Take a look at the new version of this page: [ beta version ]. Tell us what you think.

ABSTRACT

The identification of source camera is useful to improve the capability of evidence in the digital image such as distinguish the photographer taking illegal images and adopting digital images as evidence of crime. Lukáš, et al. showed the method for source camera identification based on the correlation of PNU (pixel nonuniformity) noise. However, the wavelet-based denoising filter for suppressing the random noise reduces the accuracy of camera identification. It is caused by the fact that the denoising filter diffuses the edge and makes the PNU noise less pronounced. Moreover, it is difficult to extract PNU noise from the images taken by cameras which are equipped with the image improvement functions such as motion blur correction, contrast enhancement, and noise reduction. In this paper, we propose a method for improving the camera identification accuracy by selecting pixels based on the texture complexity. We also propose a-method for improving the identification accuracy by applying the image restoration method.

REFERENCES

• 1: J Lukas, et al. "Digital Camera Identification from Sensor Pattern Noise", IEEE Trans. IFS, June 2006
• 2: M. Kivanc Mihcak , I. Kozintsev , K. Ramchandran, Spatially adaptive statistical modeling of wavelet image coefficients and its application to denoising, Proceedings of the Acoustics, Speech, and Signal Processing, 1999. on 1999 IEEE International Conference, p.3253-3256, March 15-19, 1999 [doi>10.1109/ICASSP.1999.757535]