| 三维飞机目标识别的一种新方法 |
| |
| 引用本文： | 杨蕊红,潘泉,程咏梅.三维飞机目标识别的一种新方法[J].计算机仿真,2006,23(6):82-84,216. |
| |
| 作者姓名： | 杨蕊红 潘泉 程咏梅 |
| |
| 作者单位： | 西北工业大学自动化学院,陕西,西安,710072 |
| |
| 基金项目： | 国防科技计划;国家自然科学基金 |
| |
| 摘 要： | 提出了将小波不变矩用于三维飞机目标的识别中.先采用CCD成像传感器拍摄小角度变化的不同姿态的二维飞机图像,建立完整的飞机图像模型库,然后对这些图像进行归一化处理,提取这些二维图像的小波不变矩,针对小波不变矩维数较高的特点,采用离散度和顺序前进法相结合的原则来对其进行特征优化选择,选择一组较优的特征组,最后用改进的BP神经网络为分类器进行三维飞机图像的识别.仿真结果表明,同Hu矩相比,小波矩不变量有比较好的识别效果.
|
| 关 键 词： | 图像识别 小波不变矩 特征选择 神经网络 |
| 文章编号： | 1006-9348（2006）06-0082-03 |
| 收稿时间： | 2005-05-16 |
| 修稿时间： | 2005-05-16 |
A New Method for Identifying 3D Aircraft Targets |
| |
| YANG Rui-hong,PAN Quan,CHENG Yong-mei.A New Method for Identifying 3D Aircraft Targets[J].Computer Simulation,2006,23(6):82-84,216. |
| |
| Authors: | YANG Rui-hong PAN Quan CHENG Yong-mei |
| |
| Abstract: | In this paper, wavelet invariant moments are used in the recognition of 3D Aircraft Targets. Firstly the photos of 2D Aircraft Targets are taken from different views by CCD imaging sensor. Then the 2D images are normalized and the wavelet invariant moments of the 2D images are extracted. In consideration of the large feature dimensions of the wavelet invariant moments, a set of better features are chosen by combining divergence with Sequential Forward Selection. At last the 3D Aircraft Targets are classified by modified BP neural network. Simulation results show that wavelet invariant moments are superior to Hu moments. |
| |
| Keywords: | Image recognition Wavelet invariant moments Feature selection Neural network |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
