基于灰度分布特征的层次掌纹识别方法  

A Hierarchical Palmprint Identification Method Based on Grayscale Distribution

作　　者：吴介[1] 裘正定[1] 孙冬梅[1] 

机构地区：[1]北京交通大学信息所,北京100044

出　　处：《电子与信息学报》2007年第8期1821-1825,共5页Journal of Electronics & Information Technology

摘　　要：目前掌纹识别算法主要集中在对掌纹图像所切取的ROI区域的研究,而对原始手部图像的灰度分布特征则讨论较少。在一定光照条件下,掌部不同位置的主线、皱纹和表层皮肤颜色的深浅在灰度图像上各自对应了不同的灰度级。该文提出一种利用手部尺寸和角度信息完成粗分类,借助单元信息熵的概念来分析手部图像的灰度分布特征从而完成细分类的层次掌纹识别方法。该方法不同于传统的对ROI区域进行特征提取的方法,直接利用整幅人手图像完成分类识别。在99类共990幅手部图像的数据库上进行的实验结果与PCA和LDA算法的对比表明,该算法具有比传统算法更高的鲁棒性,识别率也得到了较大幅度的提高。The research of previous palmprint identification algorithms are mainly focus on ROI districts cropped from the central part of palmprint images, but always ignore the color features like grayscale distribution. Under regular illumination, the texture and skin color of different position within a palmprint image will lead to differentiable grayscale distribution. In this paper, a novel hierarchical method of palmprint identification is presented, which extracts hand geometry and angle values as the coarse-level features, and calculates the unit information entropy of each subimage to describe the image＇s grayscale distribution as the fine-level feature. Distinctive to other identification methods existed, the proposed method do not need to extract ROI districts but utilize the skin colors distinction caused by locations of principle lines, wrinkles and minutias. The experimental results on the database containing 990 images from 99 individuals show the effectiveness and robustness of the proposed method compared with the traditional method PCA and LDA.

关 键 词：掌纹识别 手部尺度 灰度分布 单元信息熵 层次 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]