DocumentCode :
2762727
Title :
Detection of Face and Facial Features in digital Images and Video Frames
Author :
Beigzadeh, M. ; Vafadoost, M.
Author_Institution :
Dept. of Biomed. Eng., Amir Kabir Univ., Tehran
fYear :
2008
fDate :
18-20 Dec. 2008
Firstpage :
1
Lastpage :
4
Abstract :
In the recent few decades, automatic detection and tracking of face and facial features such as eyes and mouth, in image and video sequences has become an active research area in machine vision applications such as Human-Computer Interaction (HCI). In this paper, a new algorithm for detection of face and facial features is proposed that can localize eyes and mouth very accurately in images. In this method, a combination of luminance, color and edge properties of image is used. This method is compared to the method introduced by Rein Lien Hsu, in which color and luminance information is used, and it is shown that the new algorithm is more robust and accurate in locating eyes and mouth in facial images with maximum 30 degrees of lateral rotation. Both methods are implemented and tested on a database containing 103 different images of face, and it is shown that the proposed method increases the accuracy by 4 percent and reaches to 91.26% of accuracy.
Keywords :
edge detection; face recognition; feature extraction; image colour analysis; digital images; eyes; face detection; facial features; human-computer interaction; image color analysis; image edge analysis; image luminance analysis; machine vision applications; mouth; video frames; Digital images; Eyes; Face detection; Facial features; Human computer interaction; Image edge detection; Machine vision; Mouth; Robustness; Video sequences; Eye localization; Face Detection; Facial Features Detection; eye tracking;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Biomedical Engineering Conference, 2008. CIBEC 2008. Cairo International
Conference_Location :
Cairo
Print_ISBN :
978-1-4244-2694-2
Electronic_ISBN :
978-1-4244-2695-9
Type :
conf
DOI :
10.1109/CIBEC.2008.4786053
Filename :
4786053
Link To Document :
بازگشت