Face detection from complex scenes in color images

Automatic detection of faces in color images is a task of increasing importance for different multimedia applications. In this paper a 3-stage method for detecting human faces from images using color, shape and the gradient of luminance is presented. The aim of the method is to detect faces for a great variety of face sizes, positions and 3D orientations, including frontal and non-frontal views and situations when more than one face is present in the image. A measure of confidence that a region is a face is defined using parameters computed from the color, shape and local gradient of luminance features. This measure allows tuning the face detector to application requirements.