Digital Forensic Techniques for Graphic Data
MetadataShow full item record
With rapid development of hardware devices and software programs, a large amount of graphic data has been brought to or generated in digital domain, and become increasingly more widely used in our everyday life. Due to the ease of editing and distributing graphic data in the digital domain, protecting graphic data from such fraudulent operations as malicious tampering and unauthorized copying is becoming a major concern. The primary motivation of this dissertation search is to develop novel forensic techniques for digital graphic data to facilitate its proper distribution, authentication, and usage. We investigate two complementary mechanisms for performing forensic analysis on graphic data, namely, the extrinsic and intrinsic approaches. In the extrinsic approaches, we seamlessly embed into graphic data extrinsic watermarks/fingerprints, which shall later be extracted for verifying authenticity or tracing leak of the graphic data. By utilizing such extrinsic techniques via data embedding, we have studied robust digital fingerprinting for curve-based graphics such as topographic maps and drawings, in which a unique ID referred to as a digital fingerprint is robustly embedded for tracing traitors. Through proper transformations between 2-D contour curves and 3-D digital elevation maps, we have also developed an effective fingerprinting technique for digital elevation maps. In order to authenticate such graphic data as critical document and signature images, we have investigated high-payload watermark embedding for binary images, whose authenticity shall be decided through verifying integrity of the hidden watermark. In the intrinsic approaches, since scanners are a major kind of apparatus to capture graphic data, we develop a new technique of utilizing intrinsic sensor noise features for non-intrusive scanner forensics to verify the acquisition source and integrity of digital scanned images. We extract statistical features of scanning noise from scanned image samples, and construct a robust scanner identifier to determine the model of the scanner used to capture a scanned image. We further broaden the scope of acquisition forensics to differentiating scanned images from camera taken images and computer generated images, as well as perform integrity forensic analysis on scanned images using the proposed noise features, including detecting post-processing operations after scanning, and implementing steganalysis on scanned images.