杭州电子科技大学学报
杭州電子科技大學學報
항주전자과기대학학보
JOURNAL OF HANGZHOU DIANZI UNIVERSITY
2015年
3期
73-76
,共4页
冠字码%小波变换%能量差%特征提取与选择
冠字碼%小波變換%能量差%特徵提取與選擇
관자마%소파변환%능량차%특정제취여선택
serial number%wavelet transform%the method based on energy D-value%feature extraction and selection
针对第5套人民币冠字码，采用TMR传感器获取冠字码磁信号，经小波变换对信号降噪，采用能量差法截取冠字码有效磁信号，对其提取多个时域特征，构建特征判据样本库，利用BP神经网络识别冠字码磁信号，实现真假币分类。实验结果表明，所采集的不同面额纸币的冠字码磁信号稳定饱和，小波变换对磁信号降噪效果良好，能量差法可以获取完整、有效的冠字码磁信号，BP神经网络算法运算速度快、识别率达100％。
針對第5套人民幣冠字碼，採用TMR傳感器穫取冠字碼磁信號，經小波變換對信號降譟，採用能量差法截取冠字碼有效磁信號，對其提取多箇時域特徵，構建特徵判據樣本庫，利用BP神經網絡識彆冠字碼磁信號，實現真假幣分類。實驗結果錶明，所採集的不同麵額紙幣的冠字碼磁信號穩定飽和，小波變換對磁信號降譟效果良好，能量差法可以穫取完整、有效的冠字碼磁信號，BP神經網絡算法運算速度快、識彆率達100％。
침대제5투인민폐관자마，채용TMR전감기획취관자마자신호，경소파변환대신호강조，채용능량차법절취관자마유효자신호，대기제취다개시역특정，구건특정판거양본고，이용BP신경망락식별관자마자신호，실현진가폐분류。실험결과표명，소채집적불동면액지폐적관자마자신호은정포화，소파변환대자신호강조효과량호，능량차법가이획취완정、유효적관자마자신호，BP신경망락산법운산속도쾌、식별솔체100％。
According to the fifth set RMB serial numbers, tunnel magneto resistance ( TMR) is used to acquire the magnetic signal of serial numbers , the approach of wavelet transform is adopted to reduce the signal noise , the method based on energy D-value is used to cut out the effective signal , to extract several time domain features and build characteristic criterion sample library , BP neural network is used to recognize magnetic signal of serial numbers and classify counterfeit and banknote .The study shows that the collected character magnetic signal of different RMB is stable and saturate , the method of wavelet transform reduces the signal noise well , complete and effective magnetic signal can be acquired by using the method based on energy D-value , computing speed of BP neural network algorithm is fast and recognition currency is 100%.
