CAJ | 학술논문

在完整分割序列号的基础上,将所有字符图像分为训练数据和测试数据,以字符图像本身作为特征,利用支撑矢量机进行训练和识别。利用分类器对两种不同的策略进行了实验（即字母和数字混合识别;字母和数字分别识别）,结果表明,后者不但在时间上有所缩短,而且识别率更高,是更为可取的方法。
재완정분할서렬호적기출상,장소유자부도상분위훈련수거화측시수거,이자부도상본신작위특정,이용지탱시량궤진행훈련화식별。이용분류기대량충불동적책략진행료실험（즉자모화수자혼합식별;자모화수자분별식별）,결과표명,후자불단재시간상유소축단,이차식별솔경고,시경위가취적방법。
Based on the segmentation of RMB banknote number,the recognition algorithm of the serial number is discussed in the paper.All the character images of the paper money are devided into two groups,the training set and the testing set.SVM is applied to recognize the banknote number and all pixels of images are used as characteristic vector.In the experiment,the banknote numbers are recognized in two ways（The letters and numbers are recognized in mixed or separated ways）.The results show that the later is more advisable than the former.The recognition of denomination numbers are also discussed in the dissertation.