发表机构
Electiconsulting（电子咨询公司）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对盲高斯彩色图像去噪，提出结合多种要素的紧凑型无偏差ConvNeXt U-Net即BF-ConvUNeXt。通过特殊设计使其具有齐次性，能盲态泛化。训练单一模型，在多数据集多噪声水平下评估表现良好，虽略落后于最优技术，但参数少，还推动相关逆问题研究。
AI 中文摘要
我们描述并评估了BF-ConvUNeXt，这是一种用于盲加性白高斯噪声彩色图像去噪的紧凑型无偏差ConvNeXt U-Net。它结合了四种现有要素，使得单一属性端到端得以保留：一个冻结的深度可分离Gabor主干（定向带通，零可训练参数），一个拉普拉斯金字塔编码器将高频残差路由到每个跳跃连接中，一个ConvNeXt-V1 U-Net主体，以及全程无偏差构建（无加性偏差、线性头、LeakyReLU、仅方差批归一化）。这些共同使得具有0.82M参数的网络在推理时恰好是1次齐次的，即D(αy)=αD(y)，允许对残差进行宫泽/ Tweedie分数读取，并在一个模型中跨噪声水平进行盲态泛化。我们在噪声标准差课程（标准差约为6.4至64，0 - 255范围）上训练一个单一的盲模型；它能外推到该上限之外且无断崖式下降，在标准差=150时平滑降级到22.8dB，在标准差=200时为20.0dB。在四个标准彩色集（CBSD68、柯达24、麦克马斯特、Urban100）上以标准差在{15,25,50}进行评估时，它在每个集合和水平上都匹配或超过了DnCNN和FFDNet，比DnCNN平均高出约+0.7dB。与重量级CNN/Transformer当前最优技术相比，在其参数的1/15至1/39时，它以小幅度落后（根据集合不同约为0.3 - 1.7dB）。齐次性仅在推理时和特定检查点成立，且学习到的残差是局部而非全局分数（非保守雅可比矩阵），因此即插即用/RED保证不适用；但它仍推动随机采样和线性逆问题（图像修复、超分辨率、去模糊、压缩感知）。
英文摘要
We describe and evaluate BF-ConvUNeXt, a compact bias-free ConvNeXt U-Net for blind additive-white-Gaussian-noise color image denoising, combining four existing ingredients so a single property survives end to end: a frozen depthwise Gabor stem (oriented band-pass, zero trainable parameters), a Laplacian-pyramid encoder routing the high-frequency residual into each skip connection, a ConvNeXt-V1 U-Net body, and bias-free construction throughout (no additive bias, linear head, LeakyReLU, variance-only batch norm). Together these make the 0.82M-parameter network exactly degree-1 homogeneous at inference, D(alpha y) = alpha D(y), licensing a Miyasawa/Tweedie score reading of the residual and blind generalization across noise levels from one model. We train a single blind model on a noise-sigma curriculum (sigma approximately 6.4 to 64, 0-255 scale); it extrapolates past that ceiling without a cliff, degrading smoothly to 22.8 dB at sigma=150 and 20.0 dB at sigma=200. Evaluated unchanged on four standard color sets (CBSD68, Kodak24, McMaster, Urban100) at sigma in {15,25,50}, it matches or beats DnCNN and FFDNet on every set and level, averaging about +0.7 dB over DnCNN. Against heavyweight CNN/transformer state of the art it trails by a small margin (roughly 0.3-1.7 dB depending on set) at 1/15 to 1/39 of their parameters. The homogeneity is inference-only and checkpoint-specific, and the learned residual is a local, not global, score (non-conservative Jacobian), so plug-and-play/RED guarantees do not transfer; it still drives stochastic sampling and linear inverse problems (inpainting, super-resolution, deblurring, compressive sensing).
Comments15 pages, 1 figure
