This article is based on a recent development of a generalization of the Fourier transform. A color Clifford-Fourier transform using geometric algebra is considered to process color images. This transform, defined by Batard et al., generalizes the classical Fourier transform to $L^2(R^m;R^n)$ functions. It avoids a marginal processing that is generally applied in color image processing. From this advance, a new color phase correlation is dened and appears highly relevant for several applications, specifically in the image classication field.