Study On The Mathematical Model In The Digital Image Forensic Technology

Digital image blind forensics technology is aimed at identifying the authenticity of digital image without embedding digital watermarking in advance. With the flood of image tampering, the active forensics technology has become helpless. However, the digital image blind forensics technology by means of analysing inherent property of image has become a hot research topic in the field of applied mathematics and information security.This paper discussed the digital image forensic technology in detail, including the research background, the research significance, the hot research point, and the present research situation at home and abroad. On the basis of the existing problems, this paper systematically expounded the image forensic technology from the following aspects:exposing digital image forgeries by detecting inconsistence in the light source direction in infinite light source model. Professor Micah K.Johnson first brought the estimation problem of the light source direction into the digital image forensic technology, furthermore, in this paper we introduced the application of the image edge detection operator, realized the accurate positioning of the object in image, edge fitting and data analysis, and set up a mathematical model based on the estimation of the light source direction. In this model, the successful use of KKT condition could be able to obtain the light direction exactly, through the inconsistence on the light source direction of different objects in the tampered image from their respective shooting scene, so the inconsistencies of light direction could be used as the evidence of image tampering. In a large measure, the establishment of this forensic model greatly improved the feasibility and expansibility of this technology.In addition, this paper elaborated the research on exposing digital image forgeries by detecting inconsistencies on the camera parameter in the camera imaging geometry model, established the mathematical model of solving the projection matrix from a single digital image, used the real image through the camera calibration method to determine the internal and external camera parameters, at the same time the flexible application of camera calibration technique provided an excellent initial value.