# SAS-Net: Cross-Domain Image Registration as Inverse Rendering via Structure-Appearance Factorization

Source: [SAS-Net: Cross-Domain Image Registration as Inverse Rendering via Structure-Appearance Factorization](<https://www.sota2.com/research/paper/698c0cebc97a0b3e3b5e102f>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-18T05:00:04.466Z

Authors: Jiahao Qin

Published: 2026-02-06T21:01:27.000Z

## Abstract

Cross-domain image registration requires aligning images acquired under heterogeneous imaging physics, where the classical brightness constancy assumption is fundamentally violated. We formulate this problem through an image formation model I = R(s, a) + epsilon, where each observation is generated by a rendering function R acting on domain-invariant scene structure s and domain-specific appearance statistics a. Registration then reduces to an inverse rendering problem: given observations from two domains, recover the shared structure and re-render it under the target appearance to obtain the registered output. We instantiate this framework as SAS-Net (Scene-Appearance Separation Network), where instance normalization implements the structure-appearance decomposition and Adaptive Instance Normalization (AdaIN) realizes the differentiable forward renderer. A scene consistency loss enforces geometric correspondence in the factorized latent space. Experiments on EuroSAT-Reg-256 (satellite remote sensing) and FIRE-Reg-256 (retinal fundus) demonstrate state-of-the-art performance across heterogeneous imaging domains. SAS-Net (3.35M parameters) achieves 89 FPS on an RTX 5090 GPU. Code:   https://github.com/D-ST-Sword/SAS-Net.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Image Registration on OR-PAM-Reg-Temporal 26K](<https://www.sota2.com/research/sota/image-registration-on-or-pam-reg-temporal-26k.md>): Image Registration; OR-PAM-Reg-Temporal 26K; TNCC: 0.967.

- [Intra-frame Image Registration on OR-PAM-Reg-Temporal-26K (test)](<https://www.sota2.com/research/sota/intra-frame-image-registration-on-or-pam-reg-temporal-26k-test.md>): Intra-frame Image Registration; OR-PAM-Reg-Temporal-26K (test); NCC: 0.994.

- [Temporal consistency evaluation on OR-PAM Reg-Temporal 26K (test)](<https://www.sota2.com/research/sota/temporal-consistency-evaluation-on-or-pam-reg-temporal-26k-test.md>): Temporal consistency evaluation; OR-PAM-Reg-Temporal-26K (test); TNCC: 0.967.

- [OR-PAM Registration on OR-PAM-Reg 4K (test)](<https://www.sota2.com/research/sota/or-pam-registration-on-or-pam-reg-4k-test.md>): OR-PAM Registration; OR-PAM-Reg 4K (test); SSIM: 89.4.

- [Image Registration on OR-PAM](<https://www.sota2.com/research/sota/image-registration-on-or-pam.md>): Image Registration; OR-PAM; Time (ms): 11.2.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2602.09050>)

[PDF](<https://arxiv.org/pdf/2602.09050.pdf>)