A deeply supervised image fusion network for change detection in high resolution bi-temporal remote sensing images

By Chenxiao Zhang, P. Yue, D. Tapete, Liangcun Jiang, Boyi Shangguan, Li Huang, Guangchao Liu. 2020

1,163 citations recorded by the source corpus.

Abstract Change detection in high resolution remote sensing images is crucial to the understanding of land surface changes. As traditional change detection methods are not suitable for the task considering the challenges brought by the fine image details and complex texture features conveyed in high resolution images, a number of deep learning-based change detection methods have been proposed to improve the change detection performance. Although the state-of-the-art deep feature based methods outperform all the other deep learning-based change detection methods, networks in the existing deep feature based methods are mostly modified from architectures that are originally proposed for single-image semantic segmentation. Transferring these networks for change detection task still poses some key issues. In this paper, we propose a deeply supervised image fusion network (IFN) for change detection in high resolution bi-temporal remote sensing images. Specifically, highly representative deep features of bi-temporal images are firstly extracted through a fully convolutional two-stream architecture. Then, the extracted deep features are fed into a deeply supervised difference discrimination network (DDN) for change detection. To improve boundary completeness and internal compactness of objects in the output change maps, multi-level deep features of raw images are fused with image difference features by means of attention modules for change map reconstruction. DDN is further enhanced by directly introducing change map losses to intermediate layers in the network, and the whole network is trained in an end-to-end manner. IFN is applied to a publicly available dataset, as well as a challenging dataset consisting of multi-source bi-temporal images from Google Earth covering different cities in China. Both visual interpretation and quantitative assessment confirm that IFN outperforms four benchmark methods derived from the literature, by returning changed areas with complete boundaries and high internal compactness compared to the state-of-the-art methods.

Subfields: Atmospheric Science, Ecology, Media Technology

View Paper: https://www.semanticscholar.org/paper/f4e61d31666087d850a9f45f187254366d0a6419

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Atmospheric Science

In Ecology

In Media Technology

Citations in the corpus

Transformers in Remote Sensing: A Survey (2022, background): “TransUNetCD [58] Change Detection WHU [60], LEVIR [61], CCD [82], DSIFN [96], OSCD [98], S2Looking [100] F1 score Introduces a framework integrating merits of transformers and UNet through capturing enriched contextualized features which are upsampled and fused with multi-scale features to generate ”
A Deeply Supervised Attention Metric-Based Network and an Open Aerial Image Dataset for Remote Sensing Change Detection (2021, background): “Previous works have demonstrated the excellent ability of deep supervision to improve the effectiveness of hidden layers [32], [53], [54]; thus, we introduce deeply supervised nets (DSNs) [31] into our network.”
Change Detection on Remote Sensing Images Using Dual-Branch Multilevel Intertemporal Network (2023, background): “Moreover, several schemes are extensively investigated and adopted to expand the scope of the receptive field, including deepened network architectures [37], dilated convolutional strategies [39], and attention mechanisms [35], [36], [37], [40].”
A Transformer-Based Siamese Network for Change Detection (2022, methodology): “We use two publically available CD datasets for our experiments, namely LEVIR-CD [10] and DSIFN-CD [11]”
Changer: Feature Interaction is What You Need for Change Detection (2022, methodology): “IFN [53] continues the design of FC-Siam-Conc on the macro level, and on the micro level, it uses a more complex backbone, attention mechanism, and adds multiple auxiliary heads for deep supervision in the decoder to speed up the convergence of the model.”
ChangeMamba: Remote Sensing Change Detection With Spatiotemporal State Space Model (2024, methodology): “Some tricks for improving accuracy like focal loss [53], deep supervision [17], multi-scale training/testing [6], etc. are not adopted.”
Towards Geospatial Foundation Models via Continual Pretraining (2023, methodology): “We conduct experiments on both the Onera Satellite Change Detection (OSCD [5]) in Table 3 and DSIFN [43] in Table 4.”
SNUNet-CD: A Densely Connected Siamese Network for Change Detection of VHR Images (2021, methodology): “In [10], the pyramid model is used for extracting multiscale features; in [9] and [12], deep super-vision is used to enhance the representation and discrimination capabilities of shallow features; and attention mechanisms are used to reﬁne features and obtain better feature representations, such as…”
Remote Sensing Image Change Detection With Transformers (2021, methodology): “Note that our CNN backbone is only the pure ResNet and we do not apply sophisticated structures such as FPN in [2] or UNet in [9], [10], [14], and [22], which are powerful for pixel-wise prediction tasks by fusing low-level features with high spatial accuracy and high-level semantic features.”