Enhancing ROC Performance of Trustworthy Camera Source Identification
| Type of publication: | Inproceedings |
| Citation: | kang_etal_spie11 |
| Booktitle: | SPIE Conference on Media Watermarking, Security, and Forensics |
| Year: | 2011 |
| Location: | San Francisco, CA |
| Abstract: | Sensor pattern noise (SPN) extracted from digital images has been proved to be a unique fingerprint of digital camera. However, sensor pattern noises can be contaminated largely in frequency domain by image details from scenes according to Li's work and non-unique artifacts of on-sensor signal transfer, sensor design, color interpolation according to Chen et al's work, the identification performance needs to be improved especially for small image block. Motivated by their works, in order to lessen the effect of these contaminations, the unique SPN fingerprint for identifying one specific camera is assumed to be a white noise which has a flat frequency spectrum, so the SPN extracted from an image is whitened first to have a flat frequency spectrum, then is inputted to the mixed correlation detector. Camera source identification is the detection of the existence of photo-response non-uniformity (PRNU, also known as reference SPN) in the SPN extracted from an image. Compared with the correlation detection approach and Li's model based approaches on 8 cameras, 1600 photos total, each camera is responsible for 200, the experimental results show that the proposed mixed correlation detection enhance the receiver operating characteristic (ROC) performance of camera source identification, especially greatly raise the detection rate (true positive rate) in the case of trustworthy camera source identification which is with a low false positive rate. For example, the proposed mixed correlation detection raises the true positive rate from 69% to 94% at zero false positive rate on image blocks of 512x512 pixels cropped from the center of the 1600 photos. The existing two reference SPN extraction methods proposed by Fridrich's group are compared from the aspect of ROC performance. It is observed that the two methods have almost the same ROC performance, and they are equivalent to each other empirically. |
| Keywords: | |
| Authors | |
| Added by: | [ADM] |
| Total mark: | 0 |
|
Attachments
|
|
|
Code
|
|
|
|
|
|
Data
|
|
|
|
|
|
Notes
|
|
|
|
|