基于三维行为数据匹配的身份识别方法  

Identification Method Based on 3D Behavior Data Matching

作　　者：李昆仑[1] 汪滢[1] 熊婷[1] 

机构地区：[1]南昌大学科学技术学院,江西南昌330029

出　　处：《计算机仿真》2016年第8期403-406,410,共5页Computer Simulation

基　　金：南昌大学科技学院基金项目(项目编号2014-JG-04);南昌大学科技学院精品课程项目(项目编号2013JPKC020)

摘　　要：针对传统的身份识别方法由于在对采集的图像进行去噪处理时也最大程度的去除了行为前景特征,使得行为特征形式无法统一,导致身份识别不准确的问题。提出采用三维行为数据匹配的身份识别方法,通过均值滤波法对图像进行去噪处理。使用人体行为图像的一阶差分,求出各像素处的平均平方梯度矩阵,依据特征值给出角点响应确定Harris角点,利用Harris角点对人体行为特征进行提取。并将两个完全相同的摄像机构成的两个图像平面置于同一个平面上,模仿人眼获得物体位置信息的原理,获取人体行为数据的三维坐标及全部特征点的三维坐标。通过计算搜索窗口内对应点的相关系数和阈值的比较来实现特征点匹配,完成身份识别。仿真结果表明,所提方法相比传统方法具有很高的识别精度及识别效率。An identity recognition method using 3D behavior data matching is proposed. Image denoising is carried out by means of the mean filter method. By using the first order difference of the human behavior image,the average square gradient matrix of each pixel is obtained. According to the characteristic value,the angular point response is obtained,and the Harris angular point is determined. Harris angular point is used to extract human behavior features. And two image planes constituted by two identical cameras are placed in the same plane,to imitate principle of the human eye to obtain the object position information,and obtain the 3D coordinates of the human behavior data and the 3D coordinates of all feature points. By computing the correlation coefficient and the threshold value of the corresponding point in the search window,the feature points matching is realized,and the identity recognition is completed. Simulation results show that,compared with the traditional method,the proposed method has a high recognition accuracy and recognition efficiency.

关 键 词：三维行为 数据匹配 身份识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]