Image restoration is a fundamental task in computer vision that aims to recover high-quality images from degraded inputs affected by noise, blur, compression artifacts, and missing information. Traditional restoration methods often struggle to preserve fine textures and structural details, particularly under severe degradation. This paper proposes a Generative Adversarial Network (GAN)-based image restoration framework that effectively reconstructs visually realistic and high-fidelity images. The proposed model employs a generator network to restore degraded images while a discriminator network distinguishes restored images from real images, enabling adversarial learning for enhanced visual quality. In addition to adversarial loss, pixel-wise reconstruction and perceptual losses are incorporated to improve structural consistency and preserve image details. The model is trained on paired degraded and ground-truth images using extensive data augmentation techniques to enhance robustness and generalization. Experimental results demonstrate that the proposed GAN-based approach achieves superior restoration performance compared with conventional image restoration methods, producing higher Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), and improved perceptual quality. The proposed framework effectively restores fine textures, sharp edges, and natural image appearance, making it suitable for applications in medical imaging, remote sensing, surveillance, and digital photography.