Comparative Study of Image Compression Techniques for Efficient Content Delivery on Social Media Platforms

Abstract

Social media sites compress billions of images every day to minimize storage space and enhance loading speed. Although image compression optimizes the delivery process, it can negatively impact image quality, and the same image may look different on different sites because of the application of different compression algorithms. Despite the prevalence of image compression, there has been limited recent research on the evaluation of modern image compression formats for social media. This paper provides a comparative study of popular image formats like JPEG, PNG, Webp, and AVIF using a variety of real-world images such as portraits, landscapes, graphics, and composite images. The study is conducted based on file size reduction, objective quality assessment (PSNR and SSIM), subjective evaluation, usability, and page loading speed. Furthermore, a convolutional neural network is used to analyze the content of the images and help in making informed decisions about image compression. A web-based application is also developed to showcase the results and help users in making informed decisions about the optimal image compression methods for efficient social media content delivery.