人物身份认证视觉图像三维重构仿真研究  

Simulation of 3D Reconstruction of Visual Image of Character Identity Authentication

作　　者：马禄[1] 杨勃[1] 邓婷[1] 

机构地区：[1]银川能源学院信息传媒学院,宁夏银川750105

出　　处：《计算机仿真》2017年第9期288-291,共4页Computer Simulation

摘　　要：对动态视觉图像进行三维重构,能够有效保留动态人物图像信息。对人物身份认证视觉图像的三维重构需要以图像参考模板与视觉图像做最小二乘图像匹配,得到源图像对应的低高频分量,完成对视觉图像的三维重建。传统方法先对动态图像进行去噪处理,对图像尺度固定特征点进行标定匹配,但无法获得源图像的低高频分量,导致重构精度低。提出基于小波变换的人物身份认证视觉图像三维重构方法。结合Canny边缘检测算子计算边缘强度和边缘梯度方向,确定动态图像对应的边缘点特征,以给定的图像直线特征作为参考模板与实际动态图像做最小二乘图像匹配,结合小波转换理论对各幅待融合动态图像进行小波分解,得到源图像对应的低高频分量,以此完成对视觉环境下动态图像的三维重构。实验结果表明,所提方法能够有效够保留图像大部分细节信息,图像重建效果好。A three-dimensional （3D） reconstruction method is proposed for visual image of character identity au- thentication based on wavelet transform. The direction of edge strength and gradient integrated with Canny edge detec- tion operator is calculated to confirm feature of edge point corresponding to dynamic image. Image matching with least square is carried out to reality dynamic image by using given linear feature of image as reference template. The wave- let decomposition to each dynamic image waiting for fusion integrated with wavelet transform is also executed, then the high and low frequency component is obtained. Thus, the 3D reconstruction is completed. Simulation results show that the method can reserve major detail information of image effectively.

关 键 词：人物身份认证 视觉图像 图像三维重构 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]