摘要
针对纹理防伪标签在光线不足、视力不佳的情况下难以鉴别的问题,提出了一种基于DCT、感知哈希和视觉特征向量的纹理防伪标签鉴别算法。首先通过DCT和感知哈希提取图像的特征向量;然后建立特征数据库,通过特征向量之间的归一化相关系数(NC)实现纹理防伪标签的自动鉴别。实验结果表明,该方法可以有效抵抗常规攻击和几何攻击,可实现自动鉴别技术。与现有的纹理防伪技术相比,该算法可以减小占据的数据库存储空间,具有更好的实用性。
In view of the problem that texture anti-counterfeiting tags were difficult to identify to the naked eye in the case of poor eyesight or insufficient light,this paper put forward an algorithm based on DCT,perceptual hashing and human visual characteristics for the identification of texture anti-counterfeiting tags. Firstly,the method extracted the visual feature vectors of texture image through DCT and perceptual hashing,and then created a database of texture images’ feature vectors. At last the scheme compared the normalized cross-correlation（ NC） between the feature vectors to realize automatic identification of texture tags. The results of experimental study demonstrate that the proposed algorithm has strong robustness against common and geometric attacks. In addition,compared with the existing texture anti-counterfeiting technology,this algorithm can reduce the database storage space occupied and has better practicability.
出处
《计算机应用研究》
CSCD
北大核心
2014年第12期3734-3737,3741,共5页
Application Research of Computers
基金
国家自然科学基金资助项目(61263033)
海南省自然科学基金资助项目(60894)
关键词
纹理防伪标签
感知哈希
离散余弦变换
特征向量
自动鉴别
texture anti-counterfeiting tag
perceptual hashing
DCT
feature vector
automatic identification
