Non-parametric Model Camera Calibration Method And Error Analysis

Vision measurement system in a wide range of applications in the advancedindustrial manufacturing was supported by the rapid development of relatedtechnologies research. Camera calibration is an important part of vision measurementsystem, has an important impact on measuring performance. The existing cameracalibration methods based on parameters model have inherent defects and applicationrestrictions, such as parameters correlation interference, non-parametric error factors,etc. Which limit the measuring accuracy and adaptability to meet requirements ofhigh-precision and large field of view. A standard angle-curve data mapping isestablished in non-parametric model calibration method, which break limit ofparameter model calibration. The calibration platform was built; the feasibility of thecalibration method was verified in experiments. The difficult situation of large fieldview camera calibration is effectively improved.The content of the article include: universal adaptability of parameter basedcamera calibration method; description of non-parametric model calibration cameracalibration method, measurement model, and technology implementations; the criticalerror factors and key parameters needing fine adjustment. Experiments of singlecamera angle measurement and dual camera length measurement verified thefeasibility and accuracy. It shows that non-parametric method is better than theparametric model calibration method on measurement error, stability and adaptability.The innovative research includes: the realization of a large field of view cameracalibration method based on non-parametric model, the error analysis in imageprocessing, mechanical adjustment and one-dimensional target manufacturing, theverification scheme of method feasibility and correctness of the error theory.