Gaussian Splatting-Based Head and Pinna Reconstruction for Individualized HRTF Computation from Commodity Multi-View Images

Zhu H, Wu X, Qu T. Gaussian Splatting-Based Head and Pinna Reconstruction for Individualized HRTF Computation from Commodity Multi-View Images. In: Convention Paper. Audio Engineering Society; 2026.pp. 10290:1-9. Available from: https://aes.org/publications/elibrary-page/?id=23188

@inproceedings{Zhu2026_23188,
  author = {Zhu, Houlin and Wu, Xihong and Qu, Tianshu},
  title = {{Gaussian Splatting-Based Head and Pinna Reconstruction for Individualized HRTF Computation from Commodity Multi-View Images}},
  pages = {10290:1-9},
  url = {https://aes.org/publications/elibrary-page/?id=23188},
  year = {2026},
  month = may,
  publisher = {Audio Engineering Society}
}

TY  - CPAPER
TI  - Gaussian Splatting-Based Head and Pinna Reconstruction for Individualized HRTF Computation from Commodity Multi-View Images
AU  - Zhu, Houlin
AU  - Wu, Xihong
AU  - Qu, Tianshu
SP  - 1
EP  - 9
M1  - Paper 10290
PY  - 2026
DA  - 2026/05/28
UR  - https://aes.org/publications/elibrary-page/?id=23188
PB  - Audio Engineering Society
LA  - en
AB  - Individualized head-related transfer functions (HRTFs) require accurate pinna geometry, yet commodity multi-view captures leave the ear region self-occluded and weakly textured. We present a practical pipeline that couples ear-centric acquisition with 3D Gaussian splatting (3DGS) and the boundary element method (BEM) for complete HRTF computation. The protocol augments horizontal views with per-ear elevated captures under directional lighting; 3DGS training with depth-distortion regularization yields watertight meshes via truncated signed distance function (TSDF) fusion. Standardized head coordinates and ear-canal annotations interface the mesh with BEM. Experimental evaluations demonstrate that our method achieves lower ear-region geometric error and lower full-band spectral distortion compared to existing image-based personalized reconstruction baselines including AudioEar, NeuS, and Metashape MVS.
ER  -