| 基于混合特征提取的手写体数字识别方法的研究 |
| |
| 引用本文： | 杜敏 辛大欣. 基于混合特征提取的手写体数字识别方法的研究[J]. 西安交通大学学报, 1996, 30(9): 93-98 |
| |
| 作者姓名： | 杜敏 辛大欣 |
| |
| 作者单位： | 西安交通大学 |
| |
| 摘 要： | 文章对用ＢＰ神经网络实现手写体数字的识别进行了研究，提出混合特征提取的方法，提高了特征提取的效率和识别率，对一些不能识别的样本做了分析和研究，采取在样本集中增加相近样本的方法改善神经网络对这些样本的识别效果。
|
| 关 键 词： | 神经网络 特征提取 手写体 数字识别 数字图象 |
HANDWRITTEN DIGIT RECOGNITION BASED ON MIXED FEATURE EXTRACTION METHOD |
| |
| Du Min Xin Daxin. HANDWRITTEN DIGIT RECOGNITION BASED ON MIXED FEATURE EXTRACTION METHOD[J]. Journal of Xi'an Jiaotong University, 1996, 30(9): 93-98 |
| |
| Authors: | Du Min Xin Daxin |
| |
| Abstract: | Handwritten digit recognition was studied with BP neural networks and a mixed feature extraction method. It is proved by experiment that the efficiency of feature extraction and the rate of recognition has been improved by employing a mixed feature extraction method. The ability to recognize patterns has been increased by selecting some similar patterns. |
| |
| Keywords: | neural networks feature extraction pattern recognition |
| 本文献已被 CNKI 维普 等数据库收录！ |
