A fusion based approach for uneven illumination correction in retinal images using Gaussian Filtering method

Enhancing the poor-quality retinal images is one of the inevitable pre-processing steps in retinal image analysis for the accurate diagnosis and treatment of different eye-related issues such as diabetic retinopathy, macular degeneration, glaucoma, etc. This paper presents a systematic approach to address this challenge by proportionally adjusting the red, green, and blue channels to ensure consistent luminance correction, enhancement, and color preservation. The core of our approach involves generating a luminance scaling matrix using a Gaussian low-pass filter applied to the red channel. This matrix is then used to correct the luminance proportion in each channel, resulting in new luminance-enhanced channels. We then applied Contrast Limited Adaptive Histogram Equalization(CLAHE) for enhancing the structural properties of the luminance corrected retina image. Our approach is effectively validated on two publically available datasets Diaretdb0 and Diaretdb1. We have tested the effectiveness of our proposed method using different evaluation metrics such as entropy, standard deviation, Natural Image Quality Evaluator(NIQE) and Perception-based Image Quality Evaluator(PIQE) measures and also confirms that our approach significantly improves the quality of retinal images, making them more suitable for accurate analysis and diagnosis.