An Adaptive Enhancement method for Low Contrast Color Retinal Images based on Strucural Similarity

Abstract

In this paper, an adaptive luminosity and contrast adjustment technique is proposed for the improved visual perception of low contrast color retinal images. Luminosity improvement is achieved using adaptive gamma correction based on mean structural similarity index maximization performed on the luminosity channel of the original image in Hue, Saturation and Value (HSV) color space. Further, a local contrast enhancement technique is applied on the low frequency component obtained by performing discrete wavelet transform on the enhanced luminosity channel. The experimental results reveal that the image attributes are clearly defined and a better visualization of retinal defects is achieved. Quantitative evaluation based on peak signal to noise ratio, discrete entropy and Structural Similarity Index (SSIM) shows that the proposed method performs better than the existing methods considered. For comparison of results, 128 images from the proprietary database are considered.