Multimodal Personal Ear Authentication Using Multiple Sensor Information

In this study, we examine personal authentication technology using individual differences of acoustic transfer function measured from the pinna; the Pinna Related Transfer Function (PRTF). In this paper, we consider personal authentication using a smartphone, which is convenient and can be authenticated simply by holding it against the pinna. However, it is known that in the case of using a smartphone, PRTF changes according to the direction of signal emission to the pinna during measurement, so that the authentication rate is lower than that of using earphones and headphones. The position information of the smartphone is measured by the acceleration sensor and proximity sensor and used together with the PRTF for learning at the same time. We demonstrate that the proposed method can deal with the changes in the smartphone's position of each measurement and improve the robustness.