| 基于贝叶斯数据融合的多尺度目标识别 |
| |
| 引用本文： | 汪国有,邹玉兰.基于贝叶斯数据融合的多尺度目标识别[J].华中科技大学学报(自然科学版),2003,31(11):50-52. |
| |
| 作者姓名： | 汪国有 邹玉兰 |
| |
| 作者单位： | 华中科技大学图像识别与人工智能研究所,图像信息处理与智能控制教育部重点实验室 |
| |
| 基金项目： | 武器装备预研基金资助项目 |
| |
| 摘 要： | 针对红外电厂目标识别问题，提出了基于贝叶斯数据融合的多尺度目标识别方法．该方法研究了前视成像末制导过程中图像目标尺度变化引起的视点角度和特征尺度变化规律，建立了分层次的时空特征模型，根据显著性选取目标特征，采用贝叶斯网络把不同尺度下的显著性特征进行融合，得到正确的识别结果．实验表明，该方法能将多尺度目标的不精确、不完整的特征进行融合处理，从而完成了目标的可靠识别．
|
| 关 键 词： | 目标识别 贝叶斯网络 多尺度目标 特征融合 贝叶斯数据融合 |
| 文章编号： | 1671-4512(2003)11-0050-03 |
| 修稿时间： | 2003年4月15日 |
An approach to multiscale target recognitionbased on Bayesian data fusion |
| |
| Wang Guoyou Zou Yulan Prof., Institute for Pattern Recognition & AI,Huazhong Univ. of Sci. & Tech.,Wuhan ,China..An approach to multiscale target recognitionbased on Bayesian data fusion[J].JOURNAL OF HUAZHONG UNIVERSITY OF SCIENCE AND TECHNOLOGY.NATURE SCIENCE,2003,31(11):50-52. |
| |
| Authors: | Wang Guoyou Zou Yulan Prof Institute for Pattern Recognition & AI Huazhong Univ of Sci & Tech Wuhan China |
| |
| Institution: | Wang Guoyou Zou Yulan Prof., Institute for Pattern Recognition & AI,Huazhong Univ. of Sci. & Tech.,Wuhan 430074,China. |
| |
| Abstract: | An approach to multiscale target recognition based on Bayesian data fusion was proposed and used in infrared power plant recognition. The variation of the feature scale and the viewpoint caused by the scale change of image targets during forward-looking imaging terminal guidance was studied. A hierarchical space-time feature model was developed. And target features were selected according to their salience. By utilizing the Bayesian network, the salient feature fusion of different scale was done. The accurate result was obtained. The experiments showed that the new method worked well and credible target recognition was done through the fusion of the imprecise and incomplete multiscale features. |
| |
| Keywords: | target recognition Bayesian network multiscale target feature fusion |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
