Moiré and false color in surveillance images: lens and sensor matching issues
What is moiré pattern
When there are dense periodic textures in the picture (such as blinds, screens, striped clothes, dense railings), while the spatial frequency of the texture is close to the sensor pixel arrangement frequency, New low-frequency interference fringes will be produced - this is moiré, which appears as colored ripples or grids.
What is false color
Pseudo-color refers to the appearance of non-existent color patches in grayscale or low-saturation areas. Commonly seen near high-contrast edges. It is derived from the interpolation algorithm of color filter array (Bayer array) Misjudgment around edges, especially when sharpness is turned up too high.
Easy scenarios
| Scene | Performance | Processing |
|---|---|---|
| Facing screens/blinds | Large area of colorful ripples | Adjust the angle or switch to a high pixel density model |
| Dense railings/fences | Stripe deformation | Slightly adjust the pitch angle and focal length |
| Dense building facades in the distance | Grid texture | Reduce sharpness and avoid over-sharpening |
| Edges at high sharpness settings | Edge color fringing | Sharpness adjusted back to 30~45 |
Processing measures
- Physical level works best: Slightly change the pan or tilt angle of the camera, If the texture frequency deviates from the pixel frequency, moiré patterns tend to disappear immediately.
- Adjust focus: Changing the viewing range will change the frequency of texture projection on the sensor. It is the simplest adjustment method on site.
- Increase pixel density: The denser the pixels, the higher the sampling frequency, and the less likely it is to be close to the texture frequency. This is why high-end models have less moiré.
- Reduce sharpness: Reduces color artifacts, but is only a compromise.
Distinguishing from equipment failure
The moiré pattern changes as the content of the screen changes (it disappears when you change the angle). Stripes caused by sensor or lens failure are usually fixed and do not change with the scene. When a problem occurs, you can quickly distinguish it by rotating the camera angle.