基于Radon变换的人脸识别

Face Recognition Based on Radon Transform

针对目前大部分的人脸识别算法不能同时克服光照、表情、遮挡等复杂条件干扰的问题,提出一种基于Radon变换的人脸识别算法。该算法首先采用一种光照归一化算法对人脸图像进行预处理,然后采用Radon变换提取人脸的不变特征,最后采用二维最近邻分类器进行人脸识别。通过在AR人脸库的人脸识别实验,说明了该算法对多种复杂条件下的人脸识别具有一定的鲁棒性和普适性。

As most face recognition algorithms can＇t overcome interference problems of complex conditions, such as il- lumination, expression, shelter, a face recognition algorithm based on Radon transform is presented. The algorithm uses an il- lumination normalization algorithm in the first place for face image preprocessing, then uses the Radon transform to extract the face of the invariant features, finally a two-dimensional nearest neighbor classifier is used to face recognition. Face recog- nition experiments through the AR face database, show that the algorithm under various complex conditions for face recogni- tion has a certain robustness and universality.