• Corpus ID: 16431768

Deep Convolution Networks for Compression Artifacts Reduction

  title={Deep Convolution Networks for Compression Artifacts Reduction},
  author={K. Yu and Chao Dong and Chen Change Loy and Xiaoou Tang},
Lossy compression introduces complex compression artifacts, particularly blocking artifacts, ringing effects and blurring. [] Key Method To meet the speed requirement of real-world applications, we further accelerate the proposed baseline model by layer decomposition and joint use of large-stride convolutional and deconvolutional layers. This also leads to a more general CNN framework that has a close relationship with the conventional Multi-Layer Perceptron (MLP). Finally, the modified network achieves a…

Machine Learning based Post Processing Artifact Reduction in HEVC Intra Coding

A new Deep learning based algorithm on SAO filtering operation is proposed that outperforms other networks in terms on PSNR and SSIM measurements on widely available benchmark video sequences and also performs an average of 4.1 % bit rate reduction as compared to HEVC baseline.

S-Net: a scalable convolutional neural network for JPEG compression artifact reduction

Experimental results indicate that the proposed S-Net CNN outperforms other CNN-based methods and achieves state-of-the-art performance.

A Comprehensive Benchmark for Single Image Compression Artifact Reduction

A systematic listing of the reviewed methods is presented based on their basic models, including architectures, multi-domain sources, signal structures, and new targeted units, and recent deep learning-based methods based on diversified evaluation measures.

Artifacts reduction for very low bitrate image compression with generative adversarial networks

A new model is proposed, specifically trained to reduce artifacts resulting from strong lossy compressions, that can be interpreted as a post-processing step that can easily be added to any compression scheme without modifying the codec.

Attention-guided Convolutional Neural Network for Lightweight JPEG Compression Artifacts Removal

This paper proposed two methods that can improve the training performance of the compact convolution network without slowing down its inference speed, and proposes Fully Expanded Block (FEB) to replace the convolutional layer in compact network.

Boosting High-Level Vision with Joint Compression Artifacts Reduction and Super-Resolution

This paper proposes a context-aware joint CAR and SR neural network (CAJNN) that integrates both local and non-local features to solve CAR andSR in one-stage and demonstrates that CAJNN can serve as an effective image preprocessing method and improve the accuracy for real-scene text recognition and the average precision for tiny face detection.

SF-CNN: A Fast Compression Artifacts Removal via Spatial-To-Frequency Convolutional Neural Networks

SF-CNN, a fast convolutional neural network structure for JPEG image compression artifacts removal that takes Spatial input and predicts residual Frequency using downsampling operations only is proposed.

Fully Convolutional Network for Removing DCT Artefacts From Images

This article proposes three models of fully convolutional networks with different configurations and examines their abilities in reducing compression artifacts, and investigates the extent to which the results are improved for models that will process the image in a similar way to the compression algorithm.

Learning Deformable and Attentive Network for image restoration




Compression Artifacts Reduction by a Deep Convolutional Network

A compact and efficient network for seamless attenuation of different compression artifacts is formulated and it is demonstrated that a deeper model can be effectively trained with the features learned in a shallow network.

Image Super-Resolution Using Deep Convolutional Networks

We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low/high-resolution images. The mapping is represented as a deep

Efficient regression priors for reducing image compression artifacts

This paper proposes an efficient novel artifact reduction algorithm based on the adjusted anchored neighborhood regression (A+), a method from image super-resolution literature that double the relative gains in PSNR when compared with the state-of-the-art methods such as SLGP, while being order(s) of magnitude faster.

Efficient Learning of Image Super-Resolution and Compression Artifact Removal with Semi-Local Gaussian Processes

This work presents an efficient semi-local approximation scheme to large-scale Gaussian processes that allows efficient learning of task-specific image enhancements from example images without reducing quality.

A Contrast Enhancement Framework with JPEG Artifacts Suppression

This work proposes a framework that suppresses compression artifacts as an integral part of the contrast enhancement procedure and shows that this approach can produce compelling results superior to those obtained by existing JPEG artifacts removal methods for several types of contrast enhancement problems.

Natural Image Denoising with Convolutional Networks

An approach to low-level vision is presented that combines the use of convolutional networks as an image processing architecture and an unsupervised learning procedure that synthesizes training samples from specific noise models to avoid computational difficulties in MRF approaches that arise from probabilistic learning and inference.

Loss-Specific Training of Non-Parametric Image Restoration Models: A New State of the Art

This work introduces a powerful non-parametric image restoration framework based on Regression Tree Fields (RTF), a densely-connected tractable conditional random field that leverages existing methods to produce an image-dependent, globally consistent prediction.

Learning a Deep Convolutional Network for Image Super-Resolution

This work proposes a deep learning method for single image super-resolution (SR) that directly learns an end-to-end mapping between the low/high-resolution images and shows that traditional sparse-coding-based SR methods can also be viewed as a deep convolutional network.

Very Deep Convolutional Networks for Large-Scale Image Recognition

This work investigates the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting using an architecture with very small convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers.

Blocking artifacts suppression in block-coded images using overcomplete wavelet representation

A noniterative, wavelet-based deblocking algorithm that can suppress both block discontinuities and ringing artifacts effectively while preserving true edges and textural information is proposed.