基于灰度和梯度不变矩的人脸识别

提出一种新的人脸识别方法———灰度-梯度共轭不变矩法,克服了传统方法的缺点,在不增加复杂度的前提下,将灰度信息和梯度信息有效地结合起来,并将其共轭分布反映的纹理特征作为对灰度特征的补充参与特征分类.同时,对经典不变矩组进行分析并改进,消除了离散状态下比例因子的影响.以BP网络作为分类器进行试验,结果表明,与传统方法相比,该方法具有更高的识别率.

Face Recognition Based on Gray and Grads Invariant Moments

A new method of face recognition based on gray and grads is presented to overcome the drawbacks existing in traditional method.By using invariant moments of conjugate matrix,without any further complexity,gray and grads information are combined effectively,by which texture feature was reflected to be as the supplement of gray feature.At the same time,the paper modifies invariant moments to remove the influence of scale factor.Choosing BP neural network as classifier,the result shows that the effect of new method is better than that of traditional method.