Perceptual Compression

Perceptual compression aims to minimize data size while preserving the perceived quality of images and videos, prioritizing human visual perception over objective distortion metrics. Current research heavily utilizes diffusion models and generative adversarial networks (GANs), often incorporating side information like text to enhance compression and reconstruction quality. This field is significant because it addresses the growing need for efficient storage and transmission of visual data, particularly at ultra-low bitrates, with applications ranging from improved image and video codecs to enhanced multimedia experiences. Recent advancements focus on optimizing the trade-off between bitrate, distortion, and perceptual quality, often through novel loss functions and decoder architectures.