顔テクスチャ情報を用いた照明と姿勢の変動に頑健な人物抽出手法

概要

論文の詳細を見る
In recent studies, it's used as the man-machine interface to a person's behavior or physical features. Especially, the face is biometrics, which is most commonly used in everyday life. However, conventional methods for the face extraction have a problem that is vulnerable to posture and illumination changes of the face. In addition, when using the visible or the infrared camera to the input device, in the conventional method, it's necessary to use a method according to the camera. In this study, we proposed a technique that doesn't depend on the input device to extract the face area where robust to illumination changes. This technique is used to encode the texture information of the face in the facial feature patterns. Texture information is robust to changes in illumination without selecting a device is characterized for not using the color information, it is easily calculated. Our conventional technique was used to calculate the RRC, the original technique cannot be robust against locally illumination changes.Therefore, in this technique we proposed an improved technique of RRC which gave further robustness to illumination changes. As a result, this technique can be satisfactorily extracted human region became possible.
関東学院大学工学部工学会の論文