Face detection in color images, a robust and fast statistical approach

In the past decade, face detection has claimed a lot of research owing to its significance in a wide variety of applications. These include face recognition, surveillance systems, teleconferencing, video phones, gaze tracking, lip reading, video watch ... and the list goes on. This paper presents a fast and robust statistical approach towards face detection. The detection comprises of two steps: i) skin color segmentation, and ii) locating the face in the segmented region based on correlation of the vertical luminous values with two data sets; one for eyes and the other for nose. Special attention is given to make the detection as fast as possible so to apply this algorithm in real time video applications. Finally conclusions are presented on the basis of results obtained.