Posture Estimation of a Space Object Base on Line Reconstruction
from Stereo Images
Ke Shang
a
, Xiao Sun
a
, Jinwen Tian*
a
, Delie Ming
a
a
School of Automation, Huazhong University of Science & Technology, 1037 Luoyu Road, Wuhan,
China;
ABSTRACT
This paper proposes a novel posture estimation method which is composed of two stages. The first stage is
reconstructing lines from stereo images and the second stage is estimate posture by reconstructed lines. Accuracy of line
detection is better than the point detection. So our method have better accuracy than the methods base on points.
Keywords: posture estimation, stereo image, line reconstruction
1. INTRODUCTION
Visual tracking of a space object, such as malfunctioning satellite, is currently an interesting research topic. Classical
tracking methods using feature points can be applied on relatively variable lighting conditions. Classical method is very
depend point matching[1-3]. However, under direct sunlight the highly specular nature of the surface, due to metallic
parts or the multilayer insulation wrapped around, poses strong difficulties for point tracking. Our method use line
tracking to estimate pose, because that the line feature is more robust for noise than the point feature, and the accuracy of
the line tracking is better than the point. And the 3D line content more information than 3d point. We can estimate pose
by at least one line, while at least two points are needed.
The Hough transform is a popular tool for line detection due to its robustness to noise and missing data. However, the
accuracy of Hough transform isn't enough for pose estimation. So we need a method to optimize the results. Classical
least-squares method only minimum the errors in only one coordinate and can't process vertical line. We use improved
least-squares to get accuracy position of detected line. Our method gives an analytical solutions of minimum the errors in
both coordinates and can optimize all lines include vertical line.
After line detection, we use epipolar geometry to match lines over two views. The two end-points of a segment generate
two epipolar lines in the other image. These two lines define a region, called the epipolar “beam”, which necessarily
intersects or contains the corresponding segment. In parallel stereo cameras, the vertical coordinates of one end-point
over two views are same. And the pixels over two views at the same vertical coordinates on the line have the same color.
Then, we can calculate the pose and position of the 3D line from the matched lines over two views. We also use epipolar
geometry to match 3D lines over frames. Further more, the pose and position of the target can be calculated.
2. LINE RECONSTRUCTION
Line reconstruction is base on line detection and line matching. The accuracy of reconstruction depends on the accuracy
of line detection. Whether line can be reconstructed depends on line matching. Line matching is a difficult problem for
several reasons. The end points of line segment are not reliable and there is no strong disambiguating geometric
constraint available over 2 views. [4]
2.1 Epipolar geometry
Epipolar geometry is the geometry of stereo vision. When two cameras view a 3D scene from two distinct positions,
there are a number of geometric relations between the 3D points and their projections onto the 2D images that lead to
constraints between the image points. These relations are derived based on the assumption that the cameras can be
approximated by the pinhole camera model.
Proc. of SPIE Vol. 9813, 98130M · © 2015 SPIE · CCC code: 0277-786X/15/$18
Proc. of SPIE Vol. 9813 98130M-1