CAJ | 학술논문

提出了一种基于离散曲波变换和支持向量机的掌纹识别方法.首先将所有掌纹样本图像和测试图像通过基于Wrapping的快速离散曲波变换进行分解,从而获得不同尺度、不同角度的曲波变换系数;掌纹重要特征信息包含在曲波变换分解系数中的低频系数中,因此将分解系数变换形成特征向量后作为特征参数送入支持向量机中进行学习训练;最后将训练好的支持向量机用于掌纹分类.基于香港理工大学Palmprint掌纹数据库进行了大量实验,实验结果证实所提方法的识别正确率相对优于小波变换方法和其它几种经典方法.
제출료일충기우리산곡파변환화지지향량궤적장문식별방법.수선장소유장문양본도상화측시도상통과기우Wrapping적쾌속리산곡파변환진행분해,종이획득불동척도、불동각도적곡파변환계수;장문중요특정신식포함재곡파변환분해계수중적저빈계수중,인차장분해계수변환형성특정향량후작위특정삼수송입지지향량궤중진행학습훈련;최후장훈련호적지지향량궤용우장문분류.기우향항리공대학Palmprint장문수거고진행료대량실험,실험결과증실소제방법적식별정학솔상대우우소파변환방법화기타궤충경전방법.
A multiscale palmprint recognition method based on discrete curvelet transform and support vector machine was proposed. First, all palmprint images were decomposed by using discrete curvelet transform via wrapping. As a result, curvelet coefficients in different scales and various angels were obtained. The important feature information of palmprint was included in the low frequency coefficients of curvelet transform decomposition coefficients. After the decomposition coefficients were transformed into feature vectors, they were regarded as feature parameters and sent into SVM for training. Finally, SVM was used for classification of palmprint. The experiments were performed in the PolyU Palmprint database. The results indicate that the proposed method has better performance than wavelet-based method and other classical methods.