发表机构
University of Nis, Faculty of Electronic Engineering; University MB, Information Technology department; University of Arizona, Electrical & Computer Engineering (ECE) department（尼什大学电子工程学院; MB大学信息技术学院; 亚利桑那大学电气与计算机工程系）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出能量条件噪声调度与谱白化策略，使前向扩散噪声适应谱能量分布，在CIFAR-10上将FID从142.48降至100.45。
AI 中文摘要
本文提出了一种用于变换域扩散模型的能量自适应噪声调度和白化策略。现有的谱扩散方法通过系数缩放、归一化或频率优先级来处理变换系数的非均匀统计特性，而前向扩散噪声调度在很大程度上仍独立于底层谱能量分布。我们研究了前向扩散过程的时间演化是否也应遵循自然图像的谱组织。所提出的公式将全局谱白化与能量条件噪声分配相结合，该分配根据单个变换系数的能量以及随扩散时间变化的图像相关能量路径共同调节注入的噪声。由此产生的前向过程保持高斯转移，具有闭式边际分布，并且与标准DDPM和DDIM程序兼容，无需修改扩散架构。在CIFAR-10上的实验证明了所提出的能量条件噪声调度和谱白化的贡献，将紧凑型DCTdiff U-Net变体的Fréchet Inception Distance从142.48降低到100.45。
英文摘要
This paper introduces an energy-adaptive noise scheduling and whitening strategy for transform-domain diffusion models. Existing spectral diffusion methods account for the non-uniform statistics of transform coefficients through coefficient scaling, normalization, or frequency prioritization, while the forward diffusion noise schedule remains largely independent of the underlying spectral-energy distribution. We investigate whether the temporal evolution of the forward diffusion process should also follow the spectral organization of natural images. The proposed formulation combines global spectral whitening with energy-conditioned noise allocation that jointly modulates the injected noise according to the energy of individual transform coefficients and an image-dependent energy path over diffusion time. The resulting forward process preserves Gaussian transitions with closed-form marginals and remains compatible with standard DDPM and DDIM procedures without modifying the diffusion architecture. Experiments on CIFAR-10 demonstrate the contribution of the proposed energy-conditioned noise schedule and spectral whitening, reducing Fréchet Inception Distance from 142.48 for a compact DCTdiff U-Net variant to 100.45.
Comments7 pages, 5 figures, 2 tables, Manuscript submitted for publication in Elsevier Pattern Recognition Letters
