Vol. 9 pp.49 - 60.

AbstractImage enhancement plays a critical role in image processing. This paper introduces an integrated approach using three methods: Super-Resolution Generative Adversarial Net- work (SRGAN), Convolutional Autoencoder (CAE), and Zero- Reference Deep Curve Estimation (ZeroDCE), all embedded within a web application to provide image enhancement opportunities to users. SRGAN enhances image resolution by generating high-quality images from low-resolution inputs, with improved adjustments in Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) values. The Convolutional Autoencoder denoises impulse noise while preserving key image features. ZeroDCE automatically enhances low-light images by adjusting pixel values using Light-Enhancement curves (LE- curves), Deep Curve Estimation Network (DCE-Net), and Non- Reference loss functions. In conclusion, this paper contributes to the broader field of computer vision and image processing, providing academic and practical understanding of the performance and limitations of these three models, paving the way for further developments in the field. The code is available at https://github.com/robinson-pujara/Refinetograph.git.