposes a novel algorithm using score-level fusion of two feature extraction methods. Frist, text features is extracted by log-Gabor wavelets, and globle statistics features by Fourier Transform. After matching with the templates, we get recognition result by combining the matching scores from two modalities. It also presents a more firmly preprocessing approach to determine the orientation of a rotated human palmprint image. The experimental result hits a high recognition rate and shows efficiency on UST Hand Image database.