Binocular stereo vision is an important branch of machine vision, which imitates the human eye and matches the left and right images captured by the camera based on epipolar constraints. The matched disparity map can be calculated according to the camera imaging model to obtain a depth map, and then the depth map is converted to a point cloud image to obtain spatial point coordinates, thereby achieving the purpose of ranging. However, due to the influence of illumination under water, the captured images no longer meet the epipolar constraints, and the changes in imaging models make traditional calibration methods no longer applicable. Therefore, this paper proposes a new underwater real-time calibration method and a matching method based on the best search domain to improve the accuracy of underwater distance measurement using binoculars.