7. Wavelenghts
- Wilhelm Scheermann
The electromagnetic spectrum consists of several different regions defined by wavelength range. Choosing a wavelength to complement the needs of the inspection and the colors) of the workpiece is key to any machine vigion application.
Below are the standard wavelengths used in machine vision, extending from 365 nm (ultraviolet) to 940 nm (infrared).
Color is a perception rather than a physical property. The observed color of any object depends on not only the spectral distribution of the incident light, but also the optical characteristics (spectral reflectance) of the workpiece itself, and the response of the imaging system (either the human eye or the camera sensor). Thus, the same object can appear to have variations in color when viewed with the same sensor using different sources or different sensors using the same source.
Machine vision must match the colors in an image with target colors predefined for the inspection system. The CIE color system, defined by the International Commission on Illumination (CIE), is the industry standard that numerically specifies colors. Based on the theory of additive color mixing, colors can be displayed mathematically in coordinates as shown in the x, y chromaticity diagram. All possible colors fall within the horseshoe shape, specified by the coordinate position (x, y) of this plane.
7.1 White Light
The visible spectrum is the wavelength range visible to the human eye from 400 to 750 nm. White light contains all of these wavelengths, but white light sources vary depending on the spectral distribution. Two measures to evaluate a white light source are correlated color temperature and color rendering index. The correlated color temperature of a white light source is the temperature in degrees Kelvin (K) required for an ideal black body radiator to emit light with the closest spectral distribution match to the light source.
White light with a low correlated color temperature has a reddish tinge and is referred to as 'warm' light, while white light with a high correlated color temperature has a bluish tinge and is referred to as 'cool' light. White LEDs used in machine vision tend to have a cooler color temperature that produces a blue tint.
Color rendering describes how the color appearance of an object changes depending on the light shining on it, and the color rendering properties of a light source determine how the colors will appear. A light source with good color rendering properties can faithfully reproduce the perceived color of an object illuminated by a natural or ideal light source such as a CIE daylight source or blackbody radiation source with the same correlated color temperature.
Natural Light (high color rendering) LED
General white LED
NATURAL WHITE LED
LEDs often have a cooler color temperature that produces a blue tint (right). CCS has developed a high-color rendering natural light LED with a spectral distribution close to sunlight to accurately reproduce colors in an image (above left).
7.2 RGB Light
While almost any color light can be custom-made by lighting manufacturers, the standard colors available are red, green, and blue. Depending on the application, specific wavelengths can increase the contrast of features in an image. Like colors or families lighten the appearance of an object, while opposite colors or families darken the appearance. For example, shining red light on red features will reflect the light and appear brighter, while green features or blue features will absorb the light and appear darker.
RGB lights blend red, green, and blue light sources to produce white light. These lights can also switch between individual wavelengths within the same light unit. RGB lights are ideal for inspections where the object is multicolored or applications with different colored products on the same assembly line.
7.3 Infrared Light
The infrared (IR) region of the spectrum covers 700 nm – 1,000,000 nm. CCD, CMOS, and InGaAs sensors used in most machine vision cameras can detect infrared radiation, unlike the human eye, which allows them to perform inspections not possible for humans.
Infrared light has a longer wavelength and less energy than visible light. As a result, it generates less interaction, with low scattering and high transmission. These properties can cancel out the appearance of surface features such as print on packaging and imaging the inside of opaque materials.
Most infrared machine vision applications use 850 nm or 940 nm wavelengths in the near-infrared (NIR) range. Recent technological developments have also increased the use of shortwave infrared (SWIR) up to 1,700 nm.
7.4 Ultraviolet Light
UV radiation wavelengths range from 10 - 400 nm, with 365 nm or 395 nm being the most commonly used UV wavelengths in machine vision. The higher energy associated with UV radiation brings two benefits.
First, when used in conjunction with a UV-sensitive camera, it offers the possibility of resolving submicron details that are not possible using visible light. Second, it can stimulate a fluorescence emission from certain materials such as dyes and inks. This emission occurs at longer wavelengths in the visible region of the spectrum where machine vision sensors can easily detect them. In addition, when used in reflectance applications, differences in spectral reflectivity and absorption to UV radiation in a workpiece can make translucent information visible.
Fluorescing Particles with UV Light
In some cases, particles that do not appear under visible light (left) will fluoresce under ultraviolet light (right), meaning the light reflected towards the camera from the sample is about 400 nm. Using a UV cut filter will block the UV light emitted from the light unit so that only the fluoresced light will appear bright in the image.