【摘要】 在带遮挡的人脸识别中,由于人脸图像呈现出较大差异性,使得特征关键点之间产生错误匹配,对人脸识别率造成很大影响。对基于尺度不变特征变换(SIFT)的人脸识别算法进行优化,提出一种全新的匹配策略,能够减少错误匹配的特征关键点对,并将其应用于带遮挡的人脸识别中。实验结果表明,该优化算法比以往的一些带遮挡人脸识别算法都具有更好的识别结果。更多还原

【Abstract】 Face images have great diversity in the situation of occlusion,which leads to many incorrect matches between keypoints,and then brings about great impact on recognition accuracy.In this paper,an optimization of Scale Invariant Feature Transform(SIFT) was proposed and applied to face recognition,in which the number of incorrect matching feature keypoint pairs was reduced by utilizing the coordinate information between each pair of feature keypoints.Experimental results demonstrate that the performance has been significantly improved compared to several other algorithms in the face recognition with occlusions.更多还原