眼镜遮挡对人脸图像识别率影响较大, 为了提高戴眼镜人脸图像的识别率, 需要摘除正面人脸图像中眼镜。采用PCA和ICA算法提取了二阶统计信息、高阶信息, 较好地刻画了人脸的细节特征; 同时结合灰度补偿算法, 弥补了由于戴眼镜图像未参与特征空间的训练过程中所造成的合成图像中人脸表情的失真; 利用反复的迭代补偿算法进行人脸重建。通过Yale人脸数据库进行仿真实验结果表明: 该方法合成的图像没有眼镜的痕迹, 看起来更加自然, 有效地改善了戴眼镜图像的面部特性, 提高了人脸的识别率。

Glasses as the most common occluding object in facial images have a greatly influence on face recognition performance. In order to improve the recognition rate with glasses, the glasses need to be removed from frontal facial image. An algorithm based on principle component analysis and independent component analysis is proposed to extract the second order statistics feature and higher-order information, which depicting well the detailed features of a face. Gray compensating algorithm is adopted in order to remedy distortion, for which the image that wear glasses do not participate in the training of the feature space. Iterative compensation algorithm is used for face reconstruction. The simulation experimental results show that through the Yale face database, the synthesize images using the proposed method have not the signs of glasses, looks more natural, improve effectively the characteristics of face images with glasses and face recognition rate.