ビデオフレームからの口唇領域および目領域の特定ならびにその抽出

概要

論文の詳細を見る
Identification of facial parts and automatic extraction for the recognition of facial expression is presented. Beforethe recognition of facial expressions, each facial part, such as eyes, mouth, and so on must be precisely extracted inadvance to increase recognition accuracy of expressions. The identification of facial parts, however, is very difficult ingeneral, even using color and moving information. In order to identify the mouth and eyes, Q-element in a YIQ colorsystem and Qc-element in a modified HSV color model were used, respectively. Moreover, spatial relationship amongfacial parts is introduced for identifying the mouth area and to avoid misextraction of the nose area as a mouth area and,the eyes area as mouth area information. The extraction of facial parts is based on template matching. The templateused was devised so that the adequate position of the template can be easily identified and automatically certain.
2012-04-30