A novel feature points extraction method for defocused camera calibration based on phase-shift circular fringes

Abstract The depth of field in cameras is inherently limited, notably in defocused scenarios where blurred feature corners or edges can considerably decrease calibration precision. Consequently, this study proposes a novel anti-defocus and high-precision calibration method based on phase-shift circular fringes (PSCFs). This approach overcomes the susceptibility of traditional calibration patterns to defocusing. The geometric constraints of the PSCFs facilitate reliable control point detection and establish an eigenvalue minimum error voting scheme, thereby circumventing numerical instability in feature point estimation caused by edge detection and noise. Compared with the chessboard method, the calibration accuracy is improved by 27% and 84% respectively in the case of focus and defocus. In three-dimensional measurement experiments, errors in grid distance and circle radius measurements were recorded at 0.086 and 0.014 mm respectively. Results indicated that this approach provides a solution for high-precision, flexible calibration of cameras with large FOVs.