Study on the Feature Extraction of Face using Color Space

In this paper suggested the feature extraction algorithm of eye-line, lip and nose from the picture of the face getting from smart phone. The experimental picture is restricted to one person's face, any place is available as background, lighting condition is not standard and there is no racial restriction. The suggested algorithm is to extract candidate face region using Harr's classifier from the input picture and then fixed candidate region of eye. Extracted the large value using dilation operation from eye candidate region and after separating eye and eyelash from the picture detected eye-line using Optimal Ellipse Estimation Method to the exterior angle line of eye. After seeking candidate region of lips using acquired center of ellipse of 2 eyes, extracted character of lip using relative difference of a* color in La*b* color space. This algorithm shows efficiency than LipMap suggested in the space of YCbCr by Hsu which is currently used in much. After extracting character of lip, seek candidate region of nose and extracted character of nose using relative difference of the position from the candidate region. (C) 2012 Elsevier B.V. Selection and/or peer-review under responsibility of Bin Nausantar University