Comprehensive improvement of camera calibration based on mutation particle swarm optimization

• Published in: Measurement : journal of the International Measurement Confederation Vol. 187; p. 110303
• Main Authors: Lü, Xueqin, Meng, Lingzheng, Long, Liyuan, Wang, Peisong
• Format: Journal Article
• Language: English
• Published: London Elsevier Ltd 01-01-2022; Elsevier Science Ltd
• Subjects: Adaptive weight and mutation; Algorithms; Calibration; Camera calibration; Cameras; Data acquisition systems; Digital cameras; Image acquisition; Image contrast; Image enhancement; Image quality; Mutation; Optimization; Parameters; Particle swarm optimization; Pixels; Sub-pixel extraction; Adaptive weight and mutation; Sub-pixel extraction; Camera calibration; Particle swarm optimization; Image enhancement
• ISSN: 0263-2241; 1873-412X
• DOI: 10.1016/j.measurement.2021.110303

•Improving the contrast of calibrated images.•Improving the accuracy of sub-pixel angle extraction.•Optimizing camera parameters nonlinearity.
In order to meet the requirements of high-precision measurement, the method of improving camera calibration is studied. In the calibration process, the quality of the calibration image, the extraction accuracy of the calibration image corner and the nonlinear optimization effect of the camera linear parameters directly affect the calibration accuracy. First of all, in order to solve the problems in image acquisition, especially in the case of over exposure, an adaptive gamma correction method is designed to automatically adjust the image brightness, and enhance the contrast of black and white grid to improve the image acquisition quality. Secondly, a sub-pixel corner extraction algorithm based on homography matrix mapping is designed, which overcomes the error and omission of Harris corner extraction algorithm, and improves the accuracy of corner extraction. At last, adaptive weight and mutation particle swarm optimization algorithm are studied to optimize the camera parameters. Compared with other optimization algorithms, this optimization algorithm needs less parameter settings, fast convergence speed, and can obtain more accurate camera parameters. The average calibration error is 0.038 pixels.