Stereo Calibration of Heterogeneous RGB-ToF Camera for Robotics Applications

RGB-Depth (or RGB-D) cameras are increasingly being used in robotic and vision systems, from service robotics to industrial robotics applications. To effectively use RGB-D cameras in such applications, accurate intrinsic and extrinsic calibration parameters are essential. These parameters must be precise to achieve reliable and accurate results. Stereo calibration of two identical cameras is relatively straightforward compared to heterogeneous cameras (e.g., ToF-RGB) and has been studied by various researchers. This research article presents a methodology for the stereo calibration of two heterogeneous cameras with differing image resolutions and pixel sizes. It proposes a method to make pixel sizes and resolutions of two cameras identical to ensure accurate calibration. The infrared (IR) image captured by the ToF camera often suffers from low brightness and contrast, which can degrade the quality of captured images. As a result, this can lead to inaccurate calibration parameters during the calibration process. To address this, the IR images are enhanced to improve the accuracy of calibration parameters. Apart from this, this paper outlines the proper calibration process, with the effectiveness of the proposed approach assessed in terms of reprojection error. Furthermore, the calibrated images have been tested using MATLAB and the OpenCV Library to analyze the reprojection error.