search

ntnu-arl / reaqrovio

An underwater visual inertial odometry with online refractive index estimation method based on ROVIO to enable reliable state estimation without camera calibration in water.
https://ntnu-arl.github.io/refractive-camera-model-in-vio/
Other
26 stars 1 forks source link

Using domed port covers. #1

Open ChristianWesterdahl opened 3 weeks ago

ChristianWesterdahl commented 3 weeks ago

In the associated paper, and I would also assume this implementation, you describe the that the camera is located directly behind a thin (2.5 mm) flat port waterproof casing lens. The refractive distortion through this is modeled by g_r in the paper.

We would very much like to use ReAqROVIO in a setting with a stereo camera, where both cameras are positioned behind glass domes, with the camera lens directly in the centre of the dome. Theoretically, this would eliminate any refractive distortion from the glass window. Is this as easy as to set the refractiveCalibration and useObservabilityCheck both to false?

Mohit505Git commented 2 weeks ago

Yes, for dome ports with cameras at the geometric center, the refractive effects are absent (theoretically). Thus, just setting refractiveCalibration and useObservabilityCheck to false with refractive_index to 1.0 is enough.