LiTe-GS: Oracle-Efficient Next Best View Selection for 3D Gaussian Splatting

LiTe-GS：面向 3D 高斯泼溅（3D Gaussian Splatting）的预言机高效下一最佳视角选择

Abstract: Selecting informative camera views is critical for efficient training and adaptive refinement in 3D Gaussian Splatting, where each observation significantly influences model parameters. However, information-driven view-selection strategies can require repeated evaluations of expensive information-gain oracles as the number of candidate views increases.

摘要： 在 3D 高斯泼溅（3D Gaussian Splatting）中，选择具有信息量的相机视角对于高效训练和自适应优化至关重要，因为每一次观测都会显著影响模型参数。然而，随着候选视角数量的增加，基于信息驱动的视角选择策略往往需要反复评估昂贵的信息增益预言机（information-gain oracles）。

We propose LiTe-GS, an oracle-efficient method for next best view selection in 3D Gaussian Splatting. LiTe-GS reduces the number of information-oracle evaluations by performing randomized subset evaluation of candidate views rather than exhaustively scoring the full candidate pool.

我们提出了 LiTe-GS，这是一种用于 3D 高斯泼溅中下一最佳视角选择的预言机高效方法。LiTe-GS 通过对候选视角进行随机子集评估，而非对整个候选池进行穷举评分，从而减少了信息预言机的评估次数。

The resulting approach achieves expected $O(M\log(1/\epsilon))$ oracle complexity with respect to the number of candidate views $M$, independent of the selection cardinality $K$, while providing an explicit trade-off between oracle efficiency and approximation quality through $\epsilon$. We provide theoretical guarantees on oracle complexity and approximation performance under the proposed selection scheme.

该方法在候选视角数量 $M$ 方面实现了预期的 $O(M\log(1/\epsilon))$ 预言机复杂度，且与选择基数 $K$ 无关，同时通过 $\epsilon$ 在预言机效率和近似质量之间提供了明确的权衡。我们为所提出的选择方案在预言机复杂度和近似性能方面提供了理论保证。

Experiments on Blender and Mip-NeRF 360 demonstrate that LiTe-GS maintains reconstruction quality comparable to Fisher-information-based baselines while substantially reducing the number of Fisher-oracle evaluations across different acquisition settings.

在 Blender 和 Mip-NeRF 360 数据集上的实验表明，LiTe-GS 在保持与基于费舍尔信息（Fisher-information）的基准方法相当的重建质量的同时，在不同的采集设置下大幅减少了费舍尔预言机的评估次数。