PhyUnfold-Net: Advancing Remote Sensing Change Detection with Physics-Guided Deep Unfolding

Zelin Lei; Yaoxing Ren; Jiaming Chang

cs.CVMar 20, 2026

PhyUnfold-Net: Advancing Remote Sensing Change Detection with Physics-Guided Deep Unfolding

Authors: Zelin Lei, Yaoxing Ren, Jiaming Chang

Organizations: Xi’an Jiaotong University · Anhui University

Abstract

Bi-temporal change detection is highly sensitive to acquisition discrepancies, including illumination, season, and atmosphere, which often cause false alarms. We observe that genuine changes exhibit higher patch-wise singular-value entropy (SVE) than pseudo changes in the feature-difference space. Motivated by this physical prior, we propose PhyUnfold-Net, a physics-guided deep unfolding framework that formulates change detection as an explicit decomposition problem. The proposed Iterative Change Decomposition Module (ICDM) unrolls a multi-step solver to progressively separate mixed discrepancy features into a change component and a nuisance component. To stabilize this process, we introduce a staged Exploration-and-Constraint loss (S-SEC), which encourages component separation in early steps while constraining nuisance magnitude in later steps to avoid degenerate solutions. We further design a Wavelet Spectral Suppression Module (WSSM) to suppress acquisition-induced spectral mismatch before decomposition. Experiments on four benchmarks show improvements over state-of-the-art methods, with gains under challenging conditions.

Figures & tables

Figure 1: Comparison of change-detection paradigms. (a) Cross-temporal attention for invariant feature learning. (b) Token interaction with implicit difference enhancement. (c) Explicit difference-space decomposition $D=C+N$ with unrolled refinement and separation control.

Figure 2: Overview of PhyUnfold-Net. WSSM suppresses acquisition-induced variations and produces corrected features $(\hat{F}_{1},\hat{F}_{2})$ before the difference map $D=\Delta(\hat{F}_{1},\hat{F}_{2})$ is formed. ICDM then unrolls a $K$ -step solver to refine the decomposition $D=C+N$ , and the segmentation head predicts $\hat{Y}$ from $C^{K}$ . Training-only losses are shown in the lower panels.

Figure 3: SVE evolution across U-Net layers during training at Encoder-4, Decoder-1, Decoder-2, and Decoder-3.

Figure 4: Qualitative analysis of the physical prior under vegetation and land-cover variations.

Figure 5: Robustness of the physical prior under severe illumination interference.

Method	Year	Pre.(%)	Rec.(%)	F1(%)	IoU(%)
FC-Siam-Di [ 1 ]	2018	89.53	83.31	86.31	75.92
DTCDSCN [ 30 ]	2021	88.54	86.13	87.32	77.49
STANet [ 10 ]	2020	89.06	85.80	87.40	77.67
DSIFN [ 31 ]	2020	90.15	87.03	88.56	79.52
SNUNet [ 2 ]	2022	89.18	87.17	88.16	78.83
BIT [ 14 ]	2022	89.24	89.37	89.31	80.68

Table 1: Results on LEVIR-CD . Best in bold , second-best underlined .

Method	Year	Pre.(%)	Rec.(%)	F1(%)	IoU(%)
FC-Siam-Di [ 1 ]	2018	63.79	55.12	59.14	41.99
DTCDSCN [ 30 ]	2021	66.38	60.51	63.31	46.30
STANet [ 10 ]	2020	68.72	57.43	62.58	45.53
DSIFN [ 31 ]	2020	71.05	62.87	66.71	50.06
SNUNet [ 2 ]	2022	73.91	66.23	69.86	53.69
BIT [ 14 ]	2022	73.42	68.89	71.08	55.17

Table 2: Results on LEVIR-CD+ . Best in bold , second-best underlined .

Method	Year	Pre.(%)	Rec.(%)	F1(%)	IoU(%)
FC-Siam-Di [ 1 ]	2018	47.33	81.53	59.88	42.73
DTCDSCN [ 30 ]	2021	63.92	82.14	71.91	56.15
STANet [ 10 ]	2020	79.37	78.52	78.94	65.22
DSIFN [ 31 ]	2020	83.40	80.13	81.73	69.10
SNUNet [ 2 ]	2022	85.60	87.90	86.73	76.53
BIT [ 14 ]	2022	86.64	85.38	86.00	75.44

Table 3: Results on WHU-CD . Best in bold , second-best underlined .

Method	Year	Pre.(%)	Rec.(%)	F1(%)	IoU(%)
FC-Siam-Di [ 1 ]	2018	38.25	27.64	32.09	19.11
DTCDSCN [ 30 ]	2021	42.16	31.87	36.30	22.17
STANet [ 10 ]	2020	43.58	33.25	37.72	23.24
DSIFN [ 31 ]	2020	48.33	37.42	42.17	26.72
SNUNet [ 2 ]	2022	50.92	40.18	44.91	28.92
BIT [ 14 ]	2022	52.74	42.53	47.06	30.79

Table 4: Results on S2Looking . Best in bold , second-best underlined .

#	ICDM	WSSM	S-SEC	Pre.(%)	$\Delta$ Pre	Rec.(%)	$\Delta$ Rec	F1(%)	$\Delta$ F1	IoU(%)	$\Delta$ IoU
1	✗	✗	✗	53.17	–	38.62	–	44.74	–	28.82	–
2	✓	✗	✗	57.63	+4.46	43.78	+5.16	49.76	+5.02	33.12	+4.30
3	✗	✓	✗	59.41	+6.24	45.32	+6.70	51.42	+6.68	34.61	+5.79
4	✗	✗	✓	55.86	+2.69	42.17	+3.55	48.06	+3.32	31.63	+2.81
5	✓	✓	✗	65.28	+12.11	53.74	+15.12	58.95	+14.21	41.79	+12.97
6	✓	✗	✓	62.41	+9.24	50.63	+12.01	55.91	+11.17	38.80	+9.98

Table 5: Ablation on S2Looking .

\Delta

indicates the gain over baseline (#1).

Figure 6: Training-time SVE trajectories for changed and unchanged patches on raw

D

and at ICDM steps

k{=}1,2,3

. The changed-versus-unchanged SVE gap increases from

0.13

0.36

Figure 7: Patch-grid SVE maps across ICDM steps. Changed regions become progressively more compact and exhibit higher SVE, whereas unchanged regions become more diffuse and exhibit lower SVE.

Figure 8: Effect of unrolling steps

K

on LEVIR-CD and LEVIR-CD+.

Figure 9: Qualitative decomposition at $K{=}1,3,5$ on LEVIR-CD. The result is under-decomposed at $K{=}1$ , becomes cleaner at $K{=}3$ , and shows only limited additional structural improvement at $K{=}5$ .

	LEVIR-CD		S2Looking
Stage	Score $r^{k}$	Ratio	Score $r^{k}$	Ratio
ICDM $k=1$	0.412	–	0.487	–
ICDM $k=2$	0.173	0.420	0.226	0.464
ICDM $k=3$	0.068	0.393	0.104	0.460

Table 6: Normalized mismatch score

r^{k}=\|R_{\mathrm{res}}^{k}\|_{F}/\|D\|_{F}

at ICDM steps

k\in\{1,2,3\}

. “Ratio” reports

r^{k}/r^{k-1}

for

k\geq 2

Method	Year	Pre.(%)	F1(%)	IoU(%)	FLOPs (G)
STANet [ 10 ]	2020	68.72	62.58	45.53	3706.81
DSIFN [ 31 ]	2020	71.05	66.71	50.06	2632.16
SNUNet [ 2 ]	2022	73.91	69.86	53.69	1754.66
RS-Mamba [ 15 ]	2024	79.67	80.91	67.95	827.83
ChangeDA [ 32 ]	2025	81.47	80.65	67.58	1848.49
SPMNet [ 33 ]	2025	83.59	83.09	71.01	2388.91

Table 7: Results on LEVIR-CD+ with computational complexity. Best results are in bold , and second-best results are underlined .

$m$	Pre.(%)	Rec.(%)	F1(%)	IoU(%)
0.1	67.83	61.42	64.47	47.56
0.2	70.29	62.85	66.36	49.66
0.3	71.85	63.28	67.29	50.71
0.4	71.56	61.93	66.40	49.70
0.5	69.14	60.78	64.69	47.81

Table 8: Sensitivity to separation margin

m

on S2Looking.

$\tau_{\ell}$	$\tau_{u}$	Pre.(%)	Rec.(%)	F1(%)	IoU(%)
0.01	0.20	68.97	60.35	64.37	47.46
0.03	0.30	70.18	62.54	66.14	49.41
0.05	0.40	71.85	63.28	67.29	50.71
0.08	0.50	71.39	62.17	66.46	49.77
0.10	0.60	69.82	61.89	65.62	48.83

Table 9: Sensitivity to nuisance bounds

(\tau_{\ell},\tau_{u})

on S2Looking.

$\lambda_{\mathrm{e}}$	$\lambda_{\mathrm{c}}$	Pre.(%)	Rec.(%)	F1(%)	IoU(%)
0.1	0.5	68.93	61.07	64.76	47.89
0.3	0.5	70.46	62.73	66.37	49.67
0.5	1.0	71.85	63.28	67.29	50.71
0.5	2.0	71.68	62.41	66.72	50.07
1.0	1.0	70.51	62.19	66.09	49.35
1.0	2.0	69.37	60.54	64.65	47.77

Table 10: Sensitivity to loss weights

(\lambda_{\mathrm{e}},\lambda_{\mathrm{c}})

on S2Looking.

Remote Sensing Change Detection

Remote sensing change detection finds and interprets differences across Earth-observation images acquired at different times.

See all 89 papers

Beyond Quadratic: Linear-Time Change Detection with RWKV
Mar 20, 2026Zhenyu Yang, Gensheng Pei, Tao Chen +4Remote Sensing Change Detection
School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China · State Key Laboratory of Intelligent Manufacturing of Advanced Construction Machinery, China · Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, Korea
NeXt2Former-CD: Efficient Remote Sensing Change Detection with Modern Vision Architectures
Feb 21, 2026Yufan Wang, Sokratis Makrogiannis, Chandra KambhamettuRemote Sensing Change Detection Hybrid CNN-Transformer Architectures
University of South Florida · Delaware State University
FM-ChangeNet: Learning Change through Pathwise Feature Transport
Jul 6, 2026Roie Kazoom, George Leifman, Genady BeryozkinRepresentation Learning Remote Sensing Change Detection
Google Research

Mar 20, 2026J/K move · Enter open · S save

Beyond Quadratic: Linear-Time Change Detection with RWKV

Zhenyu Yang, Gensheng Pei, Tao Chen, Xia Yuan+3

School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China · State Key Laboratory of Intelligent Manufacturing of Advanced Construction Machinery, China · Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, Korea

Remote Sensing Change Detection

Open paper AskPDF arXiv

Existing paradigms for remote sensing change detection are caught in a trade-off: CNNs excel at efficiency but lack global context, while Transformers capture long-range dependencies at a prohibitive computational cost. This paper introduces ChangeRWKV, a new architecture that reconciles this conflict. By building upon the Receptance Weighted Key Value (RWKV) framework, our ChangeRWKV uniquely combines the parallelizable training of Transformers with the linear-time inference of RNNs. Our approach core features two key innovations: a hierarchical RWKV encoder that builds multi-resolution feature representation, and a novel Spatial-Temporal Fusion Module (STFM) engineered to resolve spatial misalignments across scales while distilling fine-grained temporal discrepancies. ChangeRWKV not only achieves state-of-the-art performance on the LEVIR-CD benchmark, with an 85.46% IoU and 92.16% F1 score, but does so while drastically reducing parameters and FLOPs compared to previous leading methods. This work demonstrates a new, efficient, and powerful paradigm for operational-scale change detection. Our code and model are publicly available.

Abstract

Remote Sensing Change Detection

Explore similar work

Beyond Quadratic: Linear-Time Change Detection with RWKV

NeXt2Former-CD: Efficient Remote Sensing Change Detection with Modern Vision Architectures

FM-ChangeNet: Learning Change through Pathwise Feature Transport