DWTGS: Rethinking Frequency Regularization for Sparse-view 3D Gaussian Splatting
This addresses the challenge of reconstructing high-quality novel views from sparse training data in 3D reconstruction, representing an incremental improvement in frequency regularization techniques.
The paper tackles the problem of overfitting to high-frequency details in sparse-view 3D Gaussian Splatting, resulting in improved generalization and reduced hallucinations, with DWTGS consistently outperforming Fourier-based methods in benchmarks.
Sparse-view 3D Gaussian Splatting (3DGS) presents significant challenges in reconstructing high-quality novel views, as it often overfits to the widely-varying high-frequency (HF) details of the sparse training views. While frequency regularization can be a promising approach, its typical reliance on Fourier transforms causes difficult parameter tuning and biases towards detrimental HF learning. We propose DWTGS, a framework that rethinks frequency regularization by leveraging wavelet-space losses that provide additional spatial supervision. Specifically, we supervise only the low-frequency (LF) LL subbands at multiple DWT levels, while enforcing sparsity on the HF HH subband in a self-supervised manner. Experiments across benchmarks show that DWTGS consistently outperforms Fourier-based counterparts, as this LF-centric strategy improves generalization and reduces HF hallucinations.