摘要
当前所有的智能人脸识别系统都没有考虑到去身份识别问题,对去身份后人脸特征的极具减少无法建立准确的识别模型。文章提出了以模糊网络识别技术为基础的加入去身份识别功能的人脸识别系统设计方法,给出了硬件和软件的完整设计方案;分析了去身份信息后人脸的结构变化,根据不同少有的人脸节点间的关联进行运算,保证系统在逻辑上的识别完整性,规划去身份信息后,人脸特征在节点框架范围内的变化幅度,设定一个控制规则表,克服传统系统下由于身份特征被大量去除后无具体识别依据的弊端;以传统的智能识别系统为基础进行系统测试,系统在身份信息被去除后,基本能够识别人脸信息,识别精度在30%左右,有效弥补了传统系统在此方向的不足。
In order to accurate face recognition, verifying the rationality of human identity, the main goal of the traditional methods is ac- curate modeling, there are many differences with the actual status~ Traditional basic model control method can fulfill the effective identifica- tion of face features, but it can＇t get accurate membership functions. So fuzzy network control model is proposed in this paper, on the analysis of the structure of the model, the relationship between different nodes, and using the method of the model. Planning facial features change control rules table, face feature recognition controller by fuzzy network planning. Simulation experimental results show that compared with the traditional fuzzy control method, fuzzy network model can obtain accurate membership functions, and this function has a smooth surface characteristic, conforms to the face feature to identify the requirements of the relevant controller.
出处
《计算机测量与控制》
北大核心
2013年第9期2564-2566,2603,共4页
Computer Measurement &Control
基金
上海市教育委员会科研创新基金资助项目(11YZ282)
关键词
人脸识别系统
模糊网络
去身份识别
face recognition system
fuzzy network
membership functions
