Skip to Main Content
This paper presents a new real-time face detection method that uses the skin color model in the YCrCb chrominance space to remove the non-skin-color pixels from the image from which we then extract candidate human face regions. We use the mathematical morphological filter to remove noisy regions and fill holes in the skin-color region. To locate the face region, we compute the similarity between the human face features and the candidate face regions in the image. We have implemented this algorithm in our smart media systems and found it effective in a real environment.