An Euler embedding and complementary feature modeling framework for hyperspectral change detection in coastal wetlands

doi:10.11947/j.AGCS.2026.20250409

Abstract

Abstract:

Coastal wetlands are vital ecosystems for maintaining ecological balance and mitigating climate change, and their dynamic monitoring relies heavily on remote sensing technologies. With abundant spectral information, hyperspectral remote sensing can accurately identify subtle differences in wetland surface features, providing essential data support for ecosystem protection and resource management. However, existing hyperspectral change detection methods based on convolutional neural networks are limited by their local receptive fields, making it difficult to capture long-range dependencies effectively. Although Transformer-based methods possess global modeling capability, they often fail to adequately distinguish between changed and unchanged information in change detection tasks, thereby constraining overall detection performance. To address this issue, this paper proposes an Euler embedding and complementary feature modeling method (EECFM) for hyperspectral change detection in coastal wetlands. The proposed method consists of three core modules. First, an Euler-transformation-based spatial-spectral feature extraction module (ETSS) jointly models images from horizontal, vertical, and spectral channel dimensions, fully exploiting the complementarity between spatial structures and spectral information. Second, a bi-temporal image similarity enhancement module (BSE) employs cosine similarity to extract invariant features, while a Scharr-based differential attention module (SDA) leverages Scharr operators in both horizontal and vertical directions to capture fine-grained differences, thereby improving the discriminability of change regions. Finally, a multi-stage loss function is designed: in the initial stage, cross-entropy is adopted to ensure stable convergence, and in the subsequent stage, a class-reweighted Focal loss is integrated with cross-entropy to enhance robustness to long-tailed classes and improve sensitivity to change regions. The experimental results on three representative wetland remote-sensing datasets demonstrate that EECFM delivers superior accuracy and robustness in change detection, providing a technical pathway for dynamic wetland monitoring under complex environmental conditions.

Key words: coastal wetland remote sensing, change detection, hyperspectral image, Scharr operator, complementary feature modeling

CLC Number:

P237

Lanxin WU, Jiangtao PENG, Weiwei SUN, Bing YANG. An Euler embedding and complementary feature modeling framework for hyperspectral change detection in coastal wetlands[J]. Acta Geodaetica et Cartographica Sinica, 2026, 55(4): 618-631.

Figures/Tables 15

Fig. 1

Fig. 2

Tab. 1

Fig. 3

Fig. 4

Fig. 5

Tab. 2

Fig. 6

Tab. 3

Fig. 7

Tab. 4

Fig. 8

Tab. 5

Tab. 6

Tab. 7

References 44

[1]	王宗明, 颜洋洋, 赵传朋, 等. 滨海湿地遥感研究进展与展望[J]. 遥感学报, 2025, 29(6): 1938-1962.
	WANG Zongming, YAN Yangyang, ZHAO Chuanpeng, et al. Advances and perspectives in coastal wetland remote sensing research[J]. National Remote Sensing Bulletin, 2025, 29(6): 1938-1962.
[2]	HAN Zheng, LING Ziyan, DONG Li, et al. Heterogeneity effect of human disturbances on landscape patterns in the Yellow River Delta wetland, China[J]. Journal of Geodesy and Geoinformation Science, 2024, 7(4): 75-93.
[3]	薛曌娜, 王毅勇, 霍轩琳, 等. 小微湿地遥感分类和解译研究进展[J]. 遥感信息, 2025, 40(2): 1-10.
	XUE Zhaona, WANG Yiyong, HUO Xuanlin, et al. Research progress on remote sensing classification and interpretation of small and micro wetlands[J]. Remote Sensing Information, 2025, 40(2): 1-10.
[4]	顾容, 张东, 钱林峰, 等. GEE平台下考虑潮位变化及植被物候特征的盐城滨海湿地精细化遥感分类[J]. 海洋学报, 2024, 46(5): 103-115.
	GU Rong, ZHANG Dong, QIAN Linfeng, et al. Refined remote sensing classification of Yancheng coastal wetland considering tide-level changes and vegetation phenological characteristics on the GEE platform[J]. Acta Oceanologica Sinica, 2024, 46(5): 103-115.
[5]	杨漫, 刘恩勤, 杨扬, 等. 基于Landsat时间序列遥感影像的若尔盖高原湿地演变特征及其驱动因子定量分析[J]. 环境科学, 2025, 46(7): 4461-4472.
	YANG Man, LIU Enqin, YANG Yang, et al. Quantitative analysis of wetland evolution characteristics and driving factors in Ruoergai Plateau based on landsat time series remote sensing images[J]. Environmental Science, 2025, 46(7): 4461-4472.
[6]	YANG Bing, PAN Shirui, SUN Weiwei, et al. FCFDA: fine-coarse-fine progressive graph framework with distribution alignment for hyperspectral image change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 5538714.
[7]	QUINTERO-CAMPOS P, SALVADOR-CLAVELL R, MARTÍN B, et al. Environmental monitoring of a climate change indicator (Vibrio vulnificus) in coastal wetland water samples based on field-deployable detection[J]. Science of the Total Environment, 2025, 986: 179791.
[8]	杜培军, 方宏, 郭山川, 等. 深度学习遥感变化检测研究进展：像素—对象—场景[J]. 遥感技术与应用, 2025, 40(4): 783-794.
	DU Peijun, FANG Hong, GUO Shanchuan, et al. Research progress of remote sensing change detection based on deep learning: pixel-level, object-level, and scene-level[J]. Remote Sensing Technology and Application, 2025, 40(4): 783-794.
[9]	ZHANG Xinyu, CHEN Haihua, ZHAO Yanzhi, et al. Change detection of buildings in remote sensing images using a spatially and contextually aware Siamese network[J]. Expert Systems with Applications, 2025, 276: 127110.
[10]	LIU Wensong, JI Xinyuan, LIU Jie, et al. A novel unsupervised change detection method with structure consistency and GFLICM based on UAV images[J]. Journal of Geodesy and Geoinformation Science, 2022, 5(1): 91-102.
[11]	张继贤, 顾海燕, 倪欢, 等. 遥感智能变化检测的深度学习方法：演变与发展趋势[J]. 测绘学报, 2025, 54(8): 1347-1370. DOI: . doi: 10.11947/j.AGCS.2025.20240417
	ZHANG Jixian, GU Haiyan, NI Huan, et al. Deep learning methods for remote sensing intelligent change detection: evolution and development[J]. Acta Geodaetica et Cartographica Sinica, 2025, 54(8): 1347-1370. DOI: . doi: 10.11947/j.AGCS.2025.20240417
[12]	PENG Jiangtao, SUN Weiwei, LI Hengchao, et al. Low-rank and sparse representation for hyperspectral image processing: a review[J]. IEEE Geoscience and Remote Sensing Magazine, 2022, 10(1): 10-43.
[13]	CHEN Yurong, WANG Yaonan, WANG Xiaodong, et al. LCTC: lightweight convolutional thresholding sparse coding network prior for compressive hyperspectral imaging[J]. IEEE Transactions on Image Processing, 2025, 34: 4286-4301.
[14]	KıLıÇ D K, NIELSEN P. Comparative analyses of unsupervised PCA K-means change detection algorithm from the viewpoint of follow-up plan[J]. Sensors, 2022, 22(23): 9172.
[15]	赵金奇, 李宇轩, 刘子蓉, 等. 基于相似性衡量函数优化的SAR时空极化信息一体化洪涝变化检测方法[J]. 测绘学报, 2024, 53(12): 2375-2390. DOI: . doi: 10.11947/j.AGCS.2024.20230355
	ZHAO Jinqi, LI Yuxuan, LIU Zirong, et al. Flood change detection method using optimized similarity measurement function with temporal-spatial-polarized SAR information[J]. Acta Geodaetica et Cartographica Sinica, 2024, 53(12): 2375-2390. DOI: . doi: 10.11947/j.AGCS.2024.20230355
[16]	BOVOLO F, BRUZZONE L. A theoretical framework for unsupervised change detection based on change vector analysis in the polar domain[J]. IEEE Transactions on Geoscience and Remote Sensing, 2007, 45(1): 218-236.
[17]	NIELSEN A A. The regularized iteratively reweighted MAD method for change detection in multi-and hyperspectral data[J]. IEEE Transactions on Image Processing, 2007, 16(2): 463-478.
[18]	CELIK T. Unsupervised change detection in satellite images using principal component analysis and K-means clustering[J]. IEEE Geoscience and Remote Sensing Letters, 2009, 6(4): 772-776.
[19]	BOVOLO F, MARCHESI S, BRUZZONE L. A framework for automatic and unsupervised detection of multiplechanges in multitemporal images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2012, 50(6): 2196-2212.
[20]	BI Yaxin, CHRISTODOULOU V, WILKIE G, et al. Automatic anomaly detection for Swarm observations[J]. Journal of Geodesy and Geoinformation Science, 2021, 4(1): 94-108.
[21]	李相潭, 高峰, 孙悦, 等. 针对高光谱遥感图像变化检测的混合注意力和双向门控网络[J]. 中国图象图形学报, 2025, 30(4): 1017-1026.
	LI Xiangtan, GAO Feng, SUN Yue, et al. Hybrid attention and bidirectional-gated network for hyperspectral image change detection[J]. Journal of Image and Graphics, 2025, 30(4): 1017-1026.
[22]	MOU Lichao, BRUZZONE L, ZHU Xiaoxiang. Learning spectral-spatial-temporal features via a recurrent convolutional neural network for change detection in multispectral imagery[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(2): 924-935.
[23]	LIN Yun, LI Shutao, FANG Leyuan, et al. Multispectral change detection with bilinear convolutional neural networks[J]. IEEE Geoscience and Remote Sensing Letters, 2020, 17(10): 1757-1761.
[24]	WU Lanxin, MA Lei, PENG Jiangtao, et al. DIEFEN: differential information-enhanced feature exchange network for hyperspectral change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 5533312.
[25]	董子博, 王竞雪, 卜丽静, 等. MAFNet：基于多尺度空洞融合网络的遥感影像建筑物提取方法[J]. 测绘学报, 2025, 54(6): 1094-1106. DOI: . doi: 10.11947/j.AGCS.2025.20240439
	DONG Zibo, WANG Jingxue, BU Lijing, et al. MAFNet: building extraction method from remote sensing images based on multiscale atrous fusion network[J]. Acta Geodaetica et Cartographica Sinica, 2025, 54(6): 1094-1106. DOI: . doi: 10.11947/j.AGCS.2025.20240439
[26]	SONG A, CHOI J, HAN Y, et al. Change detection in hyperspectral images using recurrent 3D fully convolutional networks[J]. Remote Sensing, 2018, 10(11): 1827.
[27]	YANG Bing, SUN Weiwei, PENG Jiangtao. SAGN: sharpening-aware graph network for hyperspectral image change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 5518812.
[28]	王超, 陈天宇, 张同, 等. 基于全局差分增强模块和平衡惩罚损失的多源光学遥感影像变化检测[J]. 测绘学报, 2025, 54(5): 873-887. DOI: . doi: 10.11947/j.AGCS.2025.20240300
	WANG Chao, CHEN Tianyu, ZHANG Tong, et al. Multi-sensor optical remote sensing image change detection based on global differential enhancement module and balance penalty loss[J]. Acta Geodaetica et Cartographica Sinica, 2025, 54(5): 873-887. DOI: . doi: 10.11947/j.AGCS.2025.20240300
[29]	WANG Qi, YUAN Zhenghang, DU Qian, et al. GETNET: a general end-to-end 2-D CNN framework for hyperspectral image change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(1): 3-13.
[30]	OU Xianfeng, LIU Liangzhen, TU Bing, et al. A CNN framework with slow-fast band selection and feature fusion grouping for hyperspectral image change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5524716.
[31]	SHI Changjiang, ZHANG Zhijie, ZHANG Wanchang, et al. Learning multiscale temporal-spatial-spectral features via a multipath convolutional LSTM neural network for change detection with hyperspectral images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5529816.
[32]	CHEN Hao, QI Zipeng, SHI Zhenwei. Remote sensing image change detection with transformers[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5607514.
[33]	WANG Ziyi, GAO Feng, DONG Junyu, et al. Global and local attention-based transformer for hyperspectral image change detection[J]. IEEE Geoscience and Remote Sensing Letters, 2025, 22: 5500405.
[34]	龚良雄, 李星华, 程远明, 等. 时空差异增强与自适应特征融合的轻量级遥感影像变化检测网络[J]. 测绘学报, 2025, 54(1): 136-153. DOI: . doi: 10.11947/j.AGCS.2025.20240299
	GONG Liangxiong, LI Xinghua, CHENG Yuanming, et al. A lightweight remote sensing image change detection network utilizing spatiotemporal difference enhancement and adaptive feature fusion[J]. Acta Geodaetica et Cartographica Sinica, 2025, 54(1): 136-153. DOI: . doi: 10.11947/j.AGCS.2025.20240299
[35]	WU Lanxin, PENG Jiangtao, YANG Bing, et al. CGMNet: a center-pixel and gated mechanism-based attention network for hyperspectral change detection[J]. IEEE Transactions on Image Processing, 2026, 35: 1951-1965.
[36]	刘文力, 高峰, 张浩鹏, 等. 基于通道变换和Transformer的高光谱图像变化检测方法[J]. 计算机学报, 2025, 48(4): 971-984.
	LIU Wenli, GAO Feng, ZHANG Haopeng, et al. Hyperspectral image change detection method based on channel transformation and transformer[J]. Chinese Journal of Computers, 2025, 48(4): 971-984.
[37]	WANG Yanheng, SHA Jianjun, GAO Lianru, et al. A semi-supervised domain alignment transformer for hyperspectral image change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 5525411.
[38]	ZHANG Wuxia, SU Liangxu, ZHANG Yuhang, et al. A spectrum-aware Transformer network for change detection in hyperspectral imagery[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 5518612.
[39]	ZHANG Cui, WANG Liejun, CHENG Shuli, et al. SwinSUNet: pure transformer network for remote sensing image change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5224713.
[40]	WU Lanxin, PENG Jiangtao, YANG Bing, et al. AIWSEN: adaptive information weighting and synchronized enhancement network for hyperspectral change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2025, 63: 5503412.
[41]	LI Qingyang, ZHONG Ruofei, DU Xin, et al. TransUNetCD: a hybrid transformer network for change detection in optical remote-sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5622519.
[42]	WANG Xianghai, ZHAO Keyun, ZHAO Xiaoyang, et al. TriTF: a triplet transformer framework based on parents and brother attention for hyperspectral image change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 5507213.
[43]	LUO Junshen, LI Jiahe, CHU Xinlin, et al. BTCDNet: Bayesian tile attention network for hyperspectral image change detection[J]. IEEE Geoscience and Remote Sensing Letters, 2025, 22: 5504205.
[44]	TANG Yingjie, FENG Shou, ZHAO Chunhui, et al. A semantic change detection network based on boundary detection and task interaction for high-resolution remote sensing images[J]. IEEE Transactions on Neural Networks and Learning Systems, 2025, 36(9): 17184-17198.

数据集	图像尺寸/像素	通道数	采集时间		传感器
数据集	图像尺寸/像素	通道数	图像1	图像2	传感器
USA-Hermiston	307×241	154	2004-05-01	2007-05-08	Hyperion
HZB	400×400	166	2021-01-29	2022-02-27	ZY-02D
Yancheng	512×512	166	2022-02-24	2023-01-20	ZY-02D

方法	变化类别准确率							OA	AA	κ
方法	未变化	1	2	3	4	5	6	OA	AA	κ
Re3FCN	97.69±0.45	0.00±0.00	0.00±0.00	37.61±21.83	91.65±4.58	0.00±0.00	37.96±30.71	83.26±0.34	28.61±1.95	52.21±1.53
BIT	97.37±0.74	29.41±5.64	78.45±7.40	46.62±6.36	85.55±0.59	42.89±3.37	45.21±9.37	89.68±0.55	60.80±3.09	72.28±1.72
ConvLSTM	98.45±0.53	0.00±0.00	4.52±2.97	0.00±0.00	86.56±7.32	0.00±0.00	10.68±5.34	84.49±0.77	31.71±4.86	54.45±4.30
GLAFormer	96.96±0.51	22.23±5.91	69.28±2.96	30.32±8.85	78.91±5.03	38.21±5.50	33.82±7.30	87.38±0.69	52.82±2.09	66.14±1.79
BTCDNet	97.63±0.73	0.00±0.00	23.17±13.37	10.22±8.57	84.50±6.12	0.13±0.07	14.60±10.85	83.47±1.08	30.95±5.47	47.29±5.18
BT-SCD	96.91±1.07	42.19±11.91	57.81±7.65	38.68±5.54	82.14±4.27	34.27±12.47	51.35±5.72	88.38±1.19	57.63±0.92	68.77±2.39
EECFM	97.95±0.28	43.22±4.03	80.53±2.01	54.11±3.24	91.82±0.09	55.75±7.17	60.95±0.14	92.11±0.12	69.19±0.60	78.97±0.18

方法	变化类别准确率										OA	AA	κ
方法	未变化	1	2	3	4	5	6	7	8	9	OA	AA	κ
Re3FCN	98.42±0.41	0.00±0.00	0.00±0.00	0.00±0.00	0.00±0.00	63.76±3.24	11.48±8.15	0.00±0.00	91.58±0.95	0.00±0.00	89.90±0.13	26.92±1.65	68.63±0.43
BIT	98.64±0.08	96.09±2.71	100.00±0.00	50.19±8.08	63.52±1.10	76.20±3.28	61.81±4.36	64.17±12.79	85.84±2.82	49.47±12.57	93.84±0.47	74.60±2.97	81.27±1.64
ConvLSTM	98.63±0.58	0.00±0.00	0.00±0.00	0.00±0.00	0.00±0.00	65.92±19.48	0.00±0.00	0.00±0.00	94.85±1.19	0.00±0.00	90.19±1.09	25.94±1.93	68.51±5.32
GLAFormer	96.67±0.22	87.07±8.97	77.36±6.54	27.15±9.91	39.09±6.62	69.66±2.07	49.60±1.92	36.10±1.15	84.07±3.34	39.49±13.68	90.66±0.33	60.62±1.56	72.12±1.23
BTCDNet	98.05±0.22	55.77±8.40	64.78±6.11	0.00±0.00	0.73±0.42	64.69±13.38	2.59±1.50	32.76±19.48	84.36±13.21	27.30±5.46	89.47±1.33	41.64±2.34	65.02±6.22
BT-SCD	97.84±0.03	92.67±5.28	95.27±1.33	58.67±6.88	78.31±3.93	75.01±1.38	60.16±0.43	64.32±1.82	90.61±4.19	62.36±8.40	93.89±0.24	77.53±1.13	82.15±0.89
EECFM	98.86±0.14	85.37±4.87	93.08±8.91	61.28±6.04	74.92±4.21	80.83±3.23	70.96±6.61	72.59±5.33	93.58±2.45	68.50±12.31	95.62±0.19	82.72±5.85	86.94±0.58

方法	变化类别准确率							OA	AA	κ
方法	未变化	1	2	3	4	5	6	OA	AA	κ
Re3FCN	97.20±0.12	49.39±10.51	79.56±6.01	86.50±2.82	65.79±6.99	48.90±9.12	79.36±4.45	90.06±0.56	72.33±0.23	78.71±1.12
BIT	98.68±0.22	50.82±18.11	65.86±18.46	55.24±10.48	71.83±9.02	71.19±4.22	77.99±0.67	89.19±1.25	70.24±2.94	76.82±2.63
ConvLSTM	97.91±0.67	72.98±4.98	92.48±2.94	96.89±0.40	86.05±4.93	31.80±17.26	84.04±2.54	90.40±5.57	78.79±3.63	86.83±1.14
GLAFormer	97.14±0.39	66.69±5.18	82.85±3.51	90.97±2.43	82.02±1.82	71.10±0.21	78.61±1.40	92.14±0.32	81.34±1.27	83.29±0.78
BTCDNet	97.59±0.44	51.00±6.60	83.59±3.80	83.26±5.16	61.60±15.15	3.68±1.65	80.90±6.31	89.78±1.29	65.66±3.72	77.89±2.68
BT-SCD	98.08±0.25	67.93±2.17	83.06±2.13	95.30±0.49	90.21±3.24	80.63±1.65	77.79±1.22	93.27±0.03	84.72±0.76	86.01±0.17
EECFM	98.76±0.23	79.18±1.94	90.22±1.58	96.90±0.72	85.55±3.65	87.26±4.59	88.31±1.68	95.90±0.09	90.64±0.56	91.01±0.19

组别	模块			数据集
	模块			USA-Hermiston			HZB			Yancheng
	ETSS	BSE+SDA	L	OA	AA	κ	OA	AA	κ	OA	AA	κ
1	×	×	×	88.10	43.92	65.04	93.65	67.21	80.67	94.01	86.12	88.12
2	√	×	×	90.27	62.44	73.34	94.87	80.89	84.85	95.09	85.96	89.41
3	√	×	√	90.95	65.10	75.52	94.73	75.95	84.12	95.29	89.20	89.82
4	×	√	√	89.84	59.33	72.88	94.86	76.85	93.62	95.42	89.75	90.81
5	√	√	×	91.19	66.58	76.89	95.25	81.29	85.85	95.67	90.26	90.91
6	√	√	√	92.11	69.19	78.97	95.62	82.72	96.94	95.90	90.64	91.01