Extraction of Facial Features from Color Images
In this paper, a method for localization and extraction of faces and characteristic facial features such as eyes, mouth and face boundaries from color image data is proposed. This approach exploits color properties of human skin to localize image regions – face candidates. The facial features extraction is performed only on preselected face-candidate regions. Likewise, for eyes and mouth localization color information and local contrast around eyes are used. The ellipse of face boundary is determined using gradient image and Hough transform. Algorithm was tested on image database Feret.
KeywordsSkin color segmentation, face detection, facial features extraction, morphological operations, ellipse detection, Hough transform, Feret database
Document typePeer reviewed
Document versionFinal PDF
SourceRadioengineering. 2008, vol. 17, č. 3, s. 115-120. ISSN 1210-2512
- 2008/3