In this paper, we develop a new method for extracting invariant features for textured color surfaces moving in a 3D environment. This method is based on separation between chrominance which characterizes color of the surface, and intensity which characterizes the texture aspect. Invariant features for color are modeled by the first moments of the histograms of dominant wavelength and purity factor computed in the CIE-XYZ space. Invariant criterion for texture is computed by considering the evolution of the auto correlation function of gray-level images of surfaces in movement. Its value corresponds to the sum of the coefficients of the discrete correlation matrix. The relevance of this approach was tested on a synthetic database and also on real images of colored textured planes moving in the 3D-space. The three criteria which have been computed on these images appeared to be relevant and can be exploited to extract invariant signature.
Jacques Brochard, Majdi Khoudeir, Anis Ben Slimane, "Invariant Features for Colored Textured Surfaces Moving in a 3D Environment" in Proc. IS&T CGIV 2004 Second European Conf. on Colour in Graphics, Imaging, and Vision, 2004, pp 91 - 95, https://doi.org/10.2352/CGIV.2004.2.1.art00020