| 离线签名验证中的多分类器集成 |
| |
| 引用本文： | 刘忠平,徐一周.离线签名验证中的多分类器集成[J].计算机工程与设计,2008,29(8):2127-2130. |
| |
| 作者姓名： | 刘忠平 徐一周 |
| |
| 作者单位： | 玉林师范学院,职业技术学院,广西,玉林,537000 |
| |
| 摘 要： | 在离线签名验证的分类器设计中,为了减少特征向量分布不均和维数过高对实验结果的影响,给出一种多分类器集成的方法.根据特征向量数量级的不同进行分组,各组分类器自适应地确定分类器权重,通过投票表决得出集成判决结果.实验结果表明,通过分组和加权后,分类正确率有明显提高.
|
| 关 键 词： | 签名验证 特征分组 分类器集成 图像处理 模式识别 |
| 文章编号： | 1000-7024(2008)08-2127-04 |
| 修稿时间： | 2007年5月20日 |
Multiple classifiers integration in off-line signature verification |
| |
| LIU Zhong-ping,XU Yi-zhou.Multiple classifiers integration in off-line signature verification[J].Computer Engineering and Design,2008,29(8):2127-2130. |
| |
| Authors: | LIU Zhong-ping XU Yi-zhou |
| |
| Affiliation: | LIU Zhong-ping,XU Yi-zhou(College of Profession Technology,Yulin Normal University,Yulin 537000,China) |
| |
| Abstract: | In the classifier designing of off-line signature verification,a method based on multiple classifiers integration is given in order to reduce the influence of feature vectors distribution and great dimensions.Firstly,feature vectors are grouped by their scales.Secondly,weights of classifiers are determined automatically and adaptively.Finally,the effective integration of each classifier's result is realized by voting.The experimental result shows that the accuracy of classification can be obviously improved... |
| |
| Keywords: | signature verification feature grouping multiple classifiers integration image processing pattern recognition |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
