摘要
本文首次提出应用局部二值模式(LBP)算法来预测人脸旋转姿态。首先建立平均人脸的特征直方图模型(正面人脸特征直方图、左转人脸特征直方图和右转人脸特征直方图),然后通过比较测试人脸特征直方图与平均人脸特征直方图的距离来预测人脸旋转类型。实验结果表明本文方法可以有效的预测人脸姿态。
In this paper,Local binary pattern（LBP） algorithm is first proposed to predict facial rotation.Firstly,the mean histogram models（frontal facial histogram,left-rotated facial histogram and right-rotated facial histogram） are generated.Then compare the distance between the test facial histogram and mean facial histogram models.According to the comparison results the facial rotation can be predicted.Experimental results prove that our method can predict the facial rotation successfully.
出处
《仪器仪表用户》
2011年第3期76-77,共2页
Instrumentation
