A new technique for combining spatial and color information in images is presented, based on a modification to conventional color histograms, to naturally and directly incorporate spatial coordinates as one more dimension. The result is a simplified representation of the original image that retains a considerable amount of relevant information for analysis and processing. As an example of the use of this type of histogram, the problem of locating the fovea in fundus images is addressed.
This repository hosts the files required for the application of the spatial color histograms to the problem of fovea localization in retinal images using the Messidor image database. The images themselves should be downloaded from the corresponding web site. The description of the methods involved in this work can be found in a paper currently under review.