Individual camera identification using correlation of fixed pattern noise in image sensors

This paper presents results of experiments related to individual video camera identification using a correlation coefficient of fixed pattern noise (FPN) in image sensors. Five color charge-coupled device (CCD) modules of the same brand were examined. Images were captured using a 12-bit monochrome video capture board and stored in a personal computer. For each module, 100 frames were captured. They were integrated to obtain FPN. The results show that a specific CCD module was distinguished among the five modules by analyzing the normalized correlation coefficient. The temporal change of the correlation coefficient during several days had only a negligible effect on identifying the modules. Furthermore, a positive relation was found between the correlation coefficient of the same modules and the number of frames that were used for image integration. Consequently, precise individual camera identification is enhanced by acquisition of as many frames as possible.