掌纹屈肌线纹型的量化研究

本文提出一组定量描述掌纹屈肌线纹型的特征向量,以描述和区分传统皮纹学中定义的6种掌纹屈肌线纹型。用该特征向量对151例(302只手掌掌纹)样本的6种屈肌线纹型进行了描述,并用Fisher判别分析进行了多类判别。结果发现,6种屈肌线纹型的分类判别正确率达到96.0%,其中桥贯型、悉尼型、通贯型、叉贯型的分类正确率为100%。为检验判别效果,用"刀切法"原则进行了回代交叉核实,验证后的正确率亦为96.0%,验证效果较好。结论认为,所采用的特征向量与相应的判别方法结合,能描述并区分不同的屈肌线纹型,并可得到较高分类正确率。

Discriminant Analysis of Quantitative Palm Principal Linear Patterns

In order to discuss a novel feature vector describing quantitatively human palm patterns requires distinguishing six traditional kinds of patterns.These patterns from 302 palms belonging to 151 human subjects were described by the feature vector and analyzed using the Fisher discriminant analysis method.Results indicated an accuracy rate of 96% with this method.In particular four out of six palm patterns(bridge line,simian line,sydney line,rudimentary creases) were classified correctly.The validation rate of 96% was achieved through the "Jackknife" method.In conclusion,the feature vector clearly and accurately describes six kinds of palm patterns.