论文阅读-基于神经辐射场的视点合成算法综述

常远,盖孟.基于神经辐射场的视点合成算法综述[J].图学学报,2021,42(03):376-384.

占位网格：

[18] MESCHEDER L, OECHSLE M, NIEMEYER M, et al.Occupancy networks: learning 3D reconstruction in functionspace[C]//2019 IEEE Conference on Computer Vision andPattern Recognition (CVPR). New York: IEEE Press, 2019:4455-4465.

[19] CHEN Z Q, ZHANG H. Learning implicit fields for generativeshape modeling[C]//2019 IEEE Conference on ComputerVision and Pattern Recognition (CVPR). New York: IEEEPress, 2019: 5932-5941.

距离函数表示：

[20] PARK J J, FLORENCE P, STRAUB J, NEWCOMBE R, et al.DeepSDF: learning continuous signed distance functions forshape representation[C]//2019 IEEE Conference on ComputerVision and Pattern Recognition (CVPR). New York: IEEEPress, 2019: 165-174.

[21] SAITO S, HUANG Z, NATSUME R, et al. Pifu: pixel-alignedimplicit function for high-resolution clothed humandigitization[C]//2019 IEEE/CVF International Conference【人体穿衣高精度】

复杂问题：

[12] LIU L J, GU J T, LIN K Z, et al. Neural sparse voxelfields[C]//2020 Advances in Neural Information ProcessingSystems. Virtual: Curran Associates Inc, 2020: 15651-15663【神经稀疏体素场、8叉树】改进速度

[13] ZHANG K, RIEGLER G, SNAVELY N, et al. Nerf++:analyzing and improving neural radiance fields[EB/OL].[2021-01-11]. https://arxiv.org/abs/2010.07492v2.【改进采样过程】

[14] YU A, YE V, TANCIK M, et al. pixelNeRF: neural radiancefields from one or few images[EB/OL]. [2021-02-01].https://arxiv.org/abs/2012.02190v1.【泛化】

[15] TANCIK M, MILDENHALL B, WANG T, et al. Learnedinitializations for optimizing coordinate-based neuralrepresentations[EB/OL]. [2021-01-15]. https://arxiv.org/abs/2012.02189v2.【泛化】

[16] LI Z Q, NIKLAUS S, SNAVELY N, et al. Neural scene flowfields for space-time view synthesis of dynamicscenes [EB/OL]. [2021-02-15]. https://arxiv.org/abs/2011.13084v1.

[17] PARK K, SINHA U, BARRON J T, et al. Deformable neuralradiance fields[EB/OL]. [2021-01-29]. https://arxiv.org/abs/2011.12948.

可微方式：

[22] NIEMEYER M, MESCHEDER L, OECHSLE M, et al.Differentiable volumetric rendering: learning implicit 3drepresentations without 3D supervision[C]//2020 IEEE/CVFConference on Computer Vision and Pattern Recognition. NewYork: IEEE Press, 2020: 3501-3512.

[23] SITZMANN V, ZOLLHöFER M, WETZSTEIN G. Scenerepresentation networks: Continuous 3d-structure-aware neuralscene representations[EB/OL]. [2021-01-18]. https://arxiv.org/abs/1906.01618?context=cs.【循环神经网络】

改进位置编码：

[28] TANCIK M, SRINIVASAN P P, BEN MILDENHALL B, etal. Fourier features let networks learn high frequency functionsin low dimensional domains[EB/OL]. [2021-01-09].https://arxiv. org/abs/2006.10739.

[29] JACOT A, GABRIEL F, HONGLER C. Neural tangent kernel:convergence and generalization in neural networks[C]//The32nd International Conference on Neural InformationProcessing Systems. New York: ACM Press, 2018: 8580-8589.

改进速度：

[31] LINDELLD B, MARTEL J N P, WETZSTEIN G. Automaticintegration for fast neural volume rendering[EB/OL].[2021-02-03]. https://arxiv.org/abs/2012.01714.

[32] NEFF T, STADLBAUER P, PARGER M, et al. DONeRF:towards real-time rendering of neural radiance fields usingdepth oracle networks[EB/OL]. [2021-01-28]. https://arxiv.org/abs/2103.03231.

泛化：

[33] TREVITHICK A, YANG B. GRF: learning a general radiancefield for 3D scene representation and rendering[EB/OL].[2021-02-10]. https://arxiv.org/abs/2010.04595.

[14]\[15]

外观编码：改进拍摄条件约束，静态变动态场景

[35] BOJANOWSKI P, JOULIN A, LOPEZ-PAS D, et al.Optimizing the latent space of generative networks[C]//The35th International Conference on Machine Learning. Princeton:International Machine Learning Society (IMLS), 2018: 599-608.

改进光照条件：

[36] SRINIVASAN P P, DENG B Y, ZHANG X M, et al. NeRV:neural reflectance and visibility fields for relighting and viewsynthesis[EB/OL]. [2021-01-02]. https://arxiv.org/abs/2012.03927.

[37] BI S, XU Z X, SRINIVASAN P, et al., Neural reflectancefields for appearance acquisition[EB/OL]. [2021-01-19].https://arxiv.org/abs/2008.03824v2.

[38] BOSS M, BRAUN R, JAMPANI V, et al. NeRD: neuralreflectance decomposition from image collections[EB/OL].[2021-02-04]. https://arxiv.org/abs/2012.03918.

动态：

[16] LI Z Q, NIKLAUS S, SNAVELY N, et al. Neural scene flowfields for space-time view synthesis of dynamicscenes [EB/OL]. [2021-02-15]. https://arxiv.org/abs/2011.13084v1.

[17] PARK K, SINHA U, BARRON J T, et al. Deformable neuralradiance fields[EB/OL]. [2021-01-29]. https://arxiv.org/abs/2011.12948.