Robust Cooperative Target Detection for a Vision-Based UAVs Autonomous Aerial Refueling Platform via the Contrast . . . Figure 4. Detection results for the canny operator and the visual detection method we proposed. (a) Original images from strong (left) and low (right) lighting conditions, respectively; corresponding detection results of (b) OSNL; (c) DL; (d) CSSIL; (e) MFL; (f) TSRL for the two original images; (g) detection results of Canny edge detection operator for the two original images. POSE ESTIMATION When the point matching operation has been accomplished, the binocular LHM algorithm [44] is applied to solve the position and orientation information for aerial refueling. In binocular LHM algorithm, the pose information is acquired by minimizing an error metric based on collinearity in object space. In our platform, the standard coordinate system is left camera coordinate system. The pose information of the right camera coordinate system relative to the left camera coordinate system, ½Rrl ; trl 3Â4 , 24 can be calculated through the camera calibration. Assume that the ith marker point in the world coordinate system is Pi ði ¼ 1; :::nÞ, the corresponding image coordinate of ith feature point measured by the camera is pci ¼ ½ xci yci 1 T . The relative pose between camera coordinate system and the world coordinate system is calculated by: EðR; tÞ ¼ arg min IEEE A&E SYSTEMS MAGAZINE 2 X n X ðI À V c ÞðRPi þ t À tc Þ 2 i c¼1 i¼1 (15) MARCH 2019