Match makers

Exact colour matching of different parts of a car has until now been a well-nigh impossible task. Although few of us notice that a bumper, say, is a different shade to the vehicle's body, German researchers have developed a way to match the colours more accurately.

The parts are painted by different suppliers, so to ensure the desired colours and special effects are accurately matched a team from the

, IFAM, has developed a computer program to distinctly characterise complex colour shades.

This information can than be electronically passed on to suppliers — a far cry from the slow and error-prone procedure requiring companies to exchange painted metal colour chips, or product samples and have them visually matched by trained colourists.

One of the biggest obstacles to exact colour matching is the lack of a means to reproduce colour impressions accurately enough using physical colour values. With special-effect pigments mixed into the paint, whose metallic or pearl lustre effect varies under different lighting and from different viewing angles, the colour can change completely. This effect — known as 'flip-flop' — causes a colour such as copper red to change to green.

Taking this in to account, the researchers developed the program to characterise these complex colour shades in the form of spectral data. They were able to do this by characterising each pixel in an electronic image, not simply by the three conventional red, green, blue (RGB) values but across the entire colour spectrum. This enabled them to measure even inhomogeneous multicoloured structures with glitter and flip-flop effects.

The researchers worked with German multispectral imaging specialist

. Multispectral imaging combines metrology and RGB imaging to carry out millions of spectral measurements on a single spot at the same time. ColorAIXperts had already been using its technology in the textile industry but it was Fraunhofer's idea to apply it to car manufacturing.

'The technology was basically developed for the textile industry for patterns on fabrics,' said IFAM's Volkmar Stenzel. 'We adapted it for use in automotive colour shades.'

RGB imaging in itself is not an accurate way to capture exact colours because it suffers from inaccuracies as a result of metamerism. Metamers are two colours that look alike but have different spectrums. The problem is that these colours, depending on the light source, can look either very similar or very distinct.

'With an RGB camera you don't have exact colour values because RGB technology works very differently to the human eye,' said Ronald Post, head of development at ColorAIXperts. 'With our camera we can capture the whole spectrum of every pixel in the image. That's the main difference.'

The ColarAIXperts system takes 16 images of a colour sample with a grey scale camera through different filters so the software can make images with different wavelengths of light. The image is then reconstructed and displayed on a computer.

The researchers used the software to register each of their colour samples from several different viewing angles. They systematically tried out various combinations of angle and illumination in a car manufacturer's colour — matching both — and chose seven that clearly characterise each shade.

After this, the spectral data can be electronically forwarded, converted back into coloured images and then compared with the manufacturer's colour specifications.

The researchers tested their software with 47 popular vehicle paint colours to prove their system is suitable for use in the car industry. They also brought in a group of experienced colourists to compare four samples of each particular shade against a reference standard — using both colour chips in the conventional way and spectral data on the computer monitor.

Throughout this testing the colour matching systems proved to be consistent. The colourists reached the same decision as the software in 82 per cent of all cases.