Spatially Self-Paced Convolutional Networks for Change Detection in Heterogeneous Images

Change detection in heterogeneous remote sensing images is a challenging problem because it is hard to make a direct comparison in the original observation spaces, and most methods rely on a set of manually labeled samples. In this article, a spatially self-paced convolutional network (SSPCN) is con...

Full description

Bibliographic Details
Main Authors: Hao Li, Maoguo Gong, Mingyang Zhang, Yue Wu
Format: Article
Language:English
Published: IEEE 2021-01-01
Series:IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Subjects:
Online Access:https://ieeexplore.ieee.org/document/9427162/
id doaj-abe93d51148444b5a738e1bdbb5625a4
record_format Article
spelling doaj-abe93d51148444b5a738e1bdbb5625a42021-06-03T23:08:12ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352021-01-01144966497910.1109/JSTARS.2021.30784379427162Spatially Self-Paced Convolutional Networks for Change Detection in Heterogeneous ImagesHao Li0https://orcid.org/0000-0002-6294-6761Maoguo Gong1https://orcid.org/0000-0002-0415-8556Mingyang Zhang2https://orcid.org/0000-0002-9768-516XYue Wu3https://orcid.org/0000-0002-3459-5079School of Electronic Engineering, Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education of China, Xidian University, Xi’an, ChinaSchool of Electronic Engineering, Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education of China, Xidian University, Xi’an, ChinaSchool of Electronic Engineering, Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education of China, Xidian University, Xi’an, ChinaSchool of Computer Science and Technology, Xidian University, Xi’an, ChinaChange detection in heterogeneous remote sensing images is a challenging problem because it is hard to make a direct comparison in the original observation spaces, and most methods rely on a set of manually labeled samples. In this article, a spatially self-paced convolutional network (SSPCN) is constructed for change detection in an unsupervised way. Self-paced learning (SPL) is incorporated into convolutional networks to dynamically select reliable samples and learn the representation of the relations between the two heterogeneous images. In the proposed method, the pseudo labels are initialized by a classification-based method, and each sample is assigned to a weight to reflect the easiness of the sample. Then, SPL is used to learn the easy samples at first and then gradually take more complex samples into account. In the training process, the sample weights are dynamically updated based on the network parameters. Finally, a binary change map is acquired based on the trained convolutional network. The proposed SSPCN has three main advantages compared to the traditional methods. First, the proposed method is robust to noisy samples because the SSPCN involves the reliable samples into training. Second, the samples have different learning rates for converging to better values, and the learning rates are dynamically changed based on the current sample weights during iterations. Finally, we take the spatial information among the samples into account for further enhancing the robustness of the proposed method. Experimental results on four pairs of heterogeneous remote sensing images confirm the effectiveness of the proposed technique.https://ieeexplore.ieee.org/document/9427162/Change detectionconvolutional neural networks (CNNs)heterogeneous imagesself-paced learning (SPL)
collection DOAJ
language English
format Article
sources DOAJ
author Hao Li
Maoguo Gong
Mingyang Zhang
Yue Wu
spellingShingle Hao Li
Maoguo Gong
Mingyang Zhang
Yue Wu
Spatially Self-Paced Convolutional Networks for Change Detection in Heterogeneous Images
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Change detection
convolutional neural networks (CNNs)
heterogeneous images
self-paced learning (SPL)
author_facet Hao Li
Maoguo Gong
Mingyang Zhang
Yue Wu
author_sort Hao Li
title Spatially Self-Paced Convolutional Networks for Change Detection in Heterogeneous Images
title_short Spatially Self-Paced Convolutional Networks for Change Detection in Heterogeneous Images
title_full Spatially Self-Paced Convolutional Networks for Change Detection in Heterogeneous Images
title_fullStr Spatially Self-Paced Convolutional Networks for Change Detection in Heterogeneous Images
title_full_unstemmed Spatially Self-Paced Convolutional Networks for Change Detection in Heterogeneous Images
title_sort spatially self-paced convolutional networks for change detection in heterogeneous images
publisher IEEE
series IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
issn 2151-1535
publishDate 2021-01-01
description Change detection in heterogeneous remote sensing images is a challenging problem because it is hard to make a direct comparison in the original observation spaces, and most methods rely on a set of manually labeled samples. In this article, a spatially self-paced convolutional network (SSPCN) is constructed for change detection in an unsupervised way. Self-paced learning (SPL) is incorporated into convolutional networks to dynamically select reliable samples and learn the representation of the relations between the two heterogeneous images. In the proposed method, the pseudo labels are initialized by a classification-based method, and each sample is assigned to a weight to reflect the easiness of the sample. Then, SPL is used to learn the easy samples at first and then gradually take more complex samples into account. In the training process, the sample weights are dynamically updated based on the network parameters. Finally, a binary change map is acquired based on the trained convolutional network. The proposed SSPCN has three main advantages compared to the traditional methods. First, the proposed method is robust to noisy samples because the SSPCN involves the reliable samples into training. Second, the samples have different learning rates for converging to better values, and the learning rates are dynamically changed based on the current sample weights during iterations. Finally, we take the spatial information among the samples into account for further enhancing the robustness of the proposed method. Experimental results on four pairs of heterogeneous remote sensing images confirm the effectiveness of the proposed technique.
topic Change detection
convolutional neural networks (CNNs)
heterogeneous images
self-paced learning (SPL)
url https://ieeexplore.ieee.org/document/9427162/
work_keys_str_mv AT haoli spatiallyselfpacedconvolutionalnetworksforchangedetectioninheterogeneousimages
AT maoguogong spatiallyselfpacedconvolutionalnetworksforchangedetectioninheterogeneousimages
AT mingyangzhang spatiallyselfpacedconvolutionalnetworksforchangedetectioninheterogeneousimages
AT yuewu spatiallyselfpacedconvolutionalnetworksforchangedetectioninheterogeneousimages
_version_ 1721398596182474752