中文摘要：

人脸识别中面部遮挡易影响识别性能,针对该问题提出一种判别共同向量方法。该方法利用DWT对训练图像进行预处理,利用DCVA提取特征以确定感兴趣区域,计算测试图像与训练图像之间的多流形距离,并利用稀疏重建系数和最近邻分类器完成识别。在AR及LFW人脸数据库上的实验结果表明,该方法的识别率可高达99%,相比其他几种较新的面部遮挡识别方法取得了更高的识别率,同时减少了识别所耗的时间。

英文摘要：

Since the face covering may influence on the recognition performance in face recognition,a discriminative common vector approach（DCVA）is proposed. The DWT is used to pre-process the training images by the approach,and DCVA is used to extract the feature to determine the region of interest（ROI）. The multi-manifold distance between the testing image and training image is calculated,and then recognized with sparse reconstruction coefficient and nearest neighbor classifier. The experimental results for AR and LFW face database show that the recognition accuracy of the proposed method can reach up to 99%. Compared with other advanced recognition methods,this approach has higher recognition accuracy,and can reduce the recognition time.