A Preliminary Research on Lightweight approach of Noise Reduction in Image Processing: Gaussian Smoothing and Super-Resolution via Resizing

In the field of image processing, noise removal and enhancement of resolution are two of the most basic issues for enhancing the quality of an image. In the following paper, the problem of noisy images and low resolution will be solved in a fast and lightweight manner by applying adaptive smoothing and cubic interpolation, respectively. The performance of the proposed method is tested on more than 100 images containing many categories to check the effectiveness and efficiency of the method.Mean squared error was one of the measurement indices used to compare the processed photos in this study with the original dataset in order to assess the findings; mean squared error values ranging from 0.54 to 144.88 were found. The SSIM fluctuated between 0.938 and 0.998, while the PSNR ranged from 26.75 to 46.76. Overall, the findings demonstrate how well the technique works to improve the quality of photos with high PSNR and SSIM values.This work also proposes a lightweight way to track outcomes, where most of the steps can be coded to log metrics for each image and a case making the process reproducible. The approach also focuses on how traditional image processing techniques can deliver high-quality results with low computational cost. It is possible that future work investigates the use of these techniques coupled with more complex machine learning algorithms to be more suited for data handling and improving overall result processing.