Abstract:
Mobile vision systems greatly benefit from the large field-of-view enabled by wide-angle lenses. Accurate and robust intrinsic calibration is a critical prerequisite for leveraging this property. Calibrating wide-angle lenses with current state-of-the-art techniques yields poor results due to extreme distortion at the edge. In this work, we present TartanCalib, an accurate and robust method for wide-angle lens calibration. Our pipeline iteratively improves an intermediate camera model during the calibration process. We achieve robust target feature detection in the highly distorted image regions by using our proposed initial detection and adaptive sub-pixel feature refinement method with an intermediate camera model. TartanCalib improves the overall reprojection error by up to 27.3%, and detects up to 29.4% more features for wide-angle lens calibration. Finally, we open-source TartanCalib as an easy-to-use toolbox for the community.

Committee Members:
Sebastian Scherer (Chair)
Michael Kaess
Deva Ramanan
Ben Eisner (PhD Student)