arXiv cs.AI· Kaiwen Wang, Frank Bieder, Yinzhe Shen, Carlos Fernandez, Jan-Hendrik Pauls, Omer Sahin Tas·· 4 小时前AI 评分42
小波相位扩散（Wavelet Phase Diffusion）实现结构与语义一致的 Sim-to-Real 转换
Wavelet Phase Diffusion for Structurally and Semantically Consistent Sim-to-Real Translation
AI 导读
研究者提出 Wavelet Phase Diffusion（ψ-PD），在 Dual-Tree Complex Wavelet Packet Transform 域实现空间自适应相位注入，并以低频随机化（LFR）摆脱合成光照先验，解决傅里叶域全局频谱耦合导致的振铃和边界泄漏问题。
正文 · 原文
Abstract:Simulation-to-reality translation must bridge the appearance gap between synthetic and real domains while preserving structural and semantic consistency. Conditioning-based methods achieve spatial alignment but introduce computationally expensive control modules. Paired-data methods achieve realism but rely on complex synthesis pipelines, often altering scene geometry and semantics. Training-free editing methods avoid both constraints but lack a learned appearance prior, limiting their perceptual quality. Recently proposed phase-preserving diffusion presents a promising alternative, but Fourier-domain formulations are constrained by global spectral coupling. This coupling induces spatial artifacts such as ringing and boundary leakage, thereby degrading structural and semantic consistency. We introduce Wavelet Phase Diffusion, which addresses this through two components. First, we operate in the Dual-Tree Complex Wavelet Packet Transform domain, whose localized wavelet packets enable spatially adaptive phase injection without global spectral interference. Second, Low-Frequency Randomization (LFR) replaces the low-frequency packet, decoupling the model from the synthetic illumination prior and enabling in-distribution real-world appearance. Both components train on unpaired open-domain data, and introduce negligible inference overhead. The spatial locality further enables instance-level translation, where individual objects or regions are translated to photorealistic appearance independently while the surrounding scene remains untranslated. On vKITTI $\to$ KITTI image translation, $\psi$-PD outperforms prior methods in realism and semantic consistency while maintaining competitive structural alignment. For CARLA video translation, $\psi$-PD approaches the realism of paired-data methods while reducing VLM planner ADE and FDE by $5.4\%$ and $5.1\%$, respectively.
| Comments: | Accepted by WACV 2027. 20 pages, 15 figures. Project page: this https URL |
| Subjects: | Artificial Intelligence (cs.AI); Image and Video Processing (eess.IV) |
| Cite as: | arXiv:2607.21628 [cs.AI] |
| (or arXiv:2607.21628v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.21628 arXiv-issued DOI via DataCite |
Submission history
From: Kaiwen Wang [view email]
[v1]
Wed, 8 Jul 2026 13:30:26 UTC (41,693 KB)
[v2]
Wed, 29 Jul 2026 19:07:44 UTC (41,691 KB)
[v3]
Tue, 27 Oct 2026 07:59:22 UTC (47,289 KB)
来源：arXiv cs.AI · arxiv.org