Human face tracking using Gaussian mixture models and fuzzy shape analysis

This paper presents a novel approach for fast human face location and tracking in color image sequences. First, the skin-like pixels are detected by modeling skin color as Gaussian mixtures. The mixture model is used to assign a probability to each pixel in an image and face-like regions are detected by grouping suitably sized areas of high probability. Then, the face-like regions are determined by fuzzy shape analysis. Experimental results show that the tracking speed is fast and can be easily integrated into a human-computer interaction system.