Face Detection Based on Skin Color Segmentation, Mathematical Morphology and Geometry

A practical algorithm is presented to detect faces in color images with complex background. An image is first classified into skin/non-skin regions. Then, morphology algorithm is used to clean the skin-like image map by a pre-defined face element structure. Following that, face shape detection is carried out based on geometry. Experimental results show that the presented algorithm is flexible to various lighting conditions, faces with certain rotation conditions, multi-faces conditions and partial face covered conditions.