MAPM:PolSAR Image Classification with Masked Autoencoder Based on Position Prediction and Memory Tokens

Deep learning methods have shown significant advantages in polarimetric synthetic aperture radar (PolSAR) image classification. However, their performances rely on a large number of labeled data. To alleviate this problem, this paper proposes a PolSAR image classification method with a Masked Autoen...

Full description

Saved in:
Bibliographic Details
Main Authors: Jianlong Wang, Yingying Li, Dou Quan, Beibei Hou, Zhensong Wang, Haifeng Sima, Junding Sun
Format: Article
Language:English
Published: MDPI AG 2024-11-01
Series:Remote Sensing
Subjects:
Online Access:https://www.mdpi.com/2072-4292/16/22/4280
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850147224792596480
author Jianlong Wang
Yingying Li
Dou Quan
Beibei Hou
Zhensong Wang
Haifeng Sima
Junding Sun
author_facet Jianlong Wang
Yingying Li
Dou Quan
Beibei Hou
Zhensong Wang
Haifeng Sima
Junding Sun
author_sort Jianlong Wang
collection DOAJ
description Deep learning methods have shown significant advantages in polarimetric synthetic aperture radar (PolSAR) image classification. However, their performances rely on a large number of labeled data. To alleviate this problem, this paper proposes a PolSAR image classification method with a Masked Autoencoder based on Position prediction and Memory tokens (MAPM). First, MAPM designs a Masked Autoencoder (MAE) based on the transformer for pre-training, which can boost feature learning and improve classification results based on the number of labeled samples. Secondly, since the transformer is relatively insensitive to the order of the input tokens, a position prediction strategy is introduced in the encoder part of the MAE. It can effectively capture subtle differences and discriminate complex, blurry boundaries in PolSAR images. In the fine-tuning stage, the addition of learnable memory tokens can improve classification performance. In addition, <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mi mathvariant="italic">L</mi><mn>1</mn></msub></semantics></math></inline-formula> loss is used for MAE optimization to enhance the robustness of the model to outliers in PolSAR data. Experimental results show the effectiveness and advantages of the proposed MAPM in PolSAR image classification. Specifically, MAPM achieves performance gains of about 1% in classification accuracy compared with existing methods.
format Article
id doaj-art-15811c15a45a400cbbd97d17e6cbe2de
institution OA Journals
issn 2072-4292
language English
publishDate 2024-11-01
publisher MDPI AG
record_format Article
series Remote Sensing
spelling doaj-art-15811c15a45a400cbbd97d17e6cbe2de2025-08-20T02:27:38ZengMDPI AGRemote Sensing2072-42922024-11-011622428010.3390/rs16224280MAPM:PolSAR Image Classification with Masked Autoencoder Based on Position Prediction and Memory TokensJianlong Wang0Yingying Li1Dou Quan2Beibei Hou3Zhensong Wang4Haifeng Sima5Junding Sun6School of Computer Science and Technology, Henan Polytechnic University, Jiaozuo 454003, ChinaSchool of Computer Science and Technology, Henan Polytechnic University, Jiaozuo 454003, ChinaKey Laboratory of Intelligent Perception and Image Understanding of Ministry of Education of China, School of Artificial Intelligence, Xidian University, Xi’an 710071, ChinaSchool of Computer Science and Technology, Henan Polytechnic University, Jiaozuo 454003, ChinaSchool of Computer Science and Technology, Henan Polytechnic University, Jiaozuo 454003, ChinaSchool of Software, Henan Polytechnic University, Jiaozuo 454003, ChinaSchool of Computer Science and Technology, Henan Polytechnic University, Jiaozuo 454003, ChinaDeep learning methods have shown significant advantages in polarimetric synthetic aperture radar (PolSAR) image classification. However, their performances rely on a large number of labeled data. To alleviate this problem, this paper proposes a PolSAR image classification method with a Masked Autoencoder based on Position prediction and Memory tokens (MAPM). First, MAPM designs a Masked Autoencoder (MAE) based on the transformer for pre-training, which can boost feature learning and improve classification results based on the number of labeled samples. Secondly, since the transformer is relatively insensitive to the order of the input tokens, a position prediction strategy is introduced in the encoder part of the MAE. It can effectively capture subtle differences and discriminate complex, blurry boundaries in PolSAR images. In the fine-tuning stage, the addition of learnable memory tokens can improve classification performance. In addition, <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mi mathvariant="italic">L</mi><mn>1</mn></msub></semantics></math></inline-formula> loss is used for MAE optimization to enhance the robustness of the model to outliers in PolSAR data. Experimental results show the effectiveness and advantages of the proposed MAPM in PolSAR image classification. Specifically, MAPM achieves performance gains of about 1% in classification accuracy compared with existing methods.https://www.mdpi.com/2072-4292/16/22/4280polarimetric SARmasked autoencoderposition prediction<i>L</i><sub>1</sub> lossmemory tokens
spellingShingle Jianlong Wang
Yingying Li
Dou Quan
Beibei Hou
Zhensong Wang
Haifeng Sima
Junding Sun
MAPM:PolSAR Image Classification with Masked Autoencoder Based on Position Prediction and Memory Tokens
Remote Sensing
polarimetric SAR
masked autoencoder
position prediction
<i>L</i><sub>1</sub> loss
memory tokens
title MAPM:PolSAR Image Classification with Masked Autoencoder Based on Position Prediction and Memory Tokens
title_full MAPM:PolSAR Image Classification with Masked Autoencoder Based on Position Prediction and Memory Tokens
title_fullStr MAPM:PolSAR Image Classification with Masked Autoencoder Based on Position Prediction and Memory Tokens
title_full_unstemmed MAPM:PolSAR Image Classification with Masked Autoencoder Based on Position Prediction and Memory Tokens
title_short MAPM:PolSAR Image Classification with Masked Autoencoder Based on Position Prediction and Memory Tokens
title_sort mapm polsar image classification with masked autoencoder based on position prediction and memory tokens
topic polarimetric SAR
masked autoencoder
position prediction
<i>L</i><sub>1</sub> loss
memory tokens
url https://www.mdpi.com/2072-4292/16/22/4280
work_keys_str_mv AT jianlongwang mapmpolsarimageclassificationwithmaskedautoencoderbasedonpositionpredictionandmemorytokens
AT yingyingli mapmpolsarimageclassificationwithmaskedautoencoderbasedonpositionpredictionandmemorytokens
AT douquan mapmpolsarimageclassificationwithmaskedautoencoderbasedonpositionpredictionandmemorytokens
AT beibeihou mapmpolsarimageclassificationwithmaskedautoencoderbasedonpositionpredictionandmemorytokens
AT zhensongwang mapmpolsarimageclassificationwithmaskedautoencoderbasedonpositionpredictionandmemorytokens
AT haifengsima mapmpolsarimageclassificationwithmaskedautoencoderbasedonpositionpredictionandmemorytokens
AT jundingsun mapmpolsarimageclassificationwithmaskedautoencoderbasedonpositionpredictionandmemorytokens