| 1 | Define the smallest defect | Detect scratches, gaps, chips, or printing defects down to a specified physical size. | Plan for at least 2 pixels across the smallest defect; 3–4 pixels generally provides more reliable detection. | Required spatial resolution = smallest defect size ÷ desired pixels per defect. | Insufficient sampling can make a real defect appear too small, blurred, or invisible. |
| 2 | Match sensor width to the inspection field | Inspect a web, sheet, roll, or product that may be wider than the camera's active sensor area. | Choose enough pixels to cover the full field of view while maintaining the required pixel size. | Pixel size on the object = field of view width ÷ active pixel count. | A wider field with too few pixels reduces defect visibility and measurement accuracy. |
| 3 | Calculate the required line rate | Inspect a continuously moving product or web without motion distortion or gaps between lines. | Required line rate should be at least web speed ÷ desired object resolution, with engineering margin. | For example, 1,000 mm/s at 0.10 mm per line requires 10,000 lines/s before adding margin. | The line rate must track product movement so each image line represents the correct distance. |
| 4 | Select suitable sensor sensitivity | Inspect colored, reflective, translucent, or low-contrast materials under available lighting. | Use silicon for approximately 400–1,000 nm applications; consider extended near-infrared sensitivity for wavelengths above 1,000 nm. | Compare the sensor's quantum efficiency with the illumination wavelength and material response. | Spectral mismatch can reduce contrast even when the nominal resolution is adequate. |
| 5 | Check exposure time and motion | Capture sharp images from fast-moving products, especially when illumination is limited. | Keep exposure short enough that motion blur remains below the allowed object-space resolution. | Approximate motion blur = product speed × exposure time; use pulsed lighting when appropriate. | Long exposure increases light collection but can smear edges along the direction of travel. |
| 6 | Choose the right pixel size | Balance sensitivity, resolution, optical performance, and the available working distance. | Larger pixels generally collect more light; smaller pixels support finer sampling when the lens can resolve it. | Verify lens resolution, pixel pitch, depth of field, and diffraction performance together. | A high pixel count does not improve results if the optics or lighting cannot resolve the detail. |
| 7 | Evaluate dynamic range | Handle shiny surfaces, shadows, variable backgrounds, or scenes containing bright and dark areas. | Prefer higher dynamic range when the application has strong illumination variation or low-contrast defects near highlights. | Review full-well capacity, read noise, bit depth, and measured signal-to-noise ratio. | Adequate dynamic range helps preserve defect information without saturating bright regions. |
| 8 | Confirm trigger and encoder support | Maintain consistent image scale when conveyor speed changes or products are spaced irregularly. | Use an encoder-based trigger for variable-speed lines and a stable hardware trigger for fixed-speed systems. | Check trigger frequency, input voltage, timing jitter, and encoder pulses per millimeter. | Precise synchronization prevents stretched, compressed, or geometrically distorted images. |
| 9 | Match the interface to data volume | Transfer high-resolution line data continuously to an inspection computer or controller. | Estimate throughput before choosing an interface; use a connection with sufficient sustained bandwidth and cable length. | Approximate raw data rate = pixels per line × bytes per pixel × line rate, excluding overhead. | A bandwidth shortfall can cause dropped lines, buffering, or reduced production speed. |
| 10 | Validate the complete imaging system | Achieve stable inspection results during real production conditions, including vibration, temperature, dust, and material variation. | Test the camera, lens, lighting, triggering, software, and mounting as one system before final selection. | Measure detection rate, false rejects, image uniformity, calibration stability, and uptime at production speed. | Real-world performance depends on the complete optical, mechanical, electrical, and software setup. |