Hemispherical Infrared - Sensor
Sensor calibration and compensation
To Be completed
TODO - aka to be put in this page:
- Sensor calibration set-up (follow PDF) ?
- Homogeneity by hand
- Outlook on calibration using overlap
- Striping defects
- Super resolution ?
Sensor calibration
As we have a good control of where the sensor looks at, we can create a thermal point source, observe it under different angles and extract a good geometric sensor model.
Which is what I’ve done, allowing to extract spherical aberration, horizontal Field of view, vertical field of view. Sensor Center was not a relevant issue here.
Sensor homogeneity
A clear improvement was visible by using a cos^4 law for the temperature relative to the angle from the center. It’s thermal vignetting so it should be more complex than optical vignetting as the “occulting” black piece is an emitter itself, so I think it should depends on the sensor package temperature. Hopefully between the internal temperature probe and the average IR temperature we should obtain a good proxy.
More work need to be done here, but please enjoy a few results.
Auto calibration using overlapping images
The great aspect of having a servo mounted sensor is that we can control the overlap of the different images. With this overlap we can adjust a model on the sensor behavior. Either a “by hand” model or a “learned” model.
Project maintained by Brice Dubost-Vedrenne Theme jekyll-theme-midnight by mattgraham modified by Brice Dubost-Vedrenne
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.