Vector Correlation of Color Images
C.E. Moxey, Todd A. Ell, Stephen J. Sangwine · Conference on Colour in Graphics Imaging and Vision · 2002
Auto-correlation and cross-correlation have been used in signal and image processing for many years, but have only recently been applied to color images. Since the correlation of individual signals and gray-scale images yields a measure of structural similarity, in each of their dimensions, the correlation of n-dimensional signals should yield similarity measures of n-dimensional space. This paper presents, for the first time, evidence that a vector correlation peak encodes both the structural similarity and also a mapping of the vector-space rotation between two color images.