Multi-RGBD-camera calibration toolbox

Abstract

Development of affordable RGBD cameras, especially Microsoft Kinect camera, created a new wave of real world applications for RGBD cameras especially in entertainment, healthcare and 3D modelling industries. However, applications which uses single RGBD camera to reconstruct 3D data tends to underperform due to inconsistency when capturing different perspectives of the model. Moreover, 3D reconstruction using Single RGBD camera means it is impossible to capture different perspectives at the same time or in other words, it is impossible to do live modelling. Thus, multiple RGBD cameras setup becomes popular for applications which requires fast results. With such setup comes with its own set of challenges when combining different perspectives. Reliable geometric calibration is needed to recognize the different positions of the cameras. This project implement a calibration toolbox for calibrating multiple RGBD cameras using Rigid Transformation method. To improves the accuracy, the uses of Local Rigid Transformation is also explored. Additionally, calibration pattern called ChAruco board is used instead of normal Chessboard to improve the efficiency when calibrating.