Steganalysis of binary images.
Date of Issue2008
School of Electrical and Electronic Engineering
With an increasing use of digital binary images including text, graphic and halftone images in daily life, security of the digital binary images such as certificate, transcript, scanning hard copy to electronic copy etc. is a big concern. Such a concern makes data hiding in binary images an important and attractive research topic in recent years. With the development of the data hiding techniques for binary images, steganalysis of binary images starts receiving attention too. The concern of using binary image data hiding techniques by terrorists for conspiracy makes the research study of steganalysis of binary images necessary and important. In this thesis, we firstly give an overview of steganography and steganalysis, followed by the introduction of the proposed steganalysis techniques for different types of binary images. Two techniques are introduced for the detection of hidden data in clean and scanned text binary images. An objective distortion measure for binary text and binary graphic images based on edge line segment similarity is proposed and used together with other distortion measures for steganalysis. Filtering based and projection based inverse halftoning methods are applied in the steganalysis of halftone binary images and their performances in detection of hidden data are compared. Finally, conclusions and recommendations for future work are given. Detection of data hidden is usually done by examining some features of the images, such as image wavelet statistics, image block statistics which are affected by data hiding. Thus most of steganalysis work is to find out these features and utilize them in the detection. Our focus is on how the embedding processes distort the images, not only visual distortion, but also the distortion to the image features such as the regularities. Exploring the regularities of the images is not only useful for steganalysis, but also beneficial for minimizing the distortions as destroying the regularities of the images may often introduce visible distortion.
DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing