| 实时二维多尺度目标识别方法及其应用 |
| |
| 引用本文： | 鲍溪清.实时二维多尺度目标识别方法及其应用[J].大连铁道学院学报,2005,26(2):52-55. |
| |
| 作者姓名： | 鲍溪清 |
| |
| 作者单位： | 大连交通大学电气信息学院,辽宁大连116028 |
| |
| 摘 要： | 序列图像中多尺度目标具有灰度值分布一致性和不变性的特点．据此，提出基于量化灰度值自结构的动态目标识别方法．自结构的形成是通过相邻采样点之间的比较实现的．量化是对比较结果进行开关运算得到的．量化自结构之间的差值表示识别结果．识别计算过程可在0．3s～0．5s内完成，比一般的识别方法较大地提高了运算速度．本方法不受图像模糊的影响，提高了识别的可靠性和鲁棒性．应用实践表明，此方法是可行的和实用的。
|
| 关 键 词： | 序列图像 多尺度 采样 动态目标识别 |
| 文章编号： | 1000-1670（2005）02-0052-04 |
| 收稿时间： | 2005-02-22 |
Approach to 2D complex Scale Object Recognition with Real Time feature |
| |
| Bao XiQing.Approach to 2D complex Scale Object Recognition with Real Time feature[J].Journal of Dalian Railway Institute,2005,26(2):52-55. |
| |
| Authors: | Bao XiQing |
| |
| Abstract: | Selfrconstruction is formed by the comparison of neighboring image values, and the quantity of qrey distribution is obtained by on/off calculations with the recognition results indicating by the difference in unit integer self-relative structures. The recognition and calculation process trace only 0. 3 ~0. 5s to complete, and is not affected by blurry images. Thus, the reliability and performance of image recognation are improved. Appliation practice proves the method is feasible and practical. |
| |
| Keywords: | serial images complex scale sampling dynamic object recognition |
| 本文献已被 CNKI 维普 等数据库收录！ |
