Ship Detection in Synthetic Aperture Radar Imagery Based on Discriminative Dictionary Learning

Yun Wang, Liang Chen, Hao Shi, Bocheng Zhang

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

3 Citations (Scopus)

Abstract

The ship target detection technology based on SAR image has important significance in military and civil field applications and is one of the research hotspots at this stage. In this paper, the research work on typical problems in SAR image ship target detection is carried out. A ship target detection algorithm based on discriminative dictionary learning is proposed, which mainly includes image denoising, candidate region extraction and candidate region identification. Firstly, an adaptive non-local filtering method is used to denoise the SAR image. Then the gradient feature map reconstruction algorithm is used to extract the candidate regions. Finally, the category constrained discriminative dictionary learning method is used to classify the candidate regions. The algorithm is based on GF-3 and Terra SAR data. The experimental results show that the proposed algorithm has strong robustness and adaptability.

Original languageEnglish
Title of host publication2019 6th Asia-Pacific Conference on Synthetic Aperture Radar, APSAR 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728129129
DOIs
Publication statusPublished - Nov 2019
Event6th Asia-Pacific Conference on Synthetic Aperture Radar, APSAR 2019 - Xiamen, China
Duration: 26 Nov 201929 Nov 2019

Publication series

Name2019 6th Asia-Pacific Conference on Synthetic Aperture Radar, APSAR 2019

Conference

Conference6th Asia-Pacific Conference on Synthetic Aperture Radar, APSAR 2019
Country/TerritoryChina
CityXiamen
Period26/11/1929/11/19

Keywords

  • Discriminative Dictionary Learning
  • SAR Imagery
  • Ship Detection

Fingerprint

Dive into the research topics of 'Ship Detection in Synthetic Aperture Radar Imagery Based on Discriminative Dictionary Learning'. Together they form a unique fingerprint.

Cite this