| 基于多神经网络分类器的军事目标识别方法研究 |
| |
| 引用本文： | 李颖,王欣威,董慧颖. 基于多神经网络分类器的军事目标识别方法研究[J]. 沈阳理工大学学报, 2005, 24(4): 18-22 |
| |
| 作者姓名： | 李颖 王欣威 董慧颖 |
| |
| 作者单位： | 沈阳理工大学,信息科学与工程学院,辽宁,沈阳,110168 |
| |
| 摘 要： | 对军事目标类型的识别是军事信息处理中的一个重要环节.本文首先对军事目标进行特征提取,以矩和圆度两种不变矩特征向量作为神经网络的输入,分别采用了BP神经网络、自组织竞争网络、Hopfield网络对军事目标进行识别,最后采用了分类器多数投票法对识别结果进行融合,仿真实验结果表明采用多神经网络分类器融合的方法比单一神经网络识别率高,这对提高军事信息处理的准确性具有重要意义.
|
| 关 键 词： | 神经网络 目标识别 多数投票法 特征提取 |
| 文章编号： | 1003-1251(2005)04-0018-05 |
| 修稿时间： | 2004-11-30 |
The Research of Military Target Recognition Method Based on Multi-Neural Network Sorter |
| |
| LI Ying,WANG Xin-wei,DONG Hui-ying. The Research of Military Target Recognition Method Based on Multi-Neural Network Sorter[J]. Transactions of Shenyang Ligong University, 2005, 24(4): 18-22 |
| |
| Authors: | LI Ying WANG Xin-wei DONG Hui-ying |
| |
| Abstract: | Military target recognition is a key step in military information processing.In this paper,the features from military targets are extracted firstly,and then the features vector formed by moment feature and roundness feature is used as the input of neural network.BP neural network,self-organization neural network and Hopfield neural network are used to recognize military target.Finally the recognition result is mixed with the majority voting method.The simulation result shows the Multi-neural network is effectual.It can reach higher identification rate than that in single neural network classification,which is significative to improve the veracity of military information processing. |
| |
| Keywords: | neural network target recognition majority voting method feature extraction |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
