光学精密工程2025，Vol.33Issue(15)：2424-2440,17.DOI:10.37188/OPE.20253315.2424
高频信息物体多层多元特征权重自适应融合三维重建网络
Multi-layer multi-feature adaptive weight fusion network for 3D reconstruction of objects with high-frequency information
摘要
Abstract
To mitigate the loss of high-frequency surface texture information and the resulting reduction in reconstruction accuracy in deep learning-based photometric stereo,a multi-layer multi-element feature weight adaptive fusion 3D reconstruction network(MMF-Net)is proposed.The network architecture builds on PS-FCN and incorporates a symmetric encoder-decoder to enhance feature learning,representa-tion capacity,and multi-level feature integration.A novel multi-element convolution layer with indepen-dent inter-layer adaptive weights is introduced;by incorporating additional trainable weights,both shape and texture cues are jointly leveraged to better capture fine surface texture variations,thereby improving stability and accuracy in scenes containing dense high-frequency information.An auxiliary skip-connection mechanism is also employed to propagate intermediate-layer features to later stages,preserving high-fre-quency details while reinforcing low-frequency structure,and enabling effective fusion of multi-band(high-and low-frequency)surface information.The method was evaluated on the DiLiGenT benchmark.MMF-Net attains an average mean angular error(MAE)of 6.94°,representing a 6%improvement over PS-FCN(Norm)at 7.39°.For objects exhibiting pronounced high-frequency surface detail,the average re-construction error is 11.03°,a 12%improvement relative to FUPS-Net at 12.52°.The results demon-strate that MMF-Net effectively captures both low-and high-frequency surface information in photometric stereo,offering a viable approach for high-precision 3D reconstruction from surface normals.关键词
深度学习/光度立体视觉/多元卷积/特征融合/自适应权重Key words
deep learning/photometric stereo vision/multivariate convolution/feature fusion/adaptive weighting分类
信息技术与安全科学引用本文复制引用
王标,李影,融百川,刘璟,张进,王永红..高频信息物体多层多元特征权重自适应融合三维重建网络[J].光学精密工程,2025,33(15):2424-2440,17.基金项目
国家自然科学基金资助项目(No.52375536) （No.52375536）
长三角科技创新共同体联合攻关资助项目(No.2022CS-JGG1300) （No.2022CS-JGG1300）
合肥市关键技术研发揭榜挂帅资助项目(No.2023SGJ020) （No.2023SGJ020）
