A Robust and Accurate Calibration Method for Out-of-focus Camera.
Xiaowei Hu,Guijin Wang,Jinnan Wang,Pengfei Sun,Jingtao Fan,Feng Chen,Yiyuan Xie
DOI: https://doi.org/10.2352/issn.2470-1173.2018.2.vipc-263
2018-01-01
Abstract:Conventional camera calibration methods regard camera as ideal pinhole model and require well-focused images, which can't be satisfied for long-range photogrammetry or low depth-of-field lens. In this paper, we propose a novel active calibration method for out-of-focus camera using LCD monitor. Firstly, we estimate the defocus map by the temporal coded binary-shift patterns, which makes our method more accurate. Secondly, based on the defocus map, we encode LCD pixel's coordinates into phase-shift patterns with optimal frequency and step properties, and then deblur captured patterns. Finally, deblurred patterns are decoded to generate dense phases map to extract accurate feature points coordinates. Our method significantly improves camera calibrations robustness to lens' defocus, noises, glass refraction compared with state-of-art methods. Experimental results demonstrate that our method is superior to conventional methods whether camera is in- or out-of- focus.