摘要
掌纹自动识别是对基于生物统计学的身份鉴别的重要补充。论文尝试了对掌纹图像代数特征的提取 ,并设计了基于 K- L变换的掌纹自动识别方法。实验表明 ,这种方法能够取得很好的识别效果 ,即使对于低分辨率下的图像也有较高的识别率。在此基础上结合子空间法模式识别理论 ,提出了双子空间掌纹自动识别方法。并用双子空间对各类样本的空间分布进行了更加精细的刻画。然后采用分层最小距离分类器实现掌纹识别 ,使识别率和鲁棒性得到了进一步提高。
Automated palmprint recognition is an important part of biometric based personal identification. This paper studied the extraction of palmprint algebraic features and designs an approach for automated palmprint recognition based on K L transform. The algorithm has been proven to be very effective in research, even if the palmprint images were low resolution. Therefore, the dual eigenspace method for automatic palmprint recognition was proposed using the subspace method of pattern recognition. The dual eigenspace method provided a more refined description for the distribution of each pattern class in the subspace. Palmprint identification is then performed with a two layer minimum distance classifier and gives better and more robust results.
出处
《清华大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2000年第9期100-103,共4页
Journal of Tsinghua University(Science and Technology)
关键词
身份鉴别
生物统计学
K-L变换
掌纹自动识别
palmprint
personal identification
biometrics
K L transform
subspace method