| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 803872 | Precision Engineering | 2015 | 10 Pages |
•A diagnostic method for detecting variations of camera internal parameters in photogrammetry applications is proposed.•The main innovative aspect concerns use of multivariate control charts.•This approach, relying only upon data acquired during calibration phase, enables quick and comprehensive stability control at each calibration.•The method can be easily integrated in calibration software without the need of additional procedures or artefacts.•The robustness and reliability of the method have been proved by tests performed on a commercial multi-camera system and a laboratory prototype.
Industrial non-contact dimensional measurements using photogrammetry rely critically upon stability in time of camera calibration. This is particularly relevant for multi-camera systems employed for continuous and/or long term monitoring of some dimensional process, e.g. dimensional checks of the same manufactured component as it comes off the production line. In most of these cases, camera calibration is updated regularly to ensure optimal accuracy. Specifically, the use of photogrammetric systems requires the knowledge of both internal and external camera parameters estimated by calibration. Constancy of both sets is required during use. Internal parameters, pertaining to camera-specific properties, require stability over the operational lifespan of the system, while external parameters, concerning location and orientation, may change between calibrations. A diagnostic method for internal parameters based on multivariate control charts is proposed. The purpose of this method is to provide a comprehensive stability control over all the performed calibrations, especially for those systems used for regular monitoring of production lines. By integrating chart building into calibration software, no additional steps are added to the operator's workload for the calibration process. A practical application of the described methodology is presented at the end of the paper.
