EVALUATING LED LIGHT SOURCES BY MEANS OF TRICHROMATIC ANALYSIS USING A DIGITAL CAMERA

Abstract

This research aimed to demonstrate that digital single lens reflex cameras (DSLR) could be used as an alternative device for evaluating the quality of LED light sources in terms of color rendering performance. The methodology presented in this work is developed from the assumption that the causes for non-linear response to color signals in digital cameras produce acceptable errors in some defined color-regions across the sRGB color-space, under determined conditions. In other words, it is assumed that when pictures are taken under particular constrained conditions, the digital cameras’ behavior can be considered linear for a variable but determined number of colors within color-space volumes of the sRGB color-space, which could be properly distributed for obtaining meaningful colorimetric results of the color-shifts calculated from digital images recorded under the reference and test light sources. The demonstration for the existence of these color-space regions was carried out with 4 artificial light sources: 2 fluorescent lamps (reference source) and 2 LED bulbs (test source), 5 DSLR cameras: 2 with CCD sensors and 3 with CMOS sensors with specific settings, a black box, a 6-neutral-wheel for determining the best shooting position and a final 30-color-wheel, which has been used throughout the color selection process. The color-wheel has 24 colors and 6 neutrals for calculating the best fit functions on the red, green and blue channels of each individual captured image. The color samples selection process was carried out for minimizing the ΔE00 between spectrophotometric CIE L*a*b* and camera calculated CIE L*a*b* to less than 6 units under both types of light sources having color temperatures of 6500 Kelvin. The linearization of R, G and B values was done by using reflectance factors of the N2, N3.5, N5, N6.5, N8 and N9.5 Munsell samples at 640nm, 530nm and 480nm respectively. The average CIE color difference 2000 formula (ΔE00) between spectroradiometric CIE L*a*b* and calculated CIE L*a*b* from testing and verifying cameras is less than 2.16 for CCD sensor camera and less than 3.82 for CMOS sensor camera under 2 types of light sources. However, the difference of ΔE00 across reference and test light source for spectroradiometer and camera are more important. The differences of the color shifts between reference light source and test light source, calculated from the digital pictures, in this study were close enough to the colorimetric measured ones in order to allow the alternative selection of digital cameras over colorimetric devices in the light quality assessment process.