Add colour to your Machine Vision Solution

Machine Vision is a technology that can automate visual processing tasks. One of the great gifts of human eyesight is the ability to differentiate colours. But colours are often subjective and unless two similar shades are seen next to each other, differences are hard to spot.

Applications

There are many ways machine vision can be used to inspect colours but one common problem is to ensure that parts of a product visually matches closely to each other. Example in paint shops we need to ensure that auto and bike parts match between all the parts being produced. In packaging, discoloured packaging, for example, may cause customers to think a product is old or defective in comparison to the surrounding correctly coloured packages. These types of perceptions can be predicted, quantified, and corrected by carefully installing a visual colour-matching system.

Specifically, manufacturers use colour vision to solve three primary vision applications:

• Color verification -- Verifying that a certain part's colour matches what the vision system is programmed to find.
• Colour sorting -- Sorting parts based on colour.
• Colour inspection -- Inspecting coloured parts for defects that grayscale image processing tools can't detect.

Colour Standards

How colour is described makes a big difference in arriving at the right results. Some standard colour spaces, such as RGB are good for specifying how a camera should detect a colour but are not correlated with how a human will perceive it. There are other standards, such as CIELab standard which are more correlated with how a human perceives colour.

RGB Color Space

RGB colour can be understood by thinking of it as all possible colours that can be made from three coloured lights for red, green, and blue. Imagine, for example, shining three lights together onto a white wall in a dark room: one red light, one green light, and one blue light, each with dimmers. If only the red light is on, the wall will be red. If only the green light is on, the wall will look green. If the red and green lights are on together, the wall will look yellow. 

Cie Lab Color Space

Unlike the RGB and CMYK colour models, Lab colour is designed to approximate human vision. CIELab colour values are ideal for visual matching systems because they represent colour the same way humans perceive it. This standard uses so-called opponent theory, which relies on the fact that an object cannot look red and green at the same time, nor can it look yellow and blue at the same time. In the CIElab system three values (L*, a* and b*) quantify a colour. The first (L*) quantifies perceived brightness. The second (a*) represents how red or green the object looks. Positive a* values represent reddish colours and negative values indicate greenish ones. The third parameter (b*) indicates yellowish versus bluish colours. Positive b* indicates yellow and negative b* indicates blue.

Colour Difference

The International Commission on Illumination (CIE) calls their distance metric ΔE  ("Delta E") which is the measure of the change in visual perception of two given colours. In other words, it is a metric for understanding how the human eye perceives colour difference. 

On a typical scale, the Delta E value will range from 0 to 100, which can be interpreted according to the table below:

Multispectral Cameras

Multispectral cameras are an upcoming breed of technology that can be used. A multispectral image sensor captures image data at specific frequencies across the electromagnetic spectrum. The wavelengths may be separated by filters or by the use of instruments which are sensitive to particular wavelengths, including light from frequencies beyond our visible sight, such as infrared. For example in the colour samples below, each wavelength can be individually extracted and plotted against its reflectance. These cameras make colour identification and differentiation a lot more accurate than measuring with a camera that can differentiate only three wavelengths (R, G and B).

Summary

Machine Vision is a great technology for various colour applications ranging from sorting or identifying parts based on colour information or even to measure colour differences and deviations from expected colour shades. Whichever application you chose to solve using this technology, make sure you select the right approach with respect to colour standards and algorithms. There are a variety of options to chose from.