Self-Calibration and Optimization of Binocular Camera Extrinsic Parameters

To the calibration of the extrinsic parameters for binocular camera in the unknown environment, we propose a novel method to estimate the rotation matrix and the translation vector based on the essential matrix. By extracting SIFT feature point pairs from stereo images which taken by binocular camera, and matching the feature point pairs with the FLANN algorithm, the essential matrix can be estimated, and the extrinsic parameters can be initially derived. On this basis, the objective function is constructed by adopting the physical distance between two spatial points of the calibrator whose size is known in 3D space, and the rotation matrix is further optimized. The experimental results present that the algorithm has a high accuracy of depth estimation, and the error is smaller than the un-optimized algorithm. This method can be used as a feasible scheme for the estimation of extrinsic parameters of binocular camera.