Color-Decoupled Photo Response Non-Uniformity for Digital Image Forensics
read more
Citations
An Overview on Image Forensics
Image Splicing Localization using a Multi-task Fully Convolutional Network (MFCN)
A Bayesian-MRF Approach for PRNU-Based Image Forgery Detection
Deep Learning for Deepfakes Creation and Detection: A Survey
Large-scale evaluation of splicing localization algorithms for web images
References
Scientific Charge-Coupled Devices
Digital camera identification from sensor pattern noise
Determining Image Origin and Integrity Using Sensor Noise
Exposing digital forgeries by detecting traces of resampling
Adaptive color plane interpolation in single sensor color electronic camera
Related Papers (5)
Frequently Asked Questions (16)
Q2. How long does it take to compute the similarity between two images of the same size?
For a system with a Pentium Core II 1.3G CPU and 3 GB RAM, it takes 0.526 seconds to compute the similarity between the PRNUs of two images of 2048 × 1536 pixels and 0.567 seconds to calculate the similarity between a pair of CDPRNUs of the same size.
Q3. What is the purpose of this experiment?
The purpose of this experiment is to demonstrate the capability of the proposed CD-PRNU in dealing with the colour interpolation noise, so geometrical transformations will not be applied in order to prevent biased evaluation from happening.
Q4. What is the main problem inherent to Eq. (4)?
The main problem inherent to Eq. (4) is that it involvesthe whole image plane, which contains both artificial andphysical components, in one noise residual extraction process.
Q5. What is the reason why the PRNU method can not detect the tampered area?
Since the tampered area is 60 × 80 pixels, approximately one quarter of the window, the method based on PRNU can perform no better than a random7guess.
Q6. What is the common method of obtaining 0I?
as mentioned earlier that the most common method [11, 14, 15, 18] of obtaining 0I is to apply the discrete wavelet transform followed by a Wiener filtering operation directly to the entire image The authorwithout differentiating physical components from artificial components and, as a result, allowing the interpolation noise in the artificial components to contaminate the real PRNU in the physical components.
Q7. What is the reason why the PRNU algorithm can not detect tampered regions?
Chen predicated in [11] that one quarter of the sliding window is the lower bound on the size of tampered regions that their algorithm can identify, and therefore areas smaller than this should be filtered in order to remove the falsely identified noise.
Q8. What is the way to extract the PRNU?
The authors have also proposed a simple, yet effective, colour-decoupled PRNU (CD-PRNU) extraction method, which can prevent the CFA interpolation error from diffusing from the artificial colour channels into the physical channels, thus improving the accuracy of the fingerprint.
Q9. Why is demosaicking a key deterministic process?
Due to the fact that demosaicking is a key deterministic process that affects the quality of colour images taken by many digital devices, demosaicking has been rigorously investigated [31, 32, 33, 35, 36].
Q10. What is the main reason for sensor pattern noise?
It is this uniqueness of manufacturing imperfections and non-uniformity of photo-electronic conversion that makes sensor pattern noise capable of identifying imaging sources to the accuracy of individual devices.
Q11. How do the authors get the CD-PRNU Wc of each colour channel?
Finally the CD-PRNU Wc of each colour channel c is formed by combining the four sub-noise residuals j,i,cW , , 0,1i j such that 222/,2/, ,, mod , mod , yjxiyxWyxW jicc (13)where 0, 1x X , 0, 1y Y and mod is the modulooperation.
Q12. Why is the interpolation noise P so equal to 1+P?
This is because, for theartificial components, the interpolation noise P is many ordersgreater than the PRNU K and K << 1, therefore (1+P)(1+K)is virtually equal to (1+P).
Q13. How many pixels are used in the sliding step/displacement?
in their experiment, the sliding step/displacement is set to 5 pixels in order to reduce the computational load without sacrificing the accuracy of the integrity verification.
Q14. What is the common CFA pattern?
Although the 2×2 Bayer CFA is the most common CFA pattern, to make the proposed CD-PRNU versatile and applicable to cameras adopting different CFA patterns, the authors makes no assumption about the CFA pattern, F, except that it is a 2 × 2 square array.
Q15. Why do the authors use the terms PRNU and noise residual?
Because the PRNU is formulated in Eq. (3) and (5) as a function of the noise residual W (i.e., Eq. (4)), in the rest of the work the authors will use the two terms, PRNU and noise residual, interchangeably whenever there is no need to differentiate them.
Q16. How do the authors know which colour components are in the DWT?
But bydecomposing cI into four sub-images, , ,c i jI , the authors know thateach of the four sub-images either contains only the physical colour or only the artificial colours.