A new 1-D colour model and its application to image filtering

This paper introduces a new 1-D model to encode coloured images for an efficient subsequent processing. This representation is equivalent to, but more compact than, the 3-D HCL conical representation. It consists in gathering all the hue, chroma and luminance information in one component, namely, the cumulative spiral angle, where the spirals in question are defined as a sampling of the solid HCL cone. We use the proposed model for joint bilateral upsampling of low-resolution depth maps. The results show that, in addition to preserving the perceptual properties of the HCL colour representation, using the proposed model leads to a solution that is more accurate than when using grayscale images.