CAJ | 학술논문

针对实时人脸识别易受光照变化影响的问题,提出了一种将局部二值模式(Local Binary Patterns,LBP)与嵌入式隐马尔可夫模型(Embedded Hidden Markov Model,EHMM)相结合的人脸识别方法.该方法首先对输入的人脸图像进行LBP预处理,接着对其进行特征向量提取,最后把提取的特征观察向量送入EHMM进行训练或识别.在多个人脸数据库上进行了实验,结果表明该文算法对光照具有较好的鲁棒性,提高了识别率.
침대실시인검식별역수광조변화영향적문제,제출료일충장국부이치모식(Local Binary Patterns,LBP)여감입식은마이가부모형(Embedded Hidden Markov Model,EHMM)상결합적인검식별방법.해방법수선대수입적인검도상진행LBP예처리,접착대기진행특정향량제취,최후파제취적특정관찰향량송입EHMM진행훈련혹식별.재다개인검수거고상진행료실험,결과표명해문산법대광조구유교호적로봉성,제고료식별솔.
A combination of Local Binary Pattern(LBP) with Embedded Hidden Markov Model(EHMM) is set up for against the sensitivity of real-time face recognition in the case of lighting variation. Face images are preprocessed according to LBP and then eigenvectors are extracted from consequences. Thereby, extracted observation vectors can do training or identification in EHMM. Experiments are conducted on several face database and numerical results verify the robustness of illumination, thus im?prove the recognition rate.