Image Denoising with Graph-Convolutional Neural Networks

Valsesia, Diego; Fracastoro, Giulia; Magli, Enrico

doi:10.1109/icip.2019.8803367

Cited by 68 publications

(48 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Instead, we are interested in assessing how a baseline design inspired by recent results in the literature can already show that the proposed approach is competitive. Further optimization is certainly possible, e.g., by exploiting non-local features [39], [40]. However, this further strengthens the main point of this paper, which is about showing that coupling a simpler on-board compressor with a CNN at the ground segment allows higher throughput and has competitive rate-distortion performance with respect to the lossy CCSDS 123.0-B-2 standard.…”

Section: A Proposed Cnnsupporting

confidence: 63%

High-Throughput Onboard Hyperspectral Image Compression With Ground-Based CNN Reconstruction

Valsesia

Magli

2019

IEEE Trans. Geosci. Remote Sensing

Self Cite

View full text Add to dashboard Cite

Compression of hyperspectral images onboard of spacecrafts is a tradeoff between the limited computational resources and the ever-growing spatial and spectral resolution of the optical instruments. As such, it requires low-complexity algorithms with good rate-distortion performance and high throughput. In recent years, the Consultative Committee for Space Data Systems (CCSDS) has focused on lossless and near-lossless compression approaches based on predictive coding, resulting in the recently published CCSDS 123.0-B-2 recommended standard. While the in-loop reconstruction of quantized prediction residuals provides excellent rate-distortion performance for the nearlossless operating mode, it significantly constrains the achievable throughput due to data dependencies. In this paper, we study the performance of a faster method based on prequantization of the image followed by a lossless predictive compressor. While this is well known to be suboptimal, one can exploit powerful signal models to reconstruct the image at the ground segment, recovering part of the suboptimality. In particular, we show that convolutional neural networks can be used for this task and that they can recover the whole SNR drop incurred at a bitrate of 2 bits per pixel.

show abstract

Section: A Proposed Cnnsupporting

confidence: 63%

High-Throughput Onboard Hyperspectral Image Compression With Ground-Based CNN Reconstruction

Valsesia

Magli

2019

IEEE Trans. Geosci. Remote Sensing

Self Cite

View full text Add to dashboard Cite

show abstract

“…Second, in addition to additive and multiplicative operations, more complex relationships can be exploited for better performance between the features and the filter parameters. Finally, how to achieve comparable quality for path tracing with "light-weight" learning is worth exploring, as generating noise-free ground truth on a large scale is rather expensive [36].…”

Section: Extension and Discussionmentioning

confidence: 99%

“…As for denoising the path tracing results, Bako et al [5] used convolutional neural networks to infer not just the filter weights but also the form of a more complex filter kernel itself. Lehtinen et al [36] proposed a strategy which was based on the noisy data only. But the denoise process needed multiple images with different sampling rate, which took much more time.…”

Section: Related Workmentioning

confidence: 99%

Path Tracing Denoising Based on SURE Adaptive Sampling and Neural Network

Xing

Chen

2020

IEEE Access

View full text Add to dashboard Cite

A novel reconstruction algorithm is presented to address the noise artifacts of path tracing. SURE (Stein's unbiased risk estimator) is adopted to estimate the noise level per pixel that guides adaptive sampling process. Modified MLPs (multilayer perceptron) network is used to predict the optimal reconstruction parameters. In sampling stage, coarse samples are firstly generated. Then each noise level is estimated with SURE. Additional samples are distributed to the pixels with high noise level. Next, we extract a few features from the results of adaptive sampling used for the subsequent reconstruction stage. In reconstruction stage, modified MLPs network is adopted to model a complex relationship between extracted features and optimal reconstruction parameters. An anisotropic filter is used to reconstruct the final images with the parameters predicted by neural networks. Compared to the state-of-the-art methods, experiment results demonstrate that our algorithm performs better than other methods in numerical error and visual image quality.

show abstract

“…As a result, to improve the performance of graph-based tasks, it is crucial and appealing to extend the convolution network to the graph data. Recently, there have been many attempts to extend convolutions to graph data [6]- [10].…”

Section: Introductionmentioning

confidence: 99%

Neighborhood Adaptive Graph Convolutional Network for Node Classification

Gong

2019

IEEE Access

View full text Add to dashboard Cite

Recently, graph convolutional neural network as an efficient and effective method has experienced significant attention and becomes the de facto method for learning node or graph representations. However, existing most methods use a fixed-order neighborhood information when integrating node representations for node classification on the graph. In this paper, we present a neighborhood adaptive graph convolutional network (NAGCN), a novel method to efficiently learn each node's representations. Particularly, we construct a convolutional kernel abstracted from the diffusion process, named as the neighborhood adaptive kernel to more precisely learn and integrate related neighborhood node information for each node. As a result, our proposed method can learn more useful information across the relevant near and distant neighbors according to the real applications. We also adopt a threshold mechanism on the constructed kernel to better reserve the most impact neighbor vertices for each node on the graph. Besides, one learnable feature refinement process is used in the model to obtain high-level node representations with sufficient expressive power. The model is also theoretically analyzed in terms of spectral convolution and message passing algorithm. Notably, extensive experiments demonstrate that our method can achieve better performance on node classification tasks compared to other related approaches.

show abstract

Image Denoising with Graph-Convolutional Neural Networks

Cited by 68 publications

References 21 publications

High-Throughput Onboard Hyperspectral Image Compression With Ground-Based CNN Reconstruction

High-Throughput Onboard Hyperspectral Image Compression With Ground-Based CNN Reconstruction

Path Tracing Denoising Based on SURE Adaptive Sampling and Neural Network

Neighborhood Adaptive Graph Convolutional Network for Node Classification

Contact Info

Product

Resources

About