The need for efficient image smoothing has grown nowadays. Patch-based processing is very popular in recent years. In the existing work, an image, which is corrupted by a single noise, is the input and by using a filter it produces a clean image. This paper proposes a new image smoothing method. The input is the clean image where it extracts the patches. Adding various types of noise to an image and by applying different filters to the noisy patches will smooth the patches. Then, reorder those patches into its original position. This attains a smooth image. This proposed approach explores the use of adding discrete levels of noise in an input image and calculates the peak signal-to-noise ratio, which measures the quality of an image and shows a promising result.