Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

WaveDH: Wavelet Sub-bands Guided ConvNet for Efficient Image Dehazing

Seongmin Hwang, Daeyoung Han, Cheolkon Jung, Moongu Jeon

TL;DR

WaveDH tackles the efficiency gap in single-image dehazing by introducing a wavelet-guided ConvNet that decomposes features into low- and high-frequency sub-bands to guide downsampling and upsampling, while a frequency-aware refinement block enhances both global structure and fine details. The architecture features wavelet-guided downsampling (WaveDown), dual upsampling (WaveUP) with a fusion module, and a frequency-aware WaveBlock that refines features in a coarse-to-fine manner, using invertible wavelet transforms to maintain information and reduce computation. Empirical results on RESIDE and I-Haze show WaveDH achieves state-of-the-art or competitive performance with substantially lower parameter counts and FLOPs, and the lightweight WaveDH-Tiny variant further demonstrates strong results at a fraction of the cost. The work also includes comprehensive ablations on the proposed components and hyperparameters, and discusses practical limitations in dense real-world hazes and directions for extending the approach to other low-level vision tasks.

Abstract

The surge in interest regarding image dehazing has led to notable advancements in deep learning-based single image dehazing approaches, exhibiting impressive performance in recent studies. Despite these strides, many existing methods fall short in meeting the efficiency demands of practical applications. In this paper, we introduce WaveDH, a novel and compact ConvNet designed to address this efficiency gap in image dehazing. Our WaveDH leverages wavelet sub-bands for guided up-and-downsampling and frequency-aware feature refinement. The key idea lies in utilizing wavelet decomposition to extract low-and-high frequency components from feature levels, allowing for faster processing while upholding high-quality reconstruction. The downsampling block employs a novel squeeze-and-attention scheme to optimize the feature downsampling process in a structurally compact manner through wavelet domain learning, preserving discriminative features while discarding noise components. In our upsampling block, we introduce a dual-upsample and fusion mechanism to enhance high-frequency component awareness, aiding in the reconstruction of high-frequency details. Departing from conventional dehazing methods that treat low-and-high frequency components equally, our feature refinement block strategically processes features with a frequency-aware approach. By employing a coarse-to-fine methodology, it not only refines the details at frequency levels but also significantly optimizes computational costs. The refinement is performed in a maximum 8x downsampled feature space, striking a favorable efficiency-vs-accuracy trade-off. Extensive experiments demonstrate that our method, WaveDH, outperforms many state-of-the-art methods on several image dehazing benchmarks with significantly reduced computational costs. Our code is available at https://github.com/AwesomeHwang/WaveDH.

WaveDH: Wavelet Sub-bands Guided ConvNet for Efficient Image Dehazing

TL;DR

WaveDH tackles the efficiency gap in single-image dehazing by introducing a wavelet-guided ConvNet that decomposes features into low- and high-frequency sub-bands to guide downsampling and upsampling, while a frequency-aware refinement block enhances both global structure and fine details. The architecture features wavelet-guided downsampling (WaveDown), dual upsampling (WaveUP) with a fusion module, and a frequency-aware WaveBlock that refines features in a coarse-to-fine manner, using invertible wavelet transforms to maintain information and reduce computation. Empirical results on RESIDE and I-Haze show WaveDH achieves state-of-the-art or competitive performance with substantially lower parameter counts and FLOPs, and the lightweight WaveDH-Tiny variant further demonstrates strong results at a fraction of the cost. The work also includes comprehensive ablations on the proposed components and hyperparameters, and discusses practical limitations in dense real-world hazes and directions for extending the approach to other low-level vision tasks.

Abstract

The surge in interest regarding image dehazing has led to notable advancements in deep learning-based single image dehazing approaches, exhibiting impressive performance in recent studies. Despite these strides, many existing methods fall short in meeting the efficiency demands of practical applications. In this paper, we introduce WaveDH, a novel and compact ConvNet designed to address this efficiency gap in image dehazing. Our WaveDH leverages wavelet sub-bands for guided up-and-downsampling and frequency-aware feature refinement. The key idea lies in utilizing wavelet decomposition to extract low-and-high frequency components from feature levels, allowing for faster processing while upholding high-quality reconstruction. The downsampling block employs a novel squeeze-and-attention scheme to optimize the feature downsampling process in a structurally compact manner through wavelet domain learning, preserving discriminative features while discarding noise components. In our upsampling block, we introduce a dual-upsample and fusion mechanism to enhance high-frequency component awareness, aiding in the reconstruction of high-frequency details. Departing from conventional dehazing methods that treat low-and-high frequency components equally, our feature refinement block strategically processes features with a frequency-aware approach. By employing a coarse-to-fine methodology, it not only refines the details at frequency levels but also significantly optimizes computational costs. The refinement is performed in a maximum 8x downsampled feature space, striking a favorable efficiency-vs-accuracy trade-off. Extensive experiments demonstrate that our method, WaveDH, outperforms many state-of-the-art methods on several image dehazing benchmarks with significantly reduced computational costs. Our code is available at https://github.com/AwesomeHwang/WaveDH.
Paper Structure (24 sections, 14 equations, 9 figures, 7 tables)

This paper contains 24 sections, 14 equations, 9 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Single Image Dehazing
  4. Wavelet-based Approaches in Computer Vision
  5. Proposed Method
  6. Overall Architecture
  7. Wavelet-guided downsampling and upsampling blocks
  8. Downsampling block
  9. Upsampling block
  10. Frequency-aware feature refinement block
  11. Experiments
  12. Datasets and implementation
  13. Datasets
  14. Implementation details
  15. Ablation Studies
  16. ...and 9 more sections

Figures (9)

  • Figure 1: Comparison of WaveDH with other image dehazing methods on the SOTS indoor set RESIDE. The circle size is proportional to the number of model parameters. Note that our WaveDH achieves superior PSNR and also maintains lower model complexity.
  • Figure 2: Overview of WaveDH architecture. (a) Depicts the overall architecture consists of various modules and their contribution to efficient image dehazing. (b) Shows our downsampling block using WaveAttention. (c) Illustrates the WaveAttention module for noise suppression and detail preservation. (d) Presents the WaveBlock for frequency-aware feature learning. (e) Represents the WaveUp block based on dual-upsample and fusion mechanism. (f) Represents the fusion module crucial for interacting dual-upsampled features. (g) Presents the modified Fused-MBConv (FMBConv) employing group convolution to optimize parameter usage and efficiency.
  • Figure 3: This figure depicts the enhanced feature map via the wavelet attention mechanism. (a) Original hazy input. (b) Visualization of attention map. (c) Visualization of feature map before WaveAttention. (d) Visualization of feature map after WaveAttention. The WaveAttention refines the feature map by suppressing noise and emphasizing informative details, contributing to improved dehazing performance.
  • Figure 4: Overview of the proposed WaveBlock components. (a) The architecture of Efficient Separable Distillation Block (ESDB), employing a series of 1x1 convolutional layers with Blueprint Shallow Residual Blocks (BSRBs) for feature distillation. (b) The architecture of Feature Mixing Block (FMB), consisting of depthwise convolution (DWConv), a shuffle mixer layer, and two (c) channel projection modules.
  • Figure 5: Feature map visualization of the dual-upsample and fusion mechanism. (a) Original hazy input images. (b) Feature maps after the Inverse Discrete Wavelet Transform (IDWT) layer. (c) Feature maps after the PixelShuffle layer. (d) Feature maps after fusion module. This highlights the effectiveness of the dual-upsample and fusion mechanism.
  • ...and 4 more figures