Source attribution based on physical defects in light path

Abstract

Recent studies in multimedia forensics show that digital images contain intrinsic patterns, traces, and marks generated by imaging pipeline components (sensor) and processes (demosaicing and color adjustment). Some of these patterns and marks, such as photo response non-uniformity noise (PRNU), are unique to individual component characteristics of imaging system. Similar to PRNU noise, physical defects in imaging pipeline such as dust particles in DSLR camera chamber, scratches on flatbed scanners also generate unique patterns in image output. Due to unique and random nature of these patterns, they can be utilized in digital image forensics problems. In this chapter, we will give an overview of state-of-the-art camera identification techniques which utilize such defects and patterns.