| 分区适应截值模型及其在交通标志识别中的应用 |
| |
| 引用本文： | 常胜江 申金媛. 分区适应截值模型及其在交通标志识别中的应用[J]. 光学学报, 1996, 16(12): 750-1756 |
| |
| 作者姓名： | 常胜江 申金媛 |
| |
| 作者单位： | 南开大学现代光学研究所 |
| |
| 基金项目： | 国家攀登计划认知科学（神经网络）重大关键项目,国家自然基金 |
| |
| 摘 要： | 针对Ｈｏｐｆｉｅｌｄ网络模型在存储模式不满足０和１状态的均匀分布及数目对等的条件下存储容量及寻址能力下降的缺点，提出并用光束方向编码光学实现了三值（１，０，－１）互连的分区适应截值模型，并把这一模型应用到交通标志的识别中，结果表明该模型及光学系统有很好的稳定性。
|
| 关 键 词： | 神经网络 互连权重 分区适应截值 光信息存储 |
| 收稿时间： | 1995-11-24 |
Locally Adaptive Clipped Model and Its Application in of Traffic Sign Recognition |
| |
| Chang Shengjiang Shen Jinyuan Zhang Yanxin. Locally Adaptive Clipped Model and Its Application in of Traffic Sign Recognition[J]. Acta Optica Sinica, 1996, 16(12): 750-1756 |
| |
| Authors: | Chang Shengjiang Shen Jinyuan Zhang Yanxin |
| |
| Abstract: | The storage capacity and addressability of Hopfield model decline greatly when stored patterns are not of independence and equal number of 1, 0 state. To remove this drawback, an neural networks of locally adaptive clipped model is proposed and realized with encoding method of beam direction. This model is also applied to the recognition of traffic sign. The experimental results show that locally adaptive clipped model and the optical system performs well. |
| |
| Keywords: | neural networks interconnection weights locally adaptive clipping orthogonal algorithm. |
| 本文献已被 CNKI 维普 等数据库收录！ |
