Side-scan sonar object detection via adaptive frequency-domain purification and multi-scale contextual fusion

Chulong Lv

doi:10.1016/j.optlastec.2026.116138

Side-scan sonar object detection via adaptive frequency-domain purification and multi-scale contextual fusion

2026-08-11

0

PRE

AI

C

Chulong Lv

H

Heng Wu *

W

Wei Meng

C

Chunhua He

H

Huapan Xiao

DOI：10.1016/j.optlastec.2026.116138

Original

Original request for help

Save

Abstract

En 中文

• A new network addresses missed detections in noisy side-scan sonar images. • Multi-scale perception distinguishes targets from complex backgrounds. • Frequency domain processing suppresses noise to purify feature maps. • Guided feature fusion improves detection of faint and ambiguous targets. • The lightweight model achieves high accuracy with low computational cost.

Keywords:

Side-scan sonar

Object detection

Lightweight network

Frequency domain analysis

Feature fusion

Side-scan sonar object detection via adaptive frequency-domain purification and multi-scale contextual fusion

Abstract

Journal

Organization

Cited Papers

Citing Papers