基于局部分割的快速人脸识别

针对人脸识别系统易受光照,表情与遮挡等因素的影响,提出一种基于局部分割的快速人脸识别算法。首先,建立高斯肤色模型,并融合几何特征快速实现人脸粗定位;然后,利用局部初次匹配与全局特征的方法排除背景环境对人脸检测的干扰以及减少匹配过程中的计算量,同时运用LBP算子对2D-Fr FT的幅度与相位特征的互补信息进行编码;最后,采用最近邻分类器进行分类识别。本文的方法在公共人脸图像数据库上进行仿真实验。结果表明,该算法简单、鲁棒性高、在速度与准确性方面具有良好的性能。

Fast Face Recognition Based on Local Segmentation

In view of the influence of face recognition system, such as illumination, facial expression and occlusion, etc, a novel fractional Fourier features method based on local segmentation for fast face recognition is put forward. Firstly, YCbCr color space model is established and it combines geometrical characters to achieve the purpose of face coarse location; And then the eyes and mouth maps are constructed for local initial matching, the amplitude and phase information of the different transform orders in fractional Fourier transform domain are extracted as facial features, utilizing LBP operator to extract texture features; Finally, In order to improve the matching speed and reduce the data traffic, it recognizes the face by the means of utilizing global feature method to secondary matching to overcome the effect of the backgrounds. The simulation results are conducted based on color face image library, and the experiment demonstrates that the proposed method is not only simple and robust stronger, but better performance in recognition speed and rate.