基于仿生模式人脸识别_琚生根.pdfVIP

2008 年2 月 四川大学学报( 自然科学版) Feb. 2008 第 5 卷第1 期 Journal of Sichuan U niversity ( Natural Science Edition) Vol. 5 No. 1 : 0 90-6756(2008) 01-0065-06 1 1 1 1, 2 琚生根 , 周激流 , 何 坤 , 王 刚 ( 1. , 610065; 2. , 6 6000) : 我们提出了基于/ 认识0 事物的人脸识别. 以同类样本全体的连续性规律为基本出发 点把人脸识别 成人脸/ 认识0 的问题而不是人脸分类划分, 与以/ 最佳划分0 为目标的传统统 计模式识别相比, 本文的方法更接近于人类/ 认识0 事物的特性. 同一人脸的分布在低维空间 中具有一定的类聚性而不同人脸的覆盖范围相互交织. 随着空间维数增加, 同类样本内聚性减 弱, 不同类样本互斥性增强; 空间维数继续增加, 同类样本内聚性与不同类样本互斥性反而都 会减弱. 将待识别人脸在某一空间进行覆盖范围的识别. 若属于多 个侯选人脸覆盖范围, 进而 运用Fisher 人脸识别出最终结果. 对ORL 人脸库实验表明, 对于未经过训练的任意对象不会 误识, 识别率为97. 5% . : 模式识别; 仿生学; 连续性规律; 多项式核函数 : TP391. 1 : A Face recognition on bionic pattern 1 1 1 1, 2 J U Sheng-Gen , ZH OU Ji-L iu , H E K un , WA N G Gang ( 1. College of Computer Science, Sichuan University, Chengdu 610065, China; 2. Sichuan Police College, Luzhou 6 6000, China) Abstract: A / matter cognition0 based face recognition model has been proposed. By taking continuity rule of samples of a same class as the starting point, face pattern recognition is considered as face pattern cognition in- stead of its classification. Compared w ith traditional best classification goaled statistic pattern recognition, it. s more similar to the character of human cognition. A person. s face distribution in low dimension space has a certain kind of cohesion, w hile face coverage of different people overlap. By the increase of space dimension, the cohesion of samples of a same class decrease, w hile the repel of samples of different classes increases. But as the increase of sp