Abstract:
A method for detecting human faces in color images was developed and implemented, consisting of three main
steps: human color statistics, adaptive thresholding process and Template Matching. A skin model determines the
most likely skin regions in an image. This skin model is made from 20 skin samples from persons of different
ethnicity, which color distribution in the chromatic color space generates a chroma chart showing likelihood of
skin colors. The image is transformed into a gray scale image using this chart; the gray value at a pixel shows the
likelihood of the pixel representing skin. Then this gray scale image is segmented to skin and non-skin regions
using an adaptive thresholding process to achieve the optimal value. A fixed threshold value is not possible to be
found, since people with different skins have different likelihood. Finally, the template matching determines if a
given skin region represents a frontal human face.
