3D photogrammetry as a low-cost and non-invasive method for acoustic modelling of animal hearing

Sound localization is important for all eared animals and the spatio-spectral cues for localization are described through the head-related transfer function (HRTF). Current state-of-the-art for obtaining the HRTF involves either direct measurement with a microphone at the eardrum, or a μCT scan to create a 3D model of the head for acoustic modelling. Both methods usually involve dead animals. We developed a low cost, portable, and adaptable 3D photogrammetry approach to create scaled 3D models of different sized animals with sufficient detail to simulate the HRTF using the boundary element method and tested it on two species of small, echolocating bats (Myotis daubentonii and M. nattereri) and a domestic pig (Sus domesticus). We directly compare the mesh models generated by our photogrammetry method to μCT scans as well as the simulated HRTFs from both with measurements using an in-ear microphone. Furthermore, we designed a set-up of 28 cameras to obtain 3D models and HRTF from live awake animals and tested it on a M. daubentonii. The geometries of the mesh models of bats match well between photogrammetry and μCT, but with increasing errors where line-of-sight is compromised for photogrammetry. The resulting HRTFs are in good agreement when comparing μCT and in-ear measurements to photogrammetry for both bats and pig. The 3D model and simulated HRTF of the live and awake bat likewise aligns well to the results from the deceased animals. Photogrammetry is a viable alternative to μCT scans for the generation of surface models of animals for the purpose of acoustic modelling. These models allow numerical modelling of HRTFs at biologically relevant frequencies, shown for animals ranging from small bats to large domestic pigs and applicable to even larger animals. Moreover, photogrammetry allows for model generation and subsequent HRTF simulation of live, awake animals, abolishing the need for euthanasia and anaesthesia. It paves the way for large-scale acquisition of 3D models for various purposes including modelling of animal hearing.