Multitask Learning-Based for SAR Image Superpixel Generation
Most of the existing synthetic aperture radar (SAR) image superpixel generation methods are designed based on the raw SAR images or artificially designed features. However, such methods have the following limitations: (1) SAR images are severely affected by speckle noise, resulting in unstable pixel...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-02-01
|
Series: | Remote Sensing |
Subjects: | |
Online Access: | https://www.mdpi.com/2072-4292/14/4/899 |
_version_ | 1797476842225008640 |
---|---|
author | Jiafei Liu Qingsong Wang Jianda Cheng Deliang Xiang Wenbo Jing |
author_facet | Jiafei Liu Qingsong Wang Jianda Cheng Deliang Xiang Wenbo Jing |
author_sort | Jiafei Liu |
collection | DOAJ |
description | Most of the existing synthetic aperture radar (SAR) image superpixel generation methods are designed based on the raw SAR images or artificially designed features. However, such methods have the following limitations: (1) SAR images are severely affected by speckle noise, resulting in unstable pixel distance estimation. (2) Artificially designed features cannot be well-adapted to complex SAR image scenes, such as the building regions. Aiming to overcome these shortcomings, we propose a multitask learning-based superpixel generation network (ML-SGN) for SAR images. ML-SGN firstly utilizes a multitask feature extractor to extract deep features, and constructs a high-dimensional feature space containing intensity information, deep semantic informantion, and spatial information. Then, we define an effective pixel distance measure based on the high-dimensional feature space. In addition, we design a differentiable soft assignment operation instead of the non-differentiable nearest neighbor operation, so that the differentiable Simple Linear Iterative Clustering (SLIC) and multitask feature extractor can be combined into an end-to-end superpixel generation network. Comprehensive evaluations are performed on two real SAR images with different bands, which demonstrate that our proposed method outperforms other state-of-the-art methods. |
first_indexed | 2024-03-09T21:09:30Z |
format | Article |
id | doaj.art-ba6670f39b564837ab261d5564d829b5 |
institution | Directory Open Access Journal |
issn | 2072-4292 |
language | English |
last_indexed | 2024-03-09T21:09:30Z |
publishDate | 2022-02-01 |
publisher | MDPI AG |
record_format | Article |
series | Remote Sensing |
spelling | doaj.art-ba6670f39b564837ab261d5564d829b52023-11-23T21:53:52ZengMDPI AGRemote Sensing2072-42922022-02-0114489910.3390/rs14040899Multitask Learning-Based for SAR Image Superpixel GenerationJiafei Liu0Qingsong Wang1Jianda Cheng2Deliang Xiang3Wenbo Jing4College of Information Science and Technology, Beijing University of Chemical Technology, Beijing 100029, ChinaSchool of Electronics and Communication Engineering, Sun Yat-sen University, Shenzhen 518107, ChinaCollege of Information Science and Technology, Beijing University of Chemical Technology, Beijing 100029, ChinaBeijing Advanced Innovation Center for Soft Matter Science and Engineering, Beijing University of Chemical Technology, Beijing 100029, ChinaCollege of Electronic Science, National University of Defense Technology, Changsha 410073, ChinaMost of the existing synthetic aperture radar (SAR) image superpixel generation methods are designed based on the raw SAR images or artificially designed features. However, such methods have the following limitations: (1) SAR images are severely affected by speckle noise, resulting in unstable pixel distance estimation. (2) Artificially designed features cannot be well-adapted to complex SAR image scenes, such as the building regions. Aiming to overcome these shortcomings, we propose a multitask learning-based superpixel generation network (ML-SGN) for SAR images. ML-SGN firstly utilizes a multitask feature extractor to extract deep features, and constructs a high-dimensional feature space containing intensity information, deep semantic informantion, and spatial information. Then, we define an effective pixel distance measure based on the high-dimensional feature space. In addition, we design a differentiable soft assignment operation instead of the non-differentiable nearest neighbor operation, so that the differentiable Simple Linear Iterative Clustering (SLIC) and multitask feature extractor can be combined into an end-to-end superpixel generation network. Comprehensive evaluations are performed on two real SAR images with different bands, which demonstrate that our proposed method outperforms other state-of-the-art methods.https://www.mdpi.com/2072-4292/14/4/899multitask learningSAR image superpixel generationhigh-dimensional feature spacepixel-superpixel soft assignment |
spellingShingle | Jiafei Liu Qingsong Wang Jianda Cheng Deliang Xiang Wenbo Jing Multitask Learning-Based for SAR Image Superpixel Generation Remote Sensing multitask learning SAR image superpixel generation high-dimensional feature space pixel-superpixel soft assignment |
title | Multitask Learning-Based for SAR Image Superpixel Generation |
title_full | Multitask Learning-Based for SAR Image Superpixel Generation |
title_fullStr | Multitask Learning-Based for SAR Image Superpixel Generation |
title_full_unstemmed | Multitask Learning-Based for SAR Image Superpixel Generation |
title_short | Multitask Learning-Based for SAR Image Superpixel Generation |
title_sort | multitask learning based for sar image superpixel generation |
topic | multitask learning SAR image superpixel generation high-dimensional feature space pixel-superpixel soft assignment |
url | https://www.mdpi.com/2072-4292/14/4/899 |
work_keys_str_mv | AT jiafeiliu multitasklearningbasedforsarimagesuperpixelgeneration AT qingsongwang multitasklearningbasedforsarimagesuperpixelgeneration AT jiandacheng multitasklearningbasedforsarimagesuperpixelgeneration AT deliangxiang multitasklearningbasedforsarimagesuperpixelgeneration AT wenbojing multitasklearningbasedforsarimagesuperpixelgeneration |