Abstract: The binocular cloud height observation system has a long baseline, so it is difficult to calibrate the external parameters of the system. An external parameter calibration method for binocular camera used in ground-based cloud height measurement system was proposed. The oriented FAST and rotated BRIEF (ORB) algorithm was used to extracte the left and right cloud image points with the same name for matching, and then calculate the rotation and translation relationship between two cameras according to the antipode geometric constraints. In order to verify the accuracy of the algorithm, a binocular cloud height system with a baseline of 60 m was set up outdoors. The odometer method was used to calibrate the external parameters, and the distance error between two cameras was 34.44 cm, the accuracy was 99.43%. Finally, by analyzing the corrected binocular cloud image and sky parallax image, the higher accuracy of the external parameter calibration method of the large baseline odometer was verified.