Natural images are complex structures. Their modeling is essential for many tasks in image processing, and progress remains to be done in this direction. Properties of self-similarity in scale were observed in the spatial organization of natural images. Beyond their spatial organization, we address here their colorimetric organization. Several tools for fractal characterization are identified and implemented. Application of these tools on the three-dimensional histograms from color images allow us to reveal and characterize fractal properties in their colorimetric organization. Furthermore, we also consider, in image synthesis, new applications of fractal concepts for color images, the control of their properties and the automatic generation of decorative patterns. Applicative collaborations have been developed during the doctoral thesis in order to valorize this work, especially in the context of the "doctorant-conseil" program, and towards the industry of textile and clothing.