基于变换域的纹理防伪标签自动鉴别算法  

Automatic Identification Algorithm of Texture Anti-Counterfeiting Tag Base on Transform Domain

作　　者：李雨佳[1] 李京兵 

机构地区：[1]海南大学信息科学技术学院,海南海口570228

出　　处：《计算机仿真》2014年第12期265-269,共5页Computer Simulation

基　　金：国家自然科学基金(61263033);海南省自然科学基金(60894)

摘　　要：研究纹理防伪标签自动识别问题,由于纹理防伪标签人眼鉴别比较困难、耗时较长,不能实现自动鉴别的问题,提出一种基于变换域的纹理防伪标签自动鉴别算法,用于实现纹理防伪标签的自动鉴别,可自动鉴别标签真伪。首先对纹理防伪标签图像进行DWT和DFT,在变换域中提取一个视觉特征向量,并存入数据库,最后通过特征向量之间的归一化相关系数来鉴别。如何提取一个抗几何攻击的特征向量是算法的难点。实验结果表明,改进方法可以有效抵抗几何攻击,可实现纹理防伪标签的自动鉴别。与现有的纹理防伪技术相比,改进算法可减小占据的数据库存储空间,缩小鉴别时间,有更好的灵活性和实用性。The identification of texture anti -counterfeiting tags is complicated and time -consuming longer by naked. There is no algorithm to achieve automatic identification. In this paper, an algorithm based on transform do- main is presented for the identification of anti - counterfeiting tags. DWT and DFT are applied to the original texture image, we can obtain the feature vector in transform domain and store it to the database. At last we compare the Nor- malized Cross - correlation （NC） between the feature vectors. The experimental results demonstrate that the proposed algorithm has strong robustness against common and geometric attacks, it can realize automatic identification. In addi- tion, though comparing with the existing texture anti - counterfeiting technology, this algorithm can reduce the data- base storage space occupied and narrow the identification time. It has better flexibility and practicability.

关 键 词：纹理防伪标签 自动鉴别 变换域 特征向量 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]