LUMINOSITY AND CONTRAST ADJUSTMENT OF FUNDUS IMAGES WITH REFLECTION

Abstract

This thesis presents an automatic correction method for luminosity and contrast variations in fundus images. One hundred retina or fundus images with various levels of exposure are selected from online databases and used to assess the effectiveness of the proposed method. There are four stages in the approach, and they are preprocessing, lowpass filtering, luminosity, and contrast adjustment and postprocessing. First, a color fundus image is read as input, and its three-color components, red (R), green (G) and blue (B), are separated into different channels or arrays. Next, the eye region or the region of interest (ROI), is identified along with its border via thresholding. After that, the original ratios of red-to-green and blue-to-green for every pixel in the ROI are computed and stored. Then, the ROI for the three channels is subjected to lowpass filtering, using one and two-dimensional inverse distance filter, to create a smooth background luminosity surface. This surface does not contain foreground objects such as blood vessels, optic discs, lesions, microaneurysms and others. The outcome is a smooth luminosity surface that estimates the luminosity surface or background brightness of the entire ROI. Once the background brightness is established, the luminosity of all pixels in the ROI is equalized, such that every pixel will have the same background brightness. Next, the contrast and stability of the green channel is further enhanced by adding details from the blue and red channels. Afterward, the histogram of the green channel is stretched using CLAHE to improve the contrast between the foreground objects and the background. Finally, in the post-filtering stage, the intensities of the blue and red channels are adjusted according to their original ratios to the green channel. When all three channels are recombined, the resulting color image looks similar to the original image but shows improved luminosity and contrast. The performance of the method is compared to a few other methods of similar complexity.