Real-Time Stereo Camera Calibration Using Stereo Synchronization and Erroneous Input Image Pair Elimination

Abstract

With the increasing usage of stereo cameras in electronic gadgets, it becomes necessary to have fast, accurate, and automatic stereo calibration during production. In this paper we propose a novel method to achieve fast, accurate and automatic stereo calibration on the video feed using with chessboard pattern. Stereo calibration requires optimal stereo pair images at various orientations without delay between capturing of left and right cameras when objects are moving. In this paper we have developed a novel software based approach to capture synchronized frames from the stereo camera setup with very minimum delay between left and right camera of stereo setup. An approach to reject unmatched stereo pairs is developed based on z-score method, so that only valid optimal image pairs are used in stereo calibration. The optimal sets of stereo synchronized error free images are used for stereo camera calibration and calibration results are stored. The entire process runs in one shot real time without human intervention thus speeding up the stereo camera calibration process.