Abstract:With the benefit of deep learning techniques, recent researches have made significant progress in image compression artifacts reduction. Despite their improved performances, prevailing methods only focus on learning a mapping from the compressed image to the original one but ignore the intrinsic attributes of the given compressed images, which greatly harms the performance of downstream parsing tasks. Different from these methods, we propose to decouple the intrinsic attributes into two complementary features for artifacts reduction, i.e., the compression-insensitive features to regularize the high-level semantic representations during training and the compression-sensitive features to be aware of the compression degree. To achieve this, we first employ adversarial training to regularize the compressed and original encoded features for retaining high-level semantics, and we then develop the compression quality-aware feature encoder for compression-sensitive features. Based on these dual complementary features, we propose a Dual Awareness Guidance Network (DAGN) to utilize these awareness features as transformation guidance during the decoding phase. In our proposed DAGN, we develop a cross-feature fusion module to maintain the consistency of compression-insensitive features by fusing compression-insensitive features into the artifacts reduction baseline. Our method achieves an average 2.06 dB PSNR gains on BSD500, outperforming state-of-the-art methods, and only requires 29.7 ms to process one image on BSD500. Besides, the experimental results on LIVE1 and LIU4K also demonstrate the efficiency, effectiveness, and superiority of the proposed method in terms of quantitative metrics, visual quality, and downstream machine vision tasks.

Learned Compression Artifact Removal by Deep Residual Networks

EARN: Toward Efficient and Robust JPEG Compression Artifact Reduction

Deep Image Compression via End-to-End Learning

Sensitivity Decouple Learning for Image Compression Artifacts Reduction

Recognition-Aware Learned Image Compression

Learning Better Lossless Compression Using Lossy Compression

Better Compression With Deep Pre-Editing

Deep learning-based Edge-aware pre and post-processing methods for JPEG compressed images

JPEG Artifacts Removal Via Compression Quality Ranker-Guided Networks.

CAS-CNN: A Deep Convolutional Neural Network for Image Compression Artifact Suppression

Deep Image Compression with Residual Learning

Deep Perceptual Compression

Compressing Weight-updates for Image Artifacts Removal Neural Networks

Practical Learned Image Compression with Online Encoder Optimization

Learned Image Compression With Gaussian-Laplacian-Logistic Mixture Model and Concatenated Residual Modules

Perceptual Quality Study on Deep Learning based Image Compression

Learning a Single Model with a Wide Range of Quality Factors for JPEG Image Artifacts Removal.

Learned Image Compression for Machine Perception

Multi-Modality Deep Network for JPEG Artifacts Reduction