An Image Quality Selection and Effective Denoising on Retinal Images Using Hybrid Approaches

Retinal image analysis has remained an essential topic of research in the last decades. Several algorithms and techniques have been developed for retinal images. Most these use benchmark datasets to evaluate performance without first exploring quality image. Hence, metrics evaluated by approaches are uncertain. In this paper, images is selected utilizing hybrid naturalness evaluator perception-based (hybrid NIQE-PIQE) approach. Here, raw input score using Hybrid NIQE-PIQE Based on value, deep learning convolutional neural network (DCNN) categorizes into low quality, medium high Then again pre-processed remove noise present The individual green channel (G-channel) extracted from RGB filtering. Moreover, modified histogram equalization homomorphic filtering (Hybrid G-MHE-HF) utilized enhanced implementation proposed scheme implemented MATLAB 2021a. method compared with other accuracy, sensitivity, specificity, precision F-score DRIMDB DRIVE datasets. scheme’s accuracy 0.9774, sensitivity 0.9562, 0.99, specificity F-measure 0.9776 dataset, respectively.