Abstract:

“Seeing is believing.” The old saying seems no longer be true in this digital era. Nowadays, digital image is very popular and can be obtained almost everywhere.  With modern image processing tools, digital image can be easily modified without any visual clue. There is an increasingly concern to determine whether an image is original or not, especially for forensic purpose. Two problems of the digital image forensics issue are addressed in this proposal: (1) JPEG (Joint Photographic Experts Group) image steganalysis and (2) passive image tampering detection.

Steganography attempts to hide information in such a way that no one else would be aware of the existence of the hidden information except the intended recipient. The aim of steganalysis, on the other hand, is to detect the very existence of information hidden by steganography. Digital image is an ideal cover media for steganography yet a big challenge for steganalysis. Since the JPEG format is the most dominant image format for image storage and exchange at this time, the JPEG steganographic techniques have attracted more and more attention. This proposal presents a novel steganalysis scheme to effectively attack the JPEG steganographic schemes. The proposed method exploits the correlations between block-DCT coefficients in both intra-block and inter-block sense. Markov empirical transition matrices are used to capture these correlations. The experimental results demonstrate that the proposed scheme outperforms the state-of-the-arts in detecting the modern steganographic methods for JPEG images.

Image tampering refers to the malicious manipulation of images to mislead the observers. Image splicing is one of the common used image tampering schemes, which joints two or more images together to construct a scene actually never exists. The image splicing process is generally considered as a highly nonlinear and non-stationary process. This proposal presents a passive image splicing detection scheme, which is effective in analyzing nonlinear and non-stationary signals. The experimental results demonstrate that the proposed scheme outperforms the existing methods.

Committee Members:

Dr. Yun Q. Shi, Professor, ECE Dept., NJIT (Advisor)

Dr. Nirwan Ansari, Professor, ECE Dept., NJIT

Dr. Swades K. De, Assistant Professor, ECE Dept., NJIT

Dr. Frank Y. Shih, Professor, CS Dept., NJIT

Dr. Edward K. Wong, Associate Professor, CIS Dept., Polytechnic University