视觉传感网络中身份特征自适应识别算法改进  

Improvement of identity adaptive recognition algorithm in visual sensor network

作　　者：李娜[1] 张晓宁[1] 朱芳娥 

机构地区：[1]石家庄铁道大学四方学院计算机系,河北石家庄050011

出　　处：《电信科学》2016年第6期110-115,共6页Telecommunications Science

摘　　要：在对视觉传感网络中身份特征进行识别时,容易受到人脸表情、光照条件及遮挡等干扰,降低了身份特征识别精度。提出了一种基于改进最小灰度差树的身份特征自适应识别算法。对待识别图像进行灰度处理后,利用最小灰度差数增强待识别图像的质量;定义基于灰度的代价函数,获取待识别人脸图像和指定人脸图像对应的各灰度对的匹配代价,建立最小灰度差树模型,计算两幅图像相似度后,直接采用最近邻匹配算法获取和视觉传感网络注册图库中最小匹配代价对应的图像身份,将其看作待识别身份,实现视觉传感网络中身份特征自适应识别。仿真实验结果表明,所提算法具有很高的身份识别精度。When recognizing the identity in visual sensor network, it ＇s easily to be interfered with facial expression,illumination condition and shelter, so as to reduce the recognition accuracy. An identity adaptive recognitionalgorithm based on the improved minimum gray difference tree in visual sensor network was put forward. After grayprocessing, the minimum gray difference was used to enhance the quality of the image; cost function based on graylevel was defined, the match price of each corresponding pair of gray of the image to be recognized and thespecified face image were achieved, minimum gray difference tree model was set up, after two image similarity werecalculated, the nearest neighbor matching algorithm was directly applied to obtain the image identity correspondingto minimum matching cost in the visual sensor network registration gallery, it was viewed as the identity to berecognized, the adaptive identity recognition was achieved in visual sensor network. Simulation results show that theproposed algorithm has high identification accuracy.

关 键 词：视觉传感网络 身份特征 自适应识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]