LowAux-RDNet：用于单图像反射去除的具有场景平衡真实世界训练的低通残差监督
LowAux-RDNet: Low-Pass Residual Supervision with Scene-Balanced Real-World Training for Single-Image Reflection Removal
浏览论文内容
中文总结 AI 辅助
研究单图像反射去除，基于RDNet构建管道，引入LowAux辅助目标，结合场景平衡真实对，构建统一基准，所提系统在多数据集上取得优异指标，性能在不同反射分布上更平衡，Postcard中语义反射仍具挑战。
中文摘要 AI 辅助
单图像反射去除旨在从通过玻璃拍摄的一张图像中恢复干净的传输层。我们研究了基于RDNet构建的显式分解管道，并引入了LowAux，一种仅用于训练的低通反射辅助目标。原始残差目标仍然是主要的反射监督，而对称滤波后的预测和目标提供了稳定的低频约束。我们进一步纳入来自RRW的场景平衡真实对，以扩大真实场景覆盖范围并提高跨数据集泛化能力。为避免由模型特定的调整大小、填充、输出量化和度量代码引起的评估差异，我们在CEILNet、Real20、Postcard、Objects和Wild上构建了一个统一的公共基准。在相同评估器下，所提出的系统在五个数据集上的宏平均PSNR为27.546 dB、SSIM为0.9220、NCC为0.9751、LMSE为0.004760，在比较的公共检查点和内部变体中实现了最高的宏平均PSNR、SSIM和NCC以及最低的LMSE。每个数据集的分析和定性分析表明，主要好处是在不同反射分布上性能更平衡，而Postcard中清晰的语义反射仍然具有挑战性。
英文摘要
Single-image reflection removal aims to recover a clean transmission layer from one image captured through glass. We study an explicit decomposition pipeline built on RDNet and introduce LowAux, a training-only low-pass reflection auxiliary objective. The original residual target remains the main reflection supervision, while symmetrically filtered prediction and target provide a stable low-frequency constraint. We further incorporate scene-balanced real pairs from RRW to broaden real-scene coverage and improve cross-dataset generalization. To avoid evaluation discrepancies caused by model-specific resizing, padding, output quantization, and metric code, we build a unified public benchmark over CEILNet, Real20, Postcard, Objects, and Wild. Under the same evaluator, the proposed system obtains a five-dataset macro average of 27.546 dB PSNR, 0.9220 SSIM, 0.9751 NCC, and 0.004760 LMSE, achieving the highest macro-average PSNR, SSIM, and NCC and the lowest LMSE among the compared public checkpoints and internal variants. Per-dataset and qualitative analyses show that the main benefit is a more balanced performance across diverse reflection distributions, while clear semantic reflections in Postcard remain challenging.
发表机构
- Fudan University（复旦大学）
机构由 AI 辅助整理，请以论文原文为准。
