图像隐写分析技术综述

摘要

通过归纳典型专用隐写分析方法和通用隐写分析方法的机制,指出在该领域中,低嵌入率的检测问题、图像源不匹配问题和隐写分析方法的适用性问题是3个亟待解决的问题,进而提出基于富模型和数字取证的隐写分析是两大研究趋势,前者合并不同域的差异特征后,利用集成分类器区分载体和含密图像,后者先用数字取方法证识别图像的类型,再采用该类的隐写分析器检测图像,由此克服图像源不匹配问题,提高检测性能.%This paper summarizes the schemes of typical targeted and universal steganalysis methods,and points out three challenges in this area:the detection of low embedding rate,the mismatch between the training image source and the test image source,and the ability to adapt unknown steganography algorithms.And it shows that steganalysis based on rich model and digital forensics are two research trends in the field.The former merges different features from different domains and then distinguishes between cover and stego-images by using ensemble classifier.The latter identifies the image type by using digital forensics in advance,and then detects the image by utilizeing stegoanalysizer of the corresponding type.So that the problem is solved and performance of detection is improved.