Robustness of Biometric Authentication Method Using Image of External Auditory Canal

It is necessary to protect the individual imformation with the advancement of information technology. The biometric authentication will be needed to solve this issue. In previous research it has shown that images from inside the external auditory canal contain features that can identify individuals. However, what seems to be lacking is a consideration of the real-world images. This paper discusses the robustness of biometric authentication using image of external auditory canal. In this study, we assumed a real-life usage environment and performed image processing on the obtained image data, and evaluated the impact of this on recognition accuracy. When using VGG16, a large loss in accuracy was observed when the noise level was 150, but with ResNet50, the loss was less than 10%. Also, it turns out that when the light source is strong, the accuracy is lower than when it is weak.