基于方差相似度计算的人脸识别技术

Face Recognition Based on Computation of Variance Similarity

针对人脸识别过程中所需的图像匹配,采用计算特征矢量方差的方法,经过一定的对应关系变化,得到了相似度计算的权值.这些权值具有很好的稳定性和准确性,有利于识别率的提高.实验采用自制人脸库进行识别测试,取得了较满意的结果,验证了此方法的有效性.

This paper deals with image match for face recognition. Variances of feature vectors are computed, giving the weights of similarity after certain translation to the variations. This method is beneficial in increasing the face recognition rate due to its stability and accuracy. Experiments are carried out using the face database developed in the authors＇ laboratory. The recognition results are satisfactory, indicating validity of the proposed method.