Camera identification based on very low bit rate videos with overall noise pattern having time varying statistics

This paper proposes a method for performing the camera identification based on very low bit rate videos with the overall noise patterns having time varying statistics. First, the overall noise pattern of each frame of each video is converted to a vector. Then, the odd order statistic moments of these vectors are computed. By performing the principal component analysis, only the most major component of each statistic moment vector is attained....