著者

中川 優樹 石田 貴正 平野 晃昭 中村 納

出版者

工学院大学

雑誌

工学院大学研究報告 (ISSN:03685098)

巻号頁・発行日

no.111, pp.85-90, 2011-10

---

This paper considers a highly accurate extracting method for facial area and facial parts from a natural background based on moving and color information. To obtain facial parts, Q-element in the YIQ color model and Quesi-chrome (Qc) in the modefied HSV color model are mainly used. In our previous method, more than 90% of extracting accracy of facial areas was obtained. However, there still remain problems to be solved such as extraction accuracy and the evaluation of the facial area extracted. In this system a robust extraction system is proposed. From the computer simulation using 14,179 images, 97.10% for facial area and using 3,562 images, 92.70% for the mouth and 89.80% for the eyes are correctly extracted, respectively.