Loop Filtering

In-loop filtering (ILF) enhances the quality of compressed video and images by refining reconstructed frames before final output. Current research heavily focuses on replacing traditional, hand-crafted filters with learned filters, often employing convolutional neural networks (CNNs) and attention mechanisms, to achieve superior compression performance. However, the high computational cost of these neural network-based approaches is a major challenge, leading to efforts in complexity reduction through techniques like look-up tables and network pruning. These advancements aim to improve video coding efficiency and visual quality across various applications, including standard definition and 360° video formats.