Automatic Camera Calibration for Stereo and Muti-view Stereo Systems


Estimation of camera geometry parameters is a fundamental task in photogrammetry and 3D vision. Especially in the field of 3D and SLAM systems a correct knowledge is required to get high accurate measurements of the environment. JOANNEUM RESEARCH developed a prototype of a mobile SLAM platform containing 3 cameras to reconstruct miscellaneous objects. The platform is currently calibrated semiautomatic with high time consuming human effort. In view of serial production and accuracy this process should be improved and automated so that also nonprofessionals can calibrate the system. Potential miss or weak calibrations must be detected before the customer uses the device in his field of application. The graduate student should implement and evaluate different approaches to develop such an automatic calibration system. The software should be written in C/C++ and Windows within JR’s software libraries, the usage of diverse license free open source libraries is possible.


C/C++ with Microsoft Visual Studio Math and computer vision skills


6 months