Colour Calibration

Overview

Images from the $200 CV camera with the default settings appear purple, which is undesired. This is most likely caused by light in the UV spectrum being visible to the camera sensor.

Images have purple hue

Adjustments

2023-10-28

After collecting images with different settings in the flight test, we decided not to apply any calibration with v4l2. All the parameters are to set as the default value.

2023-10-19

We took pictures today with the following settings:

hue=-512 white_balance_temperature=6500 white_balance_automatic=0 auto_exposure=1 exposure_time_absolute=64

Since the sun was directly facing the rooftop garden, some of the imaging was poor, particularly where there were reflections. Lowering exposure time helped reduce the impact it had. Images stored using model-training/imaging/imaging.py

2023-10-12

We took pictures today with the following settings: hue=-128, white_balance_temperature=6500, white_balance_automatic=0, and auto_exposure=1.

2023-08

We found that turning off the white_balance_automatic, auto_exposure, setting white_balance_automatic to 5500 , and everything other controls/settings to default worked the best.

Note that by adjusting and increasing white_balance_temperature the camera tint became more yellow-ish which allowed for greater visibility of the color blue.

How to adjust

List of controls + example commands

To see the list of controls run:

v4l2-ctl -d /dev/video2 -d /dev/video2 --list-ctrls

 

To change one of the controls run:

v4l2-ctl -d/dev/video2 --set-ctrl control=value

for example: v4l2-ctl -d/dev/video2 --set-ctrl hue=32

Additionally you can set it to control=default, control=min, or control=max if needed. Be careful not to include a space between the control and value.

Â