Recently, there have been advances towards providing personalised binaural audio by means of numerical cal-culations based on a 3D head mesh. Nevertheless, 3D head meshes obtained by non-contact methods such as photogrammetric reconstruction have difficulties in accurately representing the complex biological structure of the pinna. The problems manifest in noisy meshes with holes, confounding the quality of the calculations. These problems can be tackled by applying a template mesh that is deformed to the personalised but noisy target mesh. In this contribution, we review two state-of-the-art non-rigid registration algorithms applied to the deformation of two high-resolution, high-accuracy pinna point clouds. We evaluate the algorithms by discussing their geometric errors, duration of the alignment task, and their robustness to spatial down-sampling.
(See document for exact affiliation information.)
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member Join the AES. If you need to check your member status, login to the Member Portal.
