In this thesis, we tackle the engineering problem of Discrepancy Check: verifying the geometric correctness of a built object against its virtual model.
We propose to use Augmented Reality (AR) to verify the model against the built state.
Acquired images of the plant are aligned with the model to offer Mixed Views, which helps engineers to detect clashes and act upon them by judging the geometric quality of a construction.
In a first part, we present interaction techniques to use Mixed Views to detect description, as well as a registration method suited for large industrial compound based on anchor-plates.
In a second part, we develop computer vision methods to evaluate the localization error of multi-scale feature, to improve the estimation an essential matrix using intensities combined with geometric measurement, and to extend a relative pose to a full pose.
«
In this thesis, we tackle the engineering problem of Discrepancy Check: verifying the geometric correctness of a built object against its virtual model.
We propose to use Augmented Reality (AR) to verify the model against the built state.
Acquired images of the plant are aligned with the model to offer Mixed Views, which helps engineers to detect clashes and act upon them by judging the geometric quality of a construction.
In a first part, we present interaction techniques to use Mixed Views...
»