半边人脸识别方法

Half Face Recognition Scheme

针对光线变化对人脸的正确识别的影响,提出一种半边人脸识别方法。该方法把训练集的人脸图片分为左右两个部分,经过亮度补偿后分别训练左右脸的特征提取器,提取训练集中左右半边脸的特征。计算待识别图片左右半边脸的熵,选取熵较大的半边脸及对应的特征提取器,经过亮度补偿后提取该半边脸的特征,并根据提取的特征进行归类。实验结果表明,对光线有变化的人脸,该方法比用全脸识别具有更好的识别率和识别速度。

This paper proposes a half face recognition scheme to reduce effect of variable light on face recognition. The faces in training set are divided into the left half and the right half, and it trains the left feature extractor and the right feature extractor after illumination compensation respectively. One half of the probe face is chosen by entropy and one’s features are extracted by the corresponding feature extractor after illumination compensation. The face is classified by the features of the chosen half face. Experimental results show that, under variable light, the scheme gains better recognition ratio and speed than the whole face.