ERDDCI：基于双链反演的高质量图像编辑精确可逆扩散
ERDDCI: Exact Reversible Diffusion via Dual-Chain Inversion for High-Quality Image Editing
AI总结：
本文提出ERDDCI方法，通过双链反演（DCI）实现精确可逆扩散过程，避免繁琐优化，并引入动态控制策略，在50步扩散中实现高质量图像重建与编辑。
摘要：
扩散模型（DMs）已成功应用于真实图像编辑。这些模型通常将图像反演为用于重建原始图像的潜在噪声向量（称为反演），然后在推理过程中对其进行编辑。然而，近期流行的DMs往往依赖局部线性化假设，即反演过程中注入的噪声被期望近似于推理过程中去除的噪声。虽然DM在该假设下能高效生成图像，但该假设也会在扩散过程中累积误差，最终对真实图像重建和编辑的质量产生负面影响。为解决这一问题，我们提出了一种新方法，称为ERDDCI（Exact Reversible Diffusion via Dual-Chain Inversion，基于双链反演的精确可逆扩散）。ERDDCI使用新的双链反演（DCI）进行联合推理，以推导出精确可逆的扩散过程。通过使用DCI，我们的方法有效避免了现有反演方法中繁琐的优化过程，并实现了高质量图像编辑。此外，为适应高引导尺度下的图像操作，我们引入了一种动态控制策略，以实现更精细的图像重建和编辑。我们的实验表明，ERDDCI在50步扩散过程中显著优于最先进的方法。它实现了快速且精确的图像重建，SSIM为0.999，LPIPS为0.001，并在图像编辑方面也取得了有竞争力的结果。
英文摘要：
Diffusion models (DMs) have been successfully applied to real image editing. These models typically invert images into latent noise vectors used to reconstruct the original images (known as inversion), and then edit them during the inference process. However, recent popular DMs often rely on the assumption of local linearization, where the noise injected during the inversion process is expected to approximate the noise removed during the inference process. While DM efficiently generates images under this assumption, it can also accumulate errors during the diffusion process due to the assumption, ultimately negatively impacting the quality of real image reconstruction and editing. To address this issue, we propose a novel method, referred to as ERDDCI (Exact Reversible Diffusion via Dual-Chain Inversion). ERDDCI uses the new Dual-Chain Inversion (DCI) for joint inference to derive an exact reversible diffusion process. By using DCI, our method effectively avoids the cumbersome optimization process in existing inversion approaches and achieves high-quality image editing. Additionally, to accommodate image operations under high guidance scales, we introduce a dynamic control strategy that enables more refined image reconstruction and editing. Our experiments demonstrate that ERDDCI significantly outperforms state-of-the-art methods in a 50-step diffusion process. It achieves rapid and precise image reconstruction with an SSIM of 0.999 and an LPIPS of 0.001, and also delivers competitive results in image editing.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
