Face detection, identification and tracking using support vector machine and fuzzy Kalman filter

This paper presents methodologies and techniques for human face detection, identification and tracking used for a human-robot interactive system. A fuzzy skin color adjuster together with standard image processing algorithm is proposed to detect human faces, and then identify them using the nonlinear support vector machine (SVM) and the Euclidean distance measure. A fuzzy Kalman filtering scheme is presented to track the identified human faces. Experimental results are conducted to verify the effectiveness and merit of the three proposed methods.