LIU Mingyue, ZHU Huizhong, XU Xinchao, QIAO Haolei, FU Xiaotian, ZHAO Hanguang. Line extraction method with brightness-weighted partitions for planetary rover in-orbit calibration[J]. Geomatics and Information Science of Wuhan University. DOI: 10.13203/j.whugis20220475
Citation: LIU Mingyue, ZHU Huizhong, XU Xinchao, QIAO Haolei, FU Xiaotian, ZHAO Hanguang. Line extraction method with brightness-weighted partitions for planetary rover in-orbit calibration[J]. Geomatics and Information Science of Wuhan University. DOI: 10.13203/j.whugis20220475

Line extraction method with brightness-weighted partitions for planetary rover in-orbit calibration

  • Objectives: Aiming at the problem that the straight line extraction effect is poor due to the strong illumination condition during the In-orbit calibration, so a brightness weighted inter-partition straight line extraction method for In-orbit calibration of planetary rover is proposed. Methods: A line extraction method with brightness-weighted partitions for planetary rover in-orbit calibration was proposed. Firstly, the gradient magnitude and direction are counted, the gradient direction distribution model is obtained, and the adaptive interval division is carried out according to the standard deviation of Gaussian distribution. Then, taking the brightness of pixels as the weight basis, the gradient threshold of each interval is determined and the initial edge chain is extracted. Then, the seed points of each edge chain are determined according to the histogram of gradient direction, and the initial detection of straight lines are completed by dynamic tracking. Combining with the constraint of the final end point distance and the consistency of the center of gravity, the straight lines are connected, and the Helmholtz criterion is used to complete the straight lines verification. The solar panel navigation images of Chang'e-3 are used to carry out the experiments of line extraction, matching and In-orbit calibration under the constraint of lines Compared with the classic LSD and EDlines methods. Results: Through several line extraction methods, the line extraction results of navigation images show that the correct straight line extraction of this method is improved by 62.35% and 43.21% at the maximum and 37.96% and 27.76% on average, and the accuracy of stereo vision system calibration with line constraint is improved by 6.45%, which is 4.38% and 2.30% higher than that with other two line extraction results. Conclusions: The line extraction method with brightness-weighted partitions can assist in the in-orbit calibration of planetary rover stereo vision system with higher accuracy.
  • loading

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return