A NEW CHROMATIC ENCODING FOR MACHINE VISION INVARIANT TO THE CHANGE OF ILLUMINANT

The human visual system is adapted to assign constant chromatic attrib utes to objects in spite of the alterations in spectral distribution o f the illuminant or the spatio-chromatic arrangement of the visual fie ld. This visual capacity is called colour constancy. However, artifici al vision systems do not usually have this ability. So, when these vis ion systems operate outdoors for example, chromatic codification error s are usually very significant. Basing ourselves on the theories about colour constancy in the human visual system, fundamentally in the pri nciples of the Retinex theory, in this work we have implemented an alg orithm for colour constancy. For this, we propose some new colour-desc riptors for a stimulus, practically invariant to the spectral change o f the illuminant. Finally, we check experimentally the statistical sta bility of these colour-descriptors, corresponding to standard colour s amples (Munsell chips), against a range of standard illuminants, as we ll as their validity rank.