LBP直方图与PCA的欧式距离的人脸识别

Face Recognition of LBP Histogram PCA and Euclidean Distance

基于LBP算子具有旋转不变性和灰度不变性等显著特点,本文通过LBP算子的特征提取,将人脸分成子区域,然后通过连接这些子区域的LBP直方图生成人脸特征向量,由于生成的特征向量的维数过高,通过PCA算法降维压缩,最后用欧式距离分类器完成测试样本和训练样本的人脸识别,通过实验比较得出很好的人脸识别效果,此人脸识别算法过程用于火车站等各种公共场合有很好的应用效果。

LBP operator has notable features of rotation invariance and gray-scale invariance etc. This paper uses LBP operator to get feature extraction, the face image is divided into sub-regions, then connecting these sub-regions LBP histogram to generate facial feature vector, because too many dimension of facial feature vector, using PCA to reduce dimension and compression. The final step is using Euclidean distance classifier to complete face recognition. Through the experimental conclusion shows very good face recognition effect. The face recognition algorithm used for various kinds of public, like the railway station have good application effect.