In a machine vision system, camera selection is critical, as it determines image quality, inspection accuracy, system speed and overall performance. Faced with a wide variety of cameras on the market, how to select the product that fits your requirements?

1. Camera Resolution and System Accuracy

Camera resolution defines the inspection accuracy of the system. Generally, accuracy can be calculated using the formulas below:

Accuracy along X-axis = Field of View (X-axis) / Number of CCD pixels (X-axis) Accuracy along Y-axis = Field of View (Y-axis) / Number of CCD pixels (Y-axis)

Nevertheless, theoretical pixel accuracy is not completely equal to the final inspection accuracy of the system, since it is also affected by sub-pixel technology.

What is Sub-pixel?

Sub-pixel is a method to improve image resolution. A single pixel can be subdivided into multiple sub-pixels to calculate object positions with higher precision. For example, in high-precision measurement, a 10×10 sub-pixel algorithm can boost inspection accuracy to 1/10 of the original pixel precision or even higher.

2. Camera Speed and Imaging Performance

System speed is determined by imaging speed and inspection speed:

Imaging Speed = Exposure Time + Transmission Time Inspection Speed = Image Processing & Analysis Speed

If the system requires high-speed imaging (such as assembly line inspection, motion analysis, etc.), high-speed CMOS cameras with Global Shutter are recommended to avoid motion distortion and deliver sharper images.

3. CMOS vs. CCD: Which Sensor Is Better?

表格

General guidelines:

• CCD: Suitable for high-precision and low-light applications (e.g. medical imaging, scientific research)
• CMOS: Suitable for high-speed applications (e.g. industrial inspection, motion analysis)

4. 12-bit Camera vs. 8-bit Camera

Camera bit depth determines the number of distinguishable grayscale levels:

• 8-bit camera: Up to 256 grayscale levels
• 12-bit camera: Up to 4096 grayscale levels

If your application demands a high dynamic range, such as medical imaging and precision inspection, a 12-bit camera is recommended to capture richer image details.

5. Analog vs. Digital Cameras

• Analog Cameras: Transmit images via analog electrical signals. An image frame grabber is normally required for A/D conversion.
• Digital Cameras: Output digital signals directly, preventing image degradation and noise interference during transmission.

For high-precision and high-stability imaging, digital cameras are preferred, with an appropriate interface selected according to application needs.

6. How to Choose Camera Output Interfaces?

Different interface standards govern data transmission modes:

• USB3.0 / GigE: Suitable for most industrial inspection scenarios with favorable cost performance
• Camera Link: Designed for high-resolution, high-speed imaging; requires dedicated frame grabbers
• Analog Interfaces (PAL/NTSC): Mainly deployed in legacy surveillance systems
• IEEE 1394 (FireWire): Once popular for industrial cameras, gradually phased out

7. Camera Synchronization Methods

For multi-camera systems, synchronization is essential to ensure simultaneous image capture across multiple cameras. Common synchronization modes:

• Vertical Sync: Basic synchronization method for surveillance systems
• Composite Video Sync: Used for analog video synchronization, without guaranteed color accuracy
• External Sync: Allows cameras to lock to an external signal source such as a master camera
• Genlock: High-precision synchronization for broadcast and precision industrial inspection
• DC Line Lock: Synchronization via AC power supply, used in legacy surveillance systems (being phased out)

8. How to Select Shutter Type?

The shutter controls the exposure mode, mainly categorized into two types:

Global Shutter

Suitable for fast-moving objects Typical applications: industrial inspection, motion analysis

Rolling Shutter

Exposes row by row, which may lead to the rolling shutter effect (jello effect) Suitable for static scenes

For high-speed imaging, global shutter cameras are recommended to eliminate image distortion caused by motion.

9. Why CCDs Cannot Detect Light Above 1100 nm?

The working principle of CCDs limits their capability to detect infrared light (>1100 nm), for the following reasons:

• Silicon has low absorptivity for long-wave infrared light
• Infrared radiation tends to induce thermal noise
• Pixel structures cannot efficiently collect long-wavelength photons

If detection of infrared light is required, the following options are available:

• NIR-enhanced CCD
• InGaAs Sensors
• Thermal imaging cameras

10. Camera Selection Based on Application Requirements

表格

Summary

To select a suitable camera, comprehensive evaluation should cover:

• Resolution and accuracy
• Speed and shutter type
• CCD vs. CMOS comparison
• Interface selection
• Synchronization method
• Application requirements
• Budget

Different application scenarios impose different requirements on cameras. The priority is to find the product matching your practical demands, rather than simply pursuing top-tier specifications or the latest technologies.