基于合作目标的精确跟踪算法及其在图像识别中的应用
|
摘要
为降低目标动态识别跟踪系统中目标识别和定位的难度,设计了一种新型的合作目标,并提出基于结构约束的合作目标精确跟踪算法.首先,从不变性、对比度、鲁棒性等方面设计合作目标,以适应不同识别距离的要求.然后,采用自适应阈值分割与连通域标记提取候选的合作目标子部件集合,并采用基于上下文的滤波方法剔除背景干扰.最后,根据合作目标的结构模型计算其在图像中的精确位置.实验结果表明,该合作目标的设计合理,识别算法具有较好的准确性和鲁棒性.
In order to reduce the difficulty of target recognition and location in dynamic target recognition and tracking system, a new type of cooperative target is designed, and a precise tracking algorithm of cooperative target based on structural constraints is proposed. Firstly, cooperative targets are designed from the perspectives of invariance, contrast and robustness to meet the requirements of different recognition distances. Then, adaptive threshold segmentation and connected domain labeling are used to extract candidate subset of cooperative targets, and context-based filtering method is used to eliminate background interference. Finally, the exact position of the cooperative object in the image is calculated according to its structural model. The experimental results show that the design of the cooperative target is reasonable, and the recognition algorithm has good accuracy and robustness.
In order to reduce the difficulty of target recognition and location in dynamic target recognition and tracking system, a new type of cooperative target is designed, and a precise tracking algorithm of cooperative target based on structural constraints is proposed. Firstly, cooperative targets are designed from the perspectives of invariance, contrast and robustness to meet the requirements of different recognition distances. Then, adaptive threshold segmentation and connected domain labeling are used to extract candidate subset of cooperative targets, and context-based filtering method is used to eliminate background interference. Finally, the exact position of the cooperative object in the image is calculated according to its structural model. The experimental results show that the design of the cooperative target is reasonable, and the recognition algorithm has good accuracy and robustness.
引文
[1] DONG Gangqi, ZHU Z H. Autonomous robotic capture of non-cooperative target by adaptive extended Kalman filter based visual servo [J]. Acta Astronaut, 2016, 122: 209-218.
[2] 吕耀宇. 空间合作目标单目视觉位姿测量技术研究 [D]. 长春: 中国科学院大学(长春光学精密机械与物理研究所), 2018.
[3] 王向军, 曹雨, 周凯. 二维合作目标的单相机空间位姿测量方法 [J]. 光学精密工程, 2017, 25(1): 274-280.
[4] 王保丰, 李广云, 陈继华, 等. 航天器交会对接中测量靶标的两种设计方法 [J]. 宇航学报, 2008, 29(1): 162-166.
[5] HAFEZ A H A, MITHUN P, ANURAG V V, et al. Reactionless visual servoing of a multi-arm space robot combined with other manipulation tasks [J]. Robot Auton Syst, 2017, 91: 1-10.
[6] FLORES-ABAD A, MA O, PHAM K, et al. A review of space robotics technologies for on-orbit servicing [J]. Prog Aerosp Sci, 2014, 68: 1-26.
[7] ZHAI Guang, ZHANG Jingrui, ZHOU Zhicheng. Coordinated target localization base on pseudo measurement for clustered space robot [J]. Chinese J Aeronaut, 2013, 26(6): 1524-1533.
[8] LIN Shanggang, GARRATT M A, LAMBERT A J. Monocular vision-based real-time target recognition and tracking for autonomously landing an UAV in a cluttered shipboard environment [J]. Auton Robot, 2017, 41(4): 881-901.
[9] NGUYEN P H, ARSALAN M, KOO J H, et al. LightDenseYOLO: a fast and accurate marker tracker for autonomous UAV landing by visible light camera sensor on drone [J]. Sensors, 2018, 18(6): 1703-1732.
[10] WEN Zhuoman, WANG Yanjie, ARJAN K, et al. On-orbit real-time robust cooperative target identification in complex background [J]. Chinese J Aeronaut, 2015, 28(5): 1451-1463.
[11] 温卓漫. 复杂场景下合作靶标的准确快速识别与定位 [D]. 长春: 中国科学院大学(长春光学精密机械与物理研究所), 2017.
[12] 孙国鹏, 郝向阳, 张振杰, 等. 多特征判断的合作目标识别方法 [J]. 系统仿真学报, 2018, 30(6): 2377-2383.
[13] YANG Haiping, LUO Jiancheng, SHEN Zhanfeng, et al. A local voting and refinement method for circle detection [J]. Optik, 2014, 125(3): 1234-1239.
[14] RUDOLF S, TOMISLAV M. Multiple circle detection based on center-based clustering [J]. Pattern Recogn Lett, 2015, 52: 9-16.
[15] OTSU N. A threshold selection method from gray-level histograms [J]. IEEE T Syst Man Cyb, 1979, 9(1): 62-66.
[2] 吕耀宇. 空间合作目标单目视觉位姿测量技术研究 [D]. 长春: 中国科学院大学(长春光学精密机械与物理研究所), 2018.
[3] 王向军, 曹雨, 周凯. 二维合作目标的单相机空间位姿测量方法 [J]. 光学精密工程, 2017, 25(1): 274-280.
[4] 王保丰, 李广云, 陈继华, 等. 航天器交会对接中测量靶标的两种设计方法 [J]. 宇航学报, 2008, 29(1): 162-166.
[5] HAFEZ A H A, MITHUN P, ANURAG V V, et al. Reactionless visual servoing of a multi-arm space robot combined with other manipulation tasks [J]. Robot Auton Syst, 2017, 91: 1-10.
[6] FLORES-ABAD A, MA O, PHAM K, et al. A review of space robotics technologies for on-orbit servicing [J]. Prog Aerosp Sci, 2014, 68: 1-26.
[7] ZHAI Guang, ZHANG Jingrui, ZHOU Zhicheng. Coordinated target localization base on pseudo measurement for clustered space robot [J]. Chinese J Aeronaut, 2013, 26(6): 1524-1533.
[8] LIN Shanggang, GARRATT M A, LAMBERT A J. Monocular vision-based real-time target recognition and tracking for autonomously landing an UAV in a cluttered shipboard environment [J]. Auton Robot, 2017, 41(4): 881-901.
[9] NGUYEN P H, ARSALAN M, KOO J H, et al. LightDenseYOLO: a fast and accurate marker tracker for autonomous UAV landing by visible light camera sensor on drone [J]. Sensors, 2018, 18(6): 1703-1732.
[10] WEN Zhuoman, WANG Yanjie, ARJAN K, et al. On-orbit real-time robust cooperative target identification in complex background [J]. Chinese J Aeronaut, 2015, 28(5): 1451-1463.
[11] 温卓漫. 复杂场景下合作靶标的准确快速识别与定位 [D]. 长春: 中国科学院大学(长春光学精密机械与物理研究所), 2017.
[12] 孙国鹏, 郝向阳, 张振杰, 等. 多特征判断的合作目标识别方法 [J]. 系统仿真学报, 2018, 30(6): 2377-2383.
[13] YANG Haiping, LUO Jiancheng, SHEN Zhanfeng, et al. A local voting and refinement method for circle detection [J]. Optik, 2014, 125(3): 1234-1239.
[14] RUDOLF S, TOMISLAV M. Multiple circle detection based on center-based clustering [J]. Pattern Recogn Lett, 2015, 52: 9-16.
[15] OTSU N. A threshold selection method from gray-level histograms [J]. IEEE T Syst Man Cyb, 1979, 9(1): 62-66.