摘要
该文提出一种基于属性论的人脸识别方法，主要应用于提高复杂背景下的多姿态人脸检测的识别效果。首先介绍了属性论的基石——定性映射理论，讨论了定性映射、经典集合论和特征函数之间的关系，然后讨论了基于定性映射的模式生成与识别模型，经过对人脸及人脸的特征点进行定位后，即可利用属性计算网格的定性基准网格对人脸的各特征点进行建模。研究发现基于属性论的人脸识别算法，识别率较高。在复杂背景下的多姿态人脸检测（如侧面人脸、遮挡脸）仍然取得较好的效果。
Propose a face recognition method based on attribute theory, Is mainly used in the context of complex multi-Pose Face Detection recognition effect.. First introduced the cornerstone of the theory of property - qualitative mapping theory, Discussed the qualitative mapping, classical set theory and the relationship between the characteristic function, Followed by a discussion of the model based on qualitative mapping generation and recognition model, After a human face and facial feature points to locate, you can use the qualitative proper- ties of computational grid reference grid on the face of all feature points for modeling. Study found that face recognition algorithm based on attribute theory, recognition rate is higher. In a complex background of multi-profile face detection （such as the side face, obscured face） still achieved good results.
作者
张萍
冯嘉礼
ZHANG Ping, FENG Jia-li （Shanghai Maritime University, Information Engineering College, Shanghai 200135, China）
出处
《电脑知识与技术》
2010年第4期2430-2432,2434,共4页
Computer Knowledge and Technology
关键词
属性论
定性映射理论
属性计算网格
attribute theory
qualitative mapping theory
property computational grid
