A Two-Step Algorithm for Denoising Peach Tree Leaf Images

The study aims to compare the effectiveness of different denoising algorithms and filters for reducing noise in real digital images with mixed Gaussian and Poisson noise, specifically in images of peach tree leaves,and propose the solution. The general methods used in the study include a comprehensive analysis of eleven diverse noise models and twenty-three distinct denoising filters, the proposal of a novel metric function based on MSE, PSNR, and SSIM to evaluate their performance, and the proposal of a two-step procedure for reducing noise in real images using the optimal filters: the bilateral filter and the inverse harmonic mean filter. The results demonstrate the effectiveness and viability of the proposed approach, with the two-step denoising procedure achieving a denoising outcome of
$f=0.9449$
for the artificial image and successfully reducing noise in the real images of peach tree leaves while preserving the details and structure.