Visual color inspection with a hyperspectral camera: inline application for automotive parts production

Abstract

The development of cars is becoming increasingly focused on colour, which is a major challenge for both suppliers and manufacturers. Manufacturers must establish strict colour tolerances for each coloured component and specify both colour and appearance. On the other hand, suppliers must maintain colour accuracy in part production, regardless of the type of material used, to ensure acceptance of colour by automotive manufacturers. In the colour industry, colour difference formulas are used to make unbiased judgments about whether manufactured goods should be approved or disapproved. These objective judgments are based on instrumental colour measurements that must accurately predict subjective colour variations. Hyperspectral imaging (HSI) has proven to provide helpful data for a variety of applications in numerous industries, including colour science. To measure and analyse colour differences between automotive parts, we proposed to use conventional colour science algorithms in conjunction with the Specim FX10 hyperspectral camera, which has a flexible infrastructure that can receive data from a Generic Interface for Cameras (GenICam) compatible with the Gigabyte Ethernet interface. The results showed that the use of hyperspectral imaging in production offers great potential for providing real-time colour information so that colour shifts can be detected early and costly errors in the production line can be avoided.