Knowledge · Injection Moulding

Detecting Flash in Injection Moulding with a Camera: What Resolution Is Required?

How many megapixels a camera actually needs for reliable flash detection in injection moulding, and why resolution alone is not enough.

The required sensor resolution depends on the field of view (FOV), the minimum detectable flash geometry, and an optical safety factor. For process-stable industrial detection, the smallest feature should be mapped to at least 3 pixels. However, how many pixels are actually required also depends on contrast, lighting, optics, motion blur and the inspection method.

The mathematical pixel calculation

To precisely design an optical inspection system for flash detection, it is not enough to simply demand a “5-megapixel camera”. Geometric object resolution is calculated from the field of view (FOV) in relation to sensor resolution, multiplied by the industrial Nyquist safety factor for robust edge detection. This figure is one of the technical core parameters that should be clearly defined in every requirements specification for a vision system:

Formula · Required Pixel Count

Pixel count = 3 × FOVd

Practical example: 3 × 130 mm0.1 mm = 3,900 pixels

FOV = field of view [mm] · d = smallest feature [mm]

An automotive injection-moulded part with a total width of 130 mm is to be fully inspected for a fine flash of at least 0.1 mm. A standard 5-megapixel sensor typically delivers 2,448 × 2,048 pixels. It is therefore mathematically unsuitable for this task. To solve this application reliably, THOP AG configures the inspection station either with a high-resolution 12-megapixel sensor (approx. 4,096 pixels horizontal) or splits the field of view across two dedicated cameras with a smaller field of view each.

Interactive Tool

Pixel calculator: work out the required resolution

Required pixel count
Recommended resolution class

Why resolution is worthless without contrast

A sensor with an extremely high megapixel count adds no value if the flash is not optically distinguishable from the rest of the part. For flash detection, the choice of lighting geometry is critical:

  • True outer contours: Backlighting (transmitted light) is the first choice here. The part appears as a sharp, black silhouette against a homogeneous bright background. Flash breaks up the contour with maximum contrast.
  • Internal openings or flash on flat surfaces: Here, THOPinspect systems work with extremely shallow grazing light (dark field). The defect-free flat surface stays dark, while the vertical geometry of the flash reflects light and appears bright.

If even optimal lighting is insufficient because interfering ambient light overlays the image, additional measures are needed, up to a fully enclosed inspection tunnel.

💡 Practical Tip from THOP AG

Never rely purely on calculated pixel values. Use the THOP Application Lab. We test your injection-moulded parts under real cycle-time conditions and determine the exact sensor-lens combination, including proof of contrast.