CAJ | 학술논문

为了提高人脸识别率和效率，提出一种改进局部方向模式特征的人脸识别算法。首先将人脸图像分割成若干不重叠的子块，采用改进局部方向模式算法提取每个子块特征，然后对所有子块的特征进行连接，构成人脸图像的特征向量，最后采用最小二乘支持向机对人脸图像进行识别。在多个人脸库上进行仿真实验，结果表明，该算法获得了比传统算法更高的人脸识别率，而且加快了运行时间，较好地满足人脸识别实时性要求。
위료제고인검식별솔화효솔，제출일충개진국부방향모식특정적인검식별산법。수선장인검도상분할성약간불중첩적자괴，채용개진국부방향모식산법제취매개자괴특정，연후대소유자괴적특정진행련접，구성인검도상적특정향량，최후채용최소이승지지향궤대인검도상진행식별。재다개인검고상진행방진실험，결과표명，해산법획득료비전통산법경고적인검식별솔，이차가쾌료운행시간，교호지만족인검식별실시성요구。
In order to improve face recognition rate and efficiency,we proposed a novel face recognition algorithm which improves the fea-tures of local directional pattern.First,the algorithm divides the face image into non-overlapping sub-blocks,and extracts the feature of each block by improving local direction mode algorithm,and then connects the features of all sub-blocks to form feature vectors of face image,fi-nally,it uses least square support vector machine to recognise the face image.Simulation experiment is carried out on various face databases, results show that the proposed algorithm achieves higher face recognition rate than traditional algorithms,and speeds up the operation time as well,it well meets the real-time requirement of face recognition.