APSIPA Transactions on Signal and Information Processing > Vol 13 > Issue 3

Oriented Ship Detection Based on Coordinate System Projection in SAR Images

Jiangtao Wang, School of Physics and Electronic Information and School of Information, Huaibei Normal University, and Anhui Province Key Laboratory of Intelligent Computing and Applications, China, jiangtaowang@chnu.edu.cn , Mingyang Wang, School of Physics and Electronic Information, Huaibei Normal University, China
 
Suggested Citation
Jiangtao Wang and Mingyang Wang (2024), "Oriented Ship Detection Based on Coordinate System Projection in SAR Images", APSIPA Transactions on Signal and Information Processing: Vol. 13: No. 3, e201. http://dx.doi.org/10.1561/116.00000141

Publication Date: 22 Apr 2024
© 2024 J. Wang and M. Wang
 
Subjects
 
Keywords
SAR imagesship detectioncoordinate system projectiondeep learning
 

Share

Open Access

This is published under the terms of CC BY-NC.

Downloaded: 281 times

In this article:
Introduction 
Related Works 
The Proposed Methods 
Results 
Conclusions 
References 

Abstract

Ship detection is a critical and challenging task in aerial images. Due to the special generation method of SAR images, they have unique characteristics. However, different from objects in natural images, ships in SAR images are often distributed with arbitrary orientations and dense distributions. Recently, key points-based anchor-free object detection algorithms have attracted the attention of quite a few researchers. To solve the task of oriented ship detection based on key points, in this paper, we propose an oriented object detection method based on coordinate system projection (CSProjection). In this work, we first detect the key point of the ship, namely, the center point, then establish a coordinate system with the object center point as the base point, and obtain a bounding box of the oriented object through the projection information of the object. Our method can effectively reduce the number of parameters applied to determine the oriented bounding box during training and decrease the network complexity. Experimental results on several SAR ship detection datasets, including SSDD, SRSDD-v1.0 and the optical remote sensing dataset HRSC2016, indicate that our method can compete with state-of-the-art algorithms for oriented ship detection, even those with more complex backbones.

DOI:10.1561/116.00000141

Companion

APSIPA Transactions on Signal and Information Processing Special Issue - Advanced Machine Learning Techniques for Remote Sensing: Algorithms and Applications
See the other articles that are part of this special issue.