[2B1] 視覚3

[2B1-02] Segmentation of Reflection and Iris Pattern in Human Corneal Images

○Anshi Liu¹, Photchara Ratsamee¹, Tomohiro Mashita¹, Kiyoshi Kiyokawa², Yuki Uranishi¹, Haruo Takemura¹ (1. Graduate School of Information Science and Technology, Osaka University, 2. Nara Institute of Science and Technology)

Optical See-Through Head Mount Display is expected to be the future of Augmented Reality (AR). To achieve seamless integration between real and AR scene, the reflection on the human iris (i.e. what human is seeing) is an essential clue. This paper introduces a segmentation method between reflection and Iris pattern. The whole procedure includes three steps: 1) prediction of eye's location and orientation, 2) applying background subtraction algorithm to remove the influence of reflections, 3) reconstructing iris pattern by subtracting the reflections. Our preliminary experimental results in virtual environment show that the proposed method achieved segmentation in the case of dynamic eye movement with a static background.