| 基于不变矩和神经网络的目标识别方法 |
| |
| 引用本文： | 张科,罗华,王秀琴.基于不变矩和神经网络的目标识别方法[J].火力与指挥控制,2009,34(3). |
| |
| 作者姓名： | 张科 罗华 王秀琴 |
| |
| 作者单位： | 1. 西北工业大学航天学院,陕西,西安,710072
2. 西安北方光电有限公司,陕西,西安,710043 |
| |
| 基金项目： | 国家自然科学基金，航天科技创新基金，武器装备预研基金 |
| |
| 摘 要： | 针对寻的导引头成像制导中实时采集图像由于存在几何失真及像质模糊而导致目标识别精度差的问题,提出了一种基于不变矩和神经网络的目标识别方法.该方法采用一种改进的不变矩算法提取目标图像的不变矩特征量来训练BP神经网络,实现目标图像的识别和分类.仿真结果表明该不变矩特征库训练的神经网络分类器具有较高的识别精度,是一种比较实用的目标识别方法.
|
| 关 键 词： | 寻的导引头 特征提取 不变矩 目标识别 BP神经网络 |
Object Recognition Method based on Neural Network and Invariant Moments |
| |
| ZHANG Ke,LUO Hua,WANG Xiu-qin.Object Recognition Method based on Neural Network and Invariant Moments[J].Fire Control & Command Control,2009,34(3). |
| |
| Authors: | ZHANG Ke LUO Hua WANG Xiu-qin |
| |
| Institution: | 1.College of Astronautics;Northwestern Polytechnical University;Xi'an 710072;China;2.Sicong Opto-Electronic Corporation;Xi'an 710043;China |
| |
| Abstract: | Aiming at the problem of the recognition precision of real time image with geometric distortion and blur,a method of object recognition based on invariant moments and neural network is advanced.The BP net is trained by using a kind of modified algorithm of Hu's invariant moments to abstract invariants of objects to classify the targets to be recognized.The result of simulation shows that the trained BP net classifier has been improved the recognition precision,this is a practical method of object recognitio... |
| |
| Keywords: | seeker feature extraction invariant moments target recognition BP neural network |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
