1. Camera Requirements
For reliable face recognition, your camera should meet the following technical specifications:
| Parameter | Requirement |
|---|---|
| Resolution | At least 1 megapixel for each meter to the person's face. This ensures the captured image has enough detail for the system to identify facial features accurately. |
| Sensitivity | A minimum of 0.1 lux. In low-light environments, you may need a camera with even higher light sensitivity to maintain image clarity. |
| Bitrate | A stable video stream with a bitrate of at least 2 Mbps. A lower bitrate can degrade image quality and interfere with recognition. |
| Shutter Speed | Set to a maximum of 1/200 second. A faster shutter helps to prevent motion blur, which is a common cause of recognition errors. |
| PACs Integration | Must have at least one alarm (relay) output. This hardware feature is necessary to physically trigger doors, gates, or other access control systems. |
2. Installation Guidelines
Correct placement is critical for accurate recognition. Follow these guidelines:
Distance: The maximum effective distance for recognition depends on your camera's resolution. Follow the rule of 1 megapixel (MP) per meter to the subject's face. For example, to clearly identify a face 5 meters away, you will need at least a 5 MP camera.
Height: For the most natural angle, install the camera at approximately 1.8 meters (average human eye level). If there is no suitable mounting point at this height, you can use a ceiling-mounted pole or bracket to lower the camera to the correct position.
Lighting: The detection area must have uniform and diffuse lighting.
- Minimum Level: Ensure at least 150 lux of illumination in the area where people will appear. You can measure this using a light meter app on your smartphone (e.g., Lux Light Meter).
- Light Sources: If the area is too dark, install additional light fixtures. Both insufficient and excessive, harsh light can be problematic, as they make it difficult to distinguish facial features from the background.
- Avoid Shadows and Glare: Strong shadows or bright reflections on a person's face hide important visual details and can lead to recognition errors or false matches.
Angle: The camera's viewing angle is crucial for providing useful images.
- Best Practice: Position the camera so that faces are captured frontally (face-on) as much as possible.
- Angles to Avoid: Recognition accuracy drops significantly for faces seen in profile, from a high angle (looking down), or from a low angle (looking up). These angles show only a partial view of the face (e.g., a profile may show only one eye), which provides insufficient data for reliable identification.
This configuration ensures optimal performance for both face recognition and integration with PACs.