| 多目标分类识别级联神经网络的实时光电混合实现 |
| |
| 引用本文： | 李豫华,张以谟,刘文耀,刘锡久,董世洪. 多目标分类识别级联神经网络的实时光电混合实现[J]. 仪器仪表学报, 1999, 20(5): 513-516 |
| |
| 作者姓名： | 李豫华 张以谟 刘文耀 刘锡久 董世洪 |
| |
| 作者单位： | 天津大学现代光学仪器研究所,天津,300072 |
| |
| 摘 要： | 本文在级联神经网络结构的基础上,采用光电混合方式实现了多目标分类识别。在实验过程中对LCD进行了改制,并采用了窄带滤波片提高了LCD的对比度。实验测试结果表明:系统稳定并具有较高的正确识别率,从而为今后进一步仪器化奠定了基础
|
| 关 键 词： | 级联神经网络 多目标分类识别 LCD的对比度 光电混合实现 |
Real-time Multi-target Classification of the Cascaded Neural Network and Its Opto-electric Hybrid Implementation |
| |
| Li Yuhua,Zhang Yimo,Liu Wenyao,Liu Xijiu,Dong Shihong. Real-time Multi-target Classification of the Cascaded Neural Network and Its Opto-electric Hybrid Implementation[J]. Chinese Journal of Scientific Instrument, 1999, 20(5): 513-516 |
| |
| Authors: | Li Yuhua Zhang Yimo Liu Wenyao Liu Xijiu Dong Shihong |
| |
| Affiliation: | College of Optoelectronic and Precision Instrument Tianjin University Tianjin 300072 |
| |
| Abstract: | Based on the model of cascaded neural network, multi target classification using opto electric hybrid implement with real time input image is described. The binary distributed interconnection weight of first stage network is obtained based on cascaded model combining with several learning algorithm families. In the experiment, the structure of liquid crystal display (LCD) is refitted, and its contrast ratio is enhanced by using a narrow band filter. It is shown that the system is stable and reliable, and the total correct recognition ratio of the system is higher. |
| |
| Keywords: | Cascaded neural network Multi target classification Contrast ratio of LCD Opto electric hybrid implement |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
