A Kind of Human Face Region Detection and Recognition Method Based on Chrominance Information Characteristics

This paper introduced the face region segmentation and recognition algorithm based on chrominance information characteristics, the algorithm gets skin color clustering subspace by the training samples in HSI color space, and combination Gauss skin color model as the face pixel detection basis decision, Through it is detected face pixel vertical and horizontal direction projection, complemented by a priori knowledge of the face contour, eventually face region will be segmented. The algorithm is fast, without the face position, posture, facial expressions and other factors affecting the advantages. At the same time, in this algorithm, we realized corresponding light reverse compensation techniques to make the algorithm has the same good experimental effect in strong light.