Localization of Facial Regions and Features in Color Images

1 Introduction Recognition of human faces out of still images or image sequences is a research eld of fast increasing interest. There are many diierent applications for systems coping with the problem of face localization and recognition, e.g. model-based video coding, security systems, mug shot matching. Due to variations in illumination, background, visual angle and facial expressions, the problem is complex. A critical survey about existing literature on human and machine recognition of faces is given in 1]. In a rst step of face recognition, the localization of facial regions and the detection of facial features as eyes, nose and mouth is necessary. In this framework we present an approach that copes with problems of this rst step. For the detection of facial regions and facial features, approaches have been published using texture, depth, shape and color information or combinations of them ((2], 4], 3], 6]). Our approach is based on the observation that human faces are characterized by their oval shape and their skin-color, also in the case of varying light conditions. Therefore we locate face-like regions on the base of shape and color (HSV) information. The hypotheses for faces are veriied by searching for facial features inside of the face-like regions. For that we rst enhance facial features by applying morphological operations and then we extract candidates for facial features by combining thresholding and a modiied watershed method. 2 Localization of face-like regions Faces are signiicantly characterized by their oval shape and speciic skin-color. For that reason, we have decided for an approach based on shape and color information. The eeectiveness of 1