Binocular Computer Vision Pose Measurement For Non-cooperative Target With Circular Feature

With the development of space technology, relative pose measurement between two non-cooperative targets has been a focus. It’s a key technology in the field of capture of non-cooperative target. This article proposes a detection algorithm for a space circle and a pose measurement scheme for non-cooperative target with circular feature. Latter it carried out a series of experimental studies. The main contents are as follows:Through an investigation for research status at home and abroad of the capture technology in space, it has been clear that pose measurement algorithm for non-cooperative target is of significance. The paper determines the using ellipse detection algorithm by summarizing the existing ellipse detection algorithms. Because binocular vision measurement system is involved, this paper introduces the basic theory of the binocular vision. Then it mainly focuses on Zhang Zhengyou’s calibration method, and use the MATLAB toolbox to calibrate this visual system adopted in the experiment.Due to the need of pose measurement algorithm for non-cooperative target, the basic principle of existing RED algorithm is studied. Then this paper analyzes RED’s defects- invalid sampling and excessive reliance on parameters. In order to overcome these defects, this paper proposes an improved RED algorithm used in the detection of non-cooperative space circle. First, this method uses Canny operator to detect edges, and extracts and classifies these edges in the form of Freeman chain code. Then it uses the original RED algorithm to detect spatial circle of non-cooperative target. At last it introduces the elliptic iterative fitting process to determine the final ellipse parameters. Simulation results show the improved algorithm overcomes the two defects of the traditional RED algorithm, and improves the processing speed, accuracy and success rate.Based on the circular surface geometry characteristics of non-cooperative target, this paper proposes a pose measurement algorithm for non-cooperative target by a binocular vision measurement system. Now show the process of this method. First of all, gather the images of non-cooperative target through the binocular vision system, remove the two images’ noise. Then use improved RED algorithm to detect the space circle’s contour, get the center coordinates and circular surface normal vector in new camera coordinate system. Finally, convert them to the unified right camera coordinate system and match them, thus get the position information and the unknown circle radius of non-cooperative target. The experimental results show that this method is validity in the subject matter of the static non-cooperative target with circular feature.