气浮实验台位姿视觉测量方法

Pose visual measurement of the flotation experiment platform

为了实现气浮实验台位姿的实时确定,提出了一种基于单目视觉的位姿测量方法。首先,设计了一种具有旋转、平移、缩放不变性且易于探测的圆形合作靶标;其次,结合靶标尺寸、形状以及安装位置,基于Blob分析快速识别合作靶标,保证了定位点提取的准确性;然后,提出一种在待识别区域内进行"行、列"扫描统计的靶标圆心测量算法,实现靶标圆心的快速、鲁棒提取;最后,结合计算机坐标系下的圆心位置以及视觉测量系统中坐标变换关系,解算气浮实验台的实际位姿。实验结果表明:该方法测量位姿精度较高,抗噪能力强,在处理1 600 pixel×1 600 pixel图片的情况下,平均测量周期为53.086 ms(约19帧/s),能够实现对气浮实验台位姿实时、精确、鲁棒的测量。

In order to realize the real-time pose determination of the flotation experiment platform, a pose measurement method based on monocular vision was proposed. Firstly, an easily identifiable circular cooperative target was designed, which was invariant to the rotation, translation and scaling. Then, combining the size, shape and installation location, the cooperative target could be detected fast based on Blob analysis which could ensure the accuracy of the point extraction. Next, a method was proposed to extract the object circle centers fast and robustly by scanning the connected domains to be recognized. Finally, the actual pose of the flotation experiment platform could be calculated by the circle centers in the computer image coordinate system and the relations of coordinate transformation of the vision-based pose measurement system. The results showed that the proposed method had high accuracy and strong anti-noise capability, and the average period of processing the image of 1 600 pixel×1 600 pixel was 53.086 ms （about 19 frames per second）. The method can realize the real-time, accurate and robust measurement of the pose of the flotation experiment platform.