基于多源数据的双向闭合云控制立体视觉测量  

Bidirectional Closed Cloud Control for Stereo Vision Measurement Based on Multi-Source Data

作　　者：张贵阳[1] 霍炬[2] 杨明[1] 薛牧遥 Zhang Guiyang;Huo Ju;Yang Ming;Xue Muyao(School of Astronautics,Harbin Institute of Technology,Harbin,Heilongjiang 150001,China;School of Electrical Engineering and Automation,Harbin Institute of Technology,Harbin,Heilongjiang 150001,China;Space Propulsion Technology Research Institute,Shanghai Academy of Spaceflight Technology,Shanghai 201109,China)

机构地区：[1]哈尔滨工业大学航天学院,黑龙江哈尔滨150001 [2]哈尔滨工业大学电气工程及自动化学院,黑龙江哈尔滨150001 [3]上海航天技术研究院新力动力研究所,上海201109

出　　处：《光学学报》2020年第19期139-149,共11页Acta Optica Sinica

基　　金：装备预研航天科技联合基金、国防重点实验室科工局基金。

摘　　要：针对大视场范围下运动目标位姿参数测量易受模型累积误差、成像畸变和特征信息不足等因素的影响,提出一种新型视觉测量方法。首先建立适用于视觉测量过程中的高效多源特征数据融合模型,可以解决特征点单一的问题。然后构建基于特征点云信息的双向闭合测量模式,改变传统方法中从图像数据到空间特征信息的单向传递过程,将已确认的空间数据作为控制信息返回至测量处理过程中,可以有效规避测量空间越大导致测量模型累积误差越大的矛盾。最后实验结果表明,所提方法在10 m×8 m×3 m大视场空间内实现目标的姿态测量精度优于±1.5°,位置精度优于2 mm。所得的测量结果验证了双向闭合云控制测量模式获取的目标位姿参数精度高,稳定性强,能够满足实际的工程应用需求。Aiming at the fact that the pose parameter measurement of moving targets in a large field of view is susceptible to factors such as model cumulative error, imaging distortion and insufficient feature information, a new vision measurement method is proposed. First, an efficient multi-source feature data fusion model is established, and it suitable for the visual measurement process, which can solve the problem of single feature point. Then, a bidirectional closed measurement mode based on feature point cloud information is build, which changes the one-way transfer process from image data to spatial feature information in the traditional method, and returns the confirmed spatial data as control information to the measurement process, which can effectively avoid the larger the measurement space, the larger the cumulative error of the measurement model. Finally, the experimental results show that the proposed method achieves a target attitude measurement accuracy better than ±1.5° in a large field of view space of 10 m×8 m×3 m, and the position accuracy is better than 2 mm. The obtained measurement results verify that the target pose parameters obtained by the bidirectional closed cloud control measurement mode have high accuracy and strong stability, and can meet the actual engineering application requirements.

关 键 词：机器视觉 机器视觉光学 多源数据 双向闭合模式 大视场视觉测量 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]