Retinal imaging is the prominent technology among the medical imaging techniques for diagnosing various fundus diseases. The low-quality retinal images generate issues in computer-aided diagnosis systems and the clinical assessment of ophthalmologists. The excellent quality of retinal images is a crucial foundation for accurate diagnosis and prognosis for ophthalmologists in detecting various eye diseases. During image gathering, retinal images become distorted by salt-and-pepper and Gaussian noises. This article proposes efficient enhancement models using Identric Mean (IDM) and Identric Mean with wavelet transform (DWT-IDM) to remove noises in the retinal fundus images. The efficacy of the proposed method is evaluated over the DRIVE and DIARETDB0 datasets. The efficacy of the suggested method is investigated using the evaluation metrics PSNR and MSE. The experimental results proved that DWT-IDM, achieved the highest PSNR of 51.45 dB and 54.32 dB in the green channel for the DRIVE and DIARETDB0 datasets, respectively. Therefore, the proposed approaches outperformed state-of-the-art mean and median filters.