How to Choose the Right LED Color for Machine Vision

The Physical Problem

Most machine vision cameras are monochrome sensors: they measure light intensity, not hue. Under a given illumination wavelength, a red feature on a green background might reflect almost identically to how the green background reflects, producing two regions of nearly the same gray level in the image — even though a human eye, with its three types of color receptors, would separate them instantly. Pick the wrong wavelength, and a perfectly visible color-coded defect, print, or component simply disappears into the background as far as the camera is concerned.

The Optical Solution

A colored surface strongly reflects light of its own color and strongly absorbs light of the complementary color on the color wheel — a direct, well-established consequence of selective reflectance. Illuminating a red feature with red light makes it reflect brightly, appearing light in the image; illuminating that same red feature with blue or green light instead makes it absorb strongly, appearing dark. Maximizing contrast between two differently colored regions is therefore a matter of choosing a wavelength that one region reflects and the other absorbs — brightening what you want visible, and darkening what you want suppressed, before the camera ever sees a single pixel.

Integration Tools

• M12 5-Pin Wiring Diagram