Enhancement of Colour Fundus Images using Histogram Matching

N.M. Salem and A.K. Nandi (UK)

Keywords

Medical image processing, retinal images, and histogram matching.

Abstract

A new method to enhance the visual appearance of blood vessels in colour fundus images is presented. This method is based on combining the information from both red and green channels of colour images instead of using the intensity information from only one of these channels. The histogram matching is used to match the histogram of the green channel with the histogram of the red channel to obtain the advantages of both channels. This method can be used to enhance the retinal images either for human interpretation or as a pre-processing step in the automatic analysis of retinal images. Although the advantages of our method can be easily discerned by comparing the three images (red, green channels, and histogram matched image) visually, we evaluate the proposed method using a database of 20 images publicly available [1] to compare the effect of using the intensity information extracted from red, green, or both channels on the detection of the blood vessel network in retinal images. Results show that the use of the histogram matched image instead of the green channel image gives better results when using the two-dimensional matched filter introduced in [2].

Important Links:

Go Back