基于SVM的数字水印检测技术研究

A survey of digital watermarking detection technology based on SVM

在分析现有数字图像水印检测算法的优势和不足的基础上,提出了基于SVM的数字水印检测算法,提供了对图像特征向量进行归一化处理的方法,得到了特征向量约简的数据形式.实验结果表明,基于SVM的数字水印检测方法避免了相关检测算法中依赖于特定嵌入算法的缺点,可实现结构风险最小化,是有效判断图像中是否存在水印的通用检测方法.

Based on the analysis of the advantages and disadvantages of the existing digital image watermarking detection algorithms, this paper proposes the digital image watermarking detection algorithm based on SVM, and it provides a method for normalization of image feature vectors, then gets the reduction data of feature vectors. The experimental results prove that the digital watermarking detection algorithm based on SVM is a valid general method for image watermarking detection and realizes the structural risk minimization. It is free from the disadvantages in correlation detection algorithm that depends on the special embedding algorithm.