3. Status of effort: We have met and even exceeded our goals that we had outlined for our project. Our efforts have focused on two areas. The first area is LSB steganalysis for grayscale images. Here, as we had proposed (as a challenging task), we have generalized our previous steganalysis technique of sample pair analysis to a theoretical framework for the detection of the LSB steganography. The new framework exploits high-order statistics of the samples, and sheds some light on the effects of window size in steganalysis, an issue not well understood before.The second thrust of our efforts has been on steganalysis of binary documents. Last year we developed steganalysis methods to detect marked images hidden by following data hiding techniques:1. Boundary-based approaches 2. The Line, word, or character shifting 3. Fixed partitioning of image 4. Modification of character features 5. Modification of run-lengths 6. Modification of half-tone patternsIn addition to electronic documents we have also developed steganalysis methods to detect marked images when the original cover images are scanned/printed images. It is known that this is one of the most challenging problems in steganalysis literature. We have made significant progress and our work has laid the foundation to the goal of successfully detecting messages hidden in scanned/printed images. It should be noted that this goal was not part of the proposed work.4. Accomplishments/New Findings: Below we summarize the main accomplishments of our project:a. An exciting progress made is the generalization of our previous steganalysis technique of sample pair analysis to a theoretical framework for the detection of the LSB steganography. The new framework exploits high-order statistics of the samples, and sheds some light on the effects of window size in steganalysis, an issue not well understood before. We have developed new detection algorithms 20050608 055