| 一种基于边缘特征的改进ORB算法 |
| |
| 引用本文： | 陈艳,曾庆化,王云舒,刘建业,刘昇.一种基于边缘特征的改进ORB算法[J].导航与控制,2018(2):18-24. |
| |
| 作者姓名： | 陈艳 曾庆化 王云舒 刘建业 刘昇 |
| |
| 作者单位： | 南京航空航天大学自动化学院导航研究中心，南京 211106,南京航空航天大学自动化学院导航研究中心，南京 211106;卫星通信与导航协同创新中心，南京 211106,南京航空航天大学自动化学院导航研究中心，南京 211106;卫星通信与导航协同创新中心，南京 211106,南京航空航天大学自动化学院导航研究中心，南京 211106;卫星通信与导航协同创新中心，南京 211106,中航工业洛阳电光设备研究所，洛阳 471009 |
| |
| 基金项目： | 国家自然科学基金（编号：61533008, 61374115, 61328301, 61304227）；江苏省自然科学基金（编号：BK20141453）；中央高校基本科研业务费专项资金（编号：NS2015037） |
| |
| 摘 要： | ORB算法运算速度快，抗旋转变换能力好，但对可见光和红外图像匹配效果较差，为解决该问题，提出了一种基于边缘特征的改进ORB算法。该算法首先针对传统canny算子需要人工设定阈值的缺点，采用基于改进最大类间方差（Otsu）的自适应canny算子对图像进行边缘检测，然后对边缘图像进行ORB特征检测与描述，最后在最近次临近粗匹配的基础上进行RANSAC算法精匹配，提高了算法正确率。实验结果表明，该算法对可见光与红外图像匹配具有较好的鲁棒性，正确匹配点数和正确匹配率均比ORB算法高，能推广使用到可见光和红外图像融合等领域。
|
| 关 键 词： | 异源图像匹配 Otsu法 canny算子 ORB算法 |
An Improved ORB Algorithm Based on Edge Feature |
| |
| CHEN Yan,ZENG Qing hu,WANG Yun shu,LIU Jian ye,LIU Sheng,ZENG Qing-hu,WANG Yun-shu,LIU Jian-ye and LIU Sheng.An Improved ORB Algorithm Based on Edge Feature[J].Navigation and Control,2018(2):18-24. |
| |
| Authors: | CHEN Yan ZENG Qing hu WANG Yun shu LIU Jian ye LIU Sheng ZENG Qing-hu WANG Yun-shu LIU Jian-ye and LIU Sheng |
| |
| Institution: | Navigation Research Center, College of Automation, Nanjing University of Aeronautics and Astronautics, Nanjing 211106,Navigation Research Center, College of Automation, Nanjing University of Aeronautics and Astronautics, Nanjing 211106;Collaborative Innovation Center for Satellite Communication and Navigation, Nanjing 211106,Navigation Research Center, College of Automation, Nanjing University of Aeronautics and Astronautics, Nanjing 211106;Collaborative Innovation Center for Satellite Communication and Navigation, Nanjing 211106,Navigation Research Center, College of Automation, Nanjing University of Aeronautics and Astronautics, Nanjing 211106;Collaborative Innovation Center for Satellite Communication and Navigation, Nanjing 211106 and AVIC Luoyang Electro-optical Equipment Research Institute, Luoyang 471009 |
| |
| Abstract: | ORB has fast speed and the ability of anti-rotation transformation, but it may fail in visible and infrared image matching. So, an improved ORB algorithm based on edge feature is proposed. For the traditional canny operator needs to set the threshold artificially, the proposed algorithm firstly studies an adaptive canny operator based on the Otsu algorithm. And then detect and describe point features of the edge image with ORB. Lastly, nearest neighbor principle is for coarse matching and RANSAC method is for fine matching. The correct matching rate is improved. The experimental results show that the proposed algorithm is robust to the matching of infrared and infrared images. Also the number of correct matching points is larger and the correct matching rate is higher than ORB. It can be generalized in the fields of visible and infrared image fusion. |
| |
| Keywords: | multi-sensor image matching Otsu algorithm canny operator ORB algorithm |
|
| 点击此处可从《导航与控制》浏览原始摘要信息 |
| 点击此处可从《导航与控制》下载免费的PDF全文 |
|
