一种基于肤色和几何特征的人面部识别方法

A Facial Recognition Based on Complexion and Geometry

面部识别的前提条件是场景中人脸的正确定位与分割。因此 ,本文提出了一种基于肤色及面部几何特征的人脸分割方法。首先利用图像的时间差分方法确定场景中有没有运动物体 ,然后利用 BP神经网络对肤色进行识别 ,比用 HIS或 L HC颜色空间中的色调识别肤色的方法具有更强的环境适应能力 ,最后利用扫描投影算法及面部固有的几何特征定位分割细节特征。

The correct segmentation is the critical step for the correct recognition.We propose an efficient facial feature segmentation approach.The motion information is first used to determine whether there are motive objects in the scene or not.Then,a BP artificial neural networks with robust adaptability is designed to classify the complexion.Finally,a scan projection algorithm is used to segment each apparatus from the segmented face.This approach can be time saving and locate apparatus more accurately.