基于离散傅里叶不变特征的人脸识别

Face Recognition Based on Fourier Scale Invariant FeaturesCHEN Mianshu

给出了一种基于离散傅里叶不变特征的人脸识别方法。从连续傅里叶变换出发,讨论连续傅里叶变换情况下的傅里叶变换性质,给出离散傅里叶变换情况下的傅里叶变换性质。依据离散傅里叶变换性质,推导出离散傅里叶变换的不变特征,并将其用于人脸图像识别。人脸识别结果表明方法具有很好的识别能力。

A face recognition method based on discrete Fourier invariant features is presented. Starting from continuous Fourier transform, the properties of continuous Fourier transform are discussed, and the paper gives the properties of discrete Fourier transform. Then according to the properties of discrete Fourier transform, it deduces the discrete Fourier invariant features that are employed on facial image recognition. The experiment results show that the face recognition methods based on discrete Fourier invariantfeatures proposed have a good ability of recognition for the translation, scale and rotation-on-the-plane variant face images.