The main goal of this paper is to define a new colour space that represents the colour information of the image in such a way to give a more coherent spatio-chromatic representation. This space can allow to improve the performance of the algorithms of blob detection. To build the space we base colour representation on the ridges of the colour image distribution since it has been proved that they capture the essential colours of the image. Then we will define a colour space where each channel depends on one of the ridges. Finally, to select the essential channels we apply a Constraint Satisfaction algorithm that allows to get a reduced number of channels minimizing the correlation between them.