Abstract: This paper proposes a thenar palmprint classification method based on Support Vector Machine(SVM). It uses high-frequency emphasis filter to enhance the thenar palmprint image. Eight textural features which come from four directions are extracted as classification feature vectors. It compares the accuracy rate of classification in different kernel function, results show that the kernel-based SVM method which use combined feature vectors can give the best performance.

Key words: thenar palmprint, texture feature, image classification, Support Vector Machine(SVM)

摘要： 提出一种基于支持向量机(SVM)的大鱼际掌纹图像二分类法。采用高频强调滤波，对分割得到的大鱼际掌纹图像进行图像增强，提取其灰度共生矩阵4个方向的8个特征量作为分类特征向量。对比不同核函数下的分类准确率，结果表明，组合特征向量的SVM方法对大鱼际掌纹的初步二分类效果较好。

关键词: 大鱼际掌纹, 纹理特征, 图像分类, 支持向量机

CLC Number: 

• TP391