Performance Evaluation of Digital Image Denoising Techniques

In this study, we evaluate the performance of several digital image denoising algorithms applied to artificially corrupted images.High-quality PNG images are used as the reference baseline, to which controlled levels of synthetic noise are deliberately added.To assess the effectiveness of these techniques in restoring visual quality, we apply three denoising methods: the Median filter and Gaussian filter (both operating in the spatial domain), as well as Block-Matching and 3D Filtering (BM3D), which operates in the frequency domain.The performance of each algorithm is quantitatively measured using standard objective metrics, including Mean Squared Error (MSE), Mean Absolute Error (MAE), Signalto-Noise Ratio (SNR), and Peak Signal-to-Noise Ratio (PSNR).The results offer a comparative analysis of the strengths and limitations of different denoising approaches, with particular emphasis on their ability to reconstruct the original noise-free content.This study provides valuable insights into the selection of appropriate denoising techniques for applications in which image fidelity is critical.