How should you light visual egg inspection with a camera?

Why does lighting matter so much for grading?

The camera decides the class from the brightness and colour of the shell. A clean shell is already not uniform, because the curve of the egg creates a gradient, and light adds more variation. With diffuse, steady light that variation is predictable. With direct or uneven light, the system starts to see patterns that aren't there.

Vigiovo handles the natural gradient by measuring each egg against its own shading surface, and it excludes glare from the dirt measurement. That protects the result, but it doesn't replace good light: a glare that persists in two of the three measurements can still affect the decision.

Because deterministic vision measures brightness and colour against fixed thresholds, a steady light matters even more. For the comparison with neural networks, see deterministic vision vs neural networks.

What is glare, and why does it fool the camera?

Glare, or specular highlight, is the image of the light source reflected in a smooth surface. On wet, smooth or very shiny shells it appears as a bright point or a streak, and it can have the same size and shape as a dirt stain.

It moves when the egg turns on the conveyor and depends on the angle between the light and the camera. So without control of the light, glare is the main cause of false dirty grades: a clean egg comes out dirty because the camera measured a brightness that isn't dirt.

How do you set up light to reduce glare?

Use diffuse light from a large, soft source, and avoid small, intense sources that reflect off the shell. The setup that works best usually follows a few simple principles:

• Use a large, soft source, such as a diffuser panel, instead of small, intense lamps.
• Keep direct light off the shell and out of the lens, so the camera doesn't see the source reflected on the egg.
• Choose surfaces around the conveyor that don't reflect, such as dark, matte materials.
• Block outside light, especially sunlight coming through windows, because it changes during the day.
• Position the camera and the lights so the geometry is the same along the whole line.

There is no universal setup. The goal is the same on any line: a clean shell should appear with even brightness, without isolated bright points.

How do you keep the lighting steady through a shift?

Keep the intensity, colour and direction of the light constant, and treat the camera and its lens as part of the setup.

Light changes over time. Lamps lose intensity and shift colour with use, and dust on the diffuser and the lens creates marks that the camera can read as dirt. Clean the diffuser and the lens on a fixed routine and record when it was done.

Lock the camera's exposure and white balance so it doesn't readjust by itself at every change in light. A camera that readjusts changes the brightness of the whole image, and the dirt criterion, which depends on brightness, stops being valid.

If the setup is changed, for example by replacing a lamp, moving a panel or swapping the camera, repeat the calibration before returning to production.

How do you test whether the light is good?

Run these checks with the setup already installed, before trusting the results:

1. Pass clean eggs under the camera and check whether marks appear where there is nothing.
2. Turn the same egg under the light and confirm that the brightness follows the curve of the shell, rather than forming a fixed point.
3. Compare images from different shifts and confirm that the shell looks the same.
4. Review a sample of the eggs marked dirty and check whether glare is the cause of the error.

If the review shows recurring glare, adjust the light before touching the thresholds. Adjusting thresholds to compensate for a lighting problem hides the defect and makes the calibration worse. The full process is in calibrating with labelled eggs.

Frequently asked questions

Does sunlight through a window affect inspection?

It does if it varies. Daylight changes during the shift and creates glare and bright patches the setup didn't account for. Block or cover the windows above the inspection area and always use the same lighting.

Can glare make a clean egg grade as dirty?

Yes, and it is the most common cause of false dirty grades. Vigiovo excludes glare from the dirt measurement and rejects an egg only when the defect shows up in two of the three checks. That lowers the error but doesn't remove it, so lighting remains the first defence.

Do I need to recalibrate if I change the lighting?

Yes. The thresholds were tuned under the calibration light, and different lighting changes the brightness and colour of the shell. Repeat the calibration with the new setup before returning to production, and record the date of the change.

Sources

1. Como o Vigiovo classifica ovos sem IA / How Vigiovo grades eggs without AI, Vigiovo