Two Stages Haar-Cascad Face Detection With Reduced False Positive

Face detection is one of the top hottest research topics in Computer vision. Human face remains the robust un-cloned biometric identity recognition which is widely used to provide person identity and has many applications for example security access, face recognition and surveillance. A common issue in face detection is that face detection rate is maximized with low threshold but this in contrast increase the false positive rate. In this paper we present two stage framework haar cascade detection algorithm where in the first stage the detected faces are cropped and re-detected by the second stage. The result is a noticeable improvement with false alarm reduction when compared to the pure algorithm alone.