Dual-Space Skin-Color Cue Based Face Detection for Eye Location

A eye location algorithm for color images was proposed based on face detection. Combination of skin-color segmentation based on both YCbCr color space and HSV color space was used for the face detection. Morphological operations were used for further processing of the skin-like color regions. The grey level information and the geometry feature were used to determine the face boundaries. Grey level integral and gradient integral projections were used for the eye location. The experiments were implemented on both single face images and video sequences respectively. The results demonstrated that our face detection approach was better than those ones solely based on either the YCbCr space or the HSV space and the eye location algorithm was very robust.