Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Self-Supervised Compression and Artifact Correction for Streaming Underwater Imaging Sonar

Rongsheng Qian, Chi Xu, Xiaoqiang Ma, Hao Fang, Yili Jin, William I. Atlas, Jiangchuan Liu

TL;DR

This work tackles the challenge of real-time underwater imaging sonar streaming under severe bandwidth limits and pervasive artifacts. It introduces SCOPE, a self-supervised framework that jointly compresses and corrects sonar data via Adaptive Codebook Compression (ACC), Frequency-Aware Multiscale Segmentation (FAMS), and a hedging training strategy. The approach achieves state-of-the-art artifact suppression (SSIM ≈ 0.77) at extremely low bitrates (≤ 0.0118 bpp) and runs in real time (3.1 ms encoding, 97 ms decoding), with demonstrated deployments in three rivers for salmon monitoring. The results show improved downstream detection and substantial uplink bandwidth reductions, highlighting practical impact for autonomous underwater sensing and suggesting broader applicability to other imaging modalities.

Abstract

Real-time imaging sonar is crucial for underwater monitoring where optical sensing fails, but its use is limited by low uplink bandwidth and severe sonar-specific artifacts (speckle, motion blur, reverberation, acoustic shadows) affecting up to 98% of frames. We present SCOPE, a self-supervised framework that jointly performs compression and artifact correction without clean-noise pairs or synthetic assumptions. SCOPE combines (i) Adaptive Codebook Compression (ACC), which learns frequency-encoded latent representations tailored to sonar, with (ii) Frequency-Aware Multiscale Segmentation (FAMS), which decomposes frames into low-frequency structure and sparse high-frequency dynamics while suppressing rapidly fluctuating artifacts. A hedging training strategy further guides frequency-aware learning using low-pass proxy pairs generated without labels. Evaluated on months of in-situ ARIS sonar data, SCOPE achieves a structural similarity index (SSIM) of 0.77, representing a 40% improvement over prior self-supervised denoising baselines, at bitrates down to <= 0.0118 bpp. It reduces uplink bandwidth by more than 80% while improving downstream detection. The system runs in real time, with 3.1 ms encoding on an embedded GPU and 97 ms full multi-layer decoding on the server end. SCOPE has been deployed for months in three Pacific Northwest rivers to support real-time salmon enumeration and environmental monitoring in the wild. Results demonstrate that learning frequency-structured latents enables practical, low-bitrate sonar streaming with preserved signal details under real-world deployment conditions.

Self-Supervised Compression and Artifact Correction for Streaming Underwater Imaging Sonar

TL;DR

This work tackles the challenge of real-time underwater imaging sonar streaming under severe bandwidth limits and pervasive artifacts. It introduces SCOPE, a self-supervised framework that jointly compresses and corrects sonar data via Adaptive Codebook Compression (ACC), Frequency-Aware Multiscale Segmentation (FAMS), and a hedging training strategy. The approach achieves state-of-the-art artifact suppression (SSIM ≈ 0.77) at extremely low bitrates (≤ 0.0118 bpp) and runs in real time (3.1 ms encoding, 97 ms decoding), with demonstrated deployments in three rivers for salmon monitoring. The results show improved downstream detection and substantial uplink bandwidth reductions, highlighting practical impact for autonomous underwater sensing and suggesting broader applicability to other imaging modalities.

Abstract

Real-time imaging sonar is crucial for underwater monitoring where optical sensing fails, but its use is limited by low uplink bandwidth and severe sonar-specific artifacts (speckle, motion blur, reverberation, acoustic shadows) affecting up to 98% of frames. We present SCOPE, a self-supervised framework that jointly performs compression and artifact correction without clean-noise pairs or synthetic assumptions. SCOPE combines (i) Adaptive Codebook Compression (ACC), which learns frequency-encoded latent representations tailored to sonar, with (ii) Frequency-Aware Multiscale Segmentation (FAMS), which decomposes frames into low-frequency structure and sparse high-frequency dynamics while suppressing rapidly fluctuating artifacts. A hedging training strategy further guides frequency-aware learning using low-pass proxy pairs generated without labels. Evaluated on months of in-situ ARIS sonar data, SCOPE achieves a structural similarity index (SSIM) of 0.77, representing a 40% improvement over prior self-supervised denoising baselines, at bitrates down to <= 0.0118 bpp. It reduces uplink bandwidth by more than 80% while improving downstream detection. The system runs in real time, with 3.1 ms encoding on an embedded GPU and 97 ms full multi-layer decoding on the server end. SCOPE has been deployed for months in three Pacific Northwest rivers to support real-time salmon enumeration and environmental monitoring in the wild. Results demonstrate that learning frequency-structured latents enables practical, low-bitrate sonar streaming with preserved signal details under real-world deployment conditions.

Paper Structure

This paper contains 18 sections, 6 equations, 10 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Speckle-Affected Imaging Framework
  4. Learned Image Compression (LIC)
  5. Self-Supervised Image Denoising
  6. Background and Motivations
  7. Method
  8. SCOPE Overview
  9. Adaptive Codebook Compression
  10. Frequency-Aware Multiscale Segmentation
  11. Hedging Training Strategy
  12. Experiment Results
  13. Experimental Settings
  14. Evaluation of Artifact Correction
  15. Evaluation of Compression
  16. ...and 3 more sections

Figures (10)

  • Figure 1: SCOPE is a self-supervised framework for joint compression and artifact correction of imaging sonar streaming.
  • Figure 2: Artifacts in data from underwater imaging sonar.
  • Figure 3: Echograms show high-frequency signals as dynamic patterns and low-frequency signals as consistent static regions.
  • Figure 4: The overall underwater imaging sonar image compression and artifact correction architecture of SCOPE.
  • Figure 5: Pixel-wise temporal analysis of intensity fluctuations across imaging sonar frames.
  • ...and 5 more figures