Study On Robustness And Security Of Digital Image Source Detection

Digital image has gradually become an important part of people’s daily life and work,while the wide application of powerful image processing software makes it very easy tomodify digital image, resulting in a variety of forged images widely used, making peopledoubt about the authenticity of digital images. Under this situation, the digital image forensicstechnology has become a new hotspot in the field of digital image processing, and one of it’smain branch is digital image source detection, which is the research of how to get the devicenoise arising from the image device within the digital image and detect the source of digitalimage.In this paper, we will combine the theoretical analysis and the experiment to study therobustness and the security problems of digital image source detection algorithm based onsensor pattern noise. Firstly, we study the robustness problem by analysing the influence ofISO speed on the detection effect.Then we study security attack and counter-attack problem,and present an improved algorithm to affect the success of the counter-attack algorithm. Themajor work can be summarized as follows:1. An analysis of digital image source detection algorithm based on sensor pattern noiseand correlation detection is made, and then we do the simulation and analyze experimentalresults, which proves the deficiency of the algorithm on robustness and security.2. Study the robustness problem of digital image source detection algorithm based onsensor pattern noise from the view of ISO speed. When using a digital camera, ISO speed canaffect the quality of digital photos, so a detailed analysis of the influence of ISO speed ondigital image source detection algorithm is made for ordinary digital camera, digital SLRcamera and single electric camera. Experiments show that when the ISO speed is high, therobustness of the algorithm is poorer for the ordinary digital camera, but better for SLR andsingle electric camera.3. Study the security problem of digital image source detection algorithm based onsensor pattern noise. We describe the fingerprint-copy attack algorithm and the counter-attackmethod-triangle test, and propose a simplified sensor pattern noise enhancement algorithm toestimate victim’s camera sensor pattern noise for fingerprint-copy attack. Experiments showthat when an attacker uses sufficient number of images with simplified enhancementalgorithm to estimate victim’s camera sensor pattern noise for fingerprint-copy attack, thedetection effect of triangle test becomes poor or even invalid, achieving the aim of attackingthe triangle test.