Abstract:Sparse representation (SR)-based hyperspectral image (HSI) denoising methods normally average the local denoising results of multiple overlapped cubes to recover the whole HSI. Though interpretable, they rely on cumbersome hyperparameter settings and ignore the relationship between overlapped cubes, leading to poor denoising performance. This article combines SR and convolutional neural networks and introduces a deep parameterized sparse neural network (DPNet-S) to address the above issues. DPNet-S parameterizes the SR-based HSI denoising model with two modules: 1) sparse optimizer to extract sparse feature maps from noisy HSIs via recurrent usage of convolution, deconvolution, and soft shrinkage operations; and 2) image reconstructor to recover the denoised HSI from its sparse feature maps via deconvolution operations. We further replace the soft shrinkage operator with U-Net architecture to account for general HSI priors and more effectively capture the complex structures of HSIs, resulting in DPNet-U. Both networks directly learn the parameters from data and perform denoising on the whole HSI, which overcomes the limitations of SR-based methods. Moreover, our networks are generated from the denoising model and optimization procedures, thus leveraging the knowledge embedded and relying less on the number of training samples. Extensive experiments on both synthetic and real-world HSIs show that our DPNet-S and DPNet-U achieve remarkable results when compared with state-of-the-art methods. The codes will be publicly available at https://github.com/bearshng/dpnets for reproducible research.

Monte Carlo Denoising with a Sparse Auxiliary Feature Encoder

Spatial-Adaptive Network for Single Image Denoising

Adversarial Monte Carlo denoising with conditioned auxiliary feature modulation

Real-time Monte Carlo Denoising with Weight Sharing Kernel Prediction Network

A detail preserving neural network model for Monte Carlo denoising

Deep Parameterized Neural Networks for Hyperspectral Image Denoising

Pixel-wise Guidance for Utilizing Auxiliary Features in Monte Carlo Denoising

DEMC: A Deep Dual-Encoder Network for Denoising Monte Carlo Rendering

Fast Reconstruction for Monte Carlo Rendering Using Deep Convolutional Networks

Multi-Scale and Kernel-Predicting Convolutional Networks for Monte Carlo Denoising.

DMCR-GAN: Adversarial Denoising for Monte Carlo Renderings with Residual Attention Networks and Hierarchical Features Modulation of Auxiliary Buffers

Real-Time Monte Carlo Denoising With Adaptive Fusion Network

Two-Stage Monte Carlo Denoising with Adaptive Sampling and Kernel Pool

Monte Carlo Denoising Via Multi-scale Auxiliary Feature Fusion Guided Transformer.

Path‐based Monte Carlo Denoising Using a Three‐Scale Neural Network

Denoising Monte Carlo renderings via a multi-scale featured dual-residual GAN

Spatial-spectral convolutional sparse neural network for hyperspectral image denoising

Deep residual learning for denoising Monte Carlo renderings

Multitask Sparse Representation Model-Inspired Network for Hyperspectral Image Denoising

Elastic Net with Adaptive Weight for Image Denoising

RL-based Stateful Neural Adaptive Sampling and Denoising for Real-Time Path Tracing