TOWARD A VALID IMAGE PROCESSING SYSTEM THROUGH COLOUR STANDARDS

After 30 years of development in image processing and in parallel in colour sciences, the two communities have few common works. In image processing domain, colour information is classically considered as vector information in vector space, where addition and scalar multiplication are valid. To overpass this actual limit, we show how to construct an image processing framework which is valid upon the colour standards. We chose to build the approach on the mathematical morphology formalism. Our formalism is based on standard colour distance functions and notion of convergence coordinates. In this paper, we explain the theoretical construction and show some results on different applications. Then we conclude on the formalism generalization for spectral images and more complex distances functions taking into account the local neighbourhood.