Color Retinal Image Enhancement using Exposure Fusion Framework

Color retinal image quality is an important parameter for ophthalmologists and computer-aided diagnosis. Thus, this study tries to enhance its quality using the Exposure Fusion Framework (EFF). The EFF is a new image enhancement technique that is introduced by Ying et al. in 2017. This enhancement technique will be compared with the Contrast-Limited Adaptive Histogram Equalization (CLAHE). The enhancement performance is assessed using extracted Retinal Blood Vessel (RBV) images. In this study, the Coye algorithm is used to extract the RBV. The Peak Signal-to-Noise Ratio (PSNR) value of the extracted RBV is used to assess the enhancement performance. Furthermore, this study utilized Structural Similarity Index (SSIM) as the second metric. The average values of PNSR and SSIM of the CLAHE-enhanced image about 59 dB and 0.993. Furthermore, the average values for the EFF-enhanced image are about 61 dB and 0.996. In the qualitative assessment, the EFF enhancement technique performs better than CLAHE. In general, the EFF enhancement technique performs better than the CLAHE. Besides, EFF requires a much longer time to process the image, around 120 times. The EFF can be used to improve retinal-based eye disease detection. Particularly in computer-assisted diagnostic systems.