Inpainting-Based camera anonymization

Mandelli, Sara; Bondi, Luca; Lameri, Silvia; Lipari, Vincenzo; Bestagini, Paolo; Tubaro, Stefano

doi:10.1109/icip.2017.8296536

Cited by 9 publications

(28 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The first dataset is the Dresden Image Database [17], which collects both uncompressed and compressed images from more than 50 diverse devices. Following the same procedure done in past works proposed in literature [13], [12], we select images from 6 different camera instances, precisely Nikon D70, Nikon D70s, Nikon D200, two devices each. Second dataset is the recently released Vision Dataset [18], which includes JPEG compressed images captured from 35 devices.…”

Section: A Datasetsmentioning

confidence: 99%

Section: B State-of-the-art Solutionsmentioning

confidence: 99%

“…Regarding PRNU-blind strategies, the most recent contribution is that proposed by [13], which demonstrates to outperform results of [12] in a PRNU-blind scenario. For the implementation of [13], we consider the parameter configurations achieving the best anonymization results, i.e., the strategies defined as (5) 1 in the original paper. Moreover, to show that simple denoising does not achieve good anonymization performances [10], we implement the well-known DnCNN denoiser [30] which represents a modern data-driven solution among image denoising strategies.…”

Section: B State-of-the-art Solutionsmentioning

confidence: 99%

“…Alternatively, [11] applies seam-carving to change pixel locations, and [12] exploits patch-match techniques to scramble pixel positions. More recently, [13] proposes an inpaintingbased method which deletes and reconstructs image pixels such that final images are anonymized with respect to their source PRNU.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

DIPPAS: A Deep Image Prior PRNU Anonymization Scheme

Picetti¹,

Mandelli²,

Bestagini³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Source device identification is an important topic in image forensics since it allows to trace back the origin of an image. Its forensics counter-part is source device anonymization, that is, to mask any trace on the image that can be useful for identifying the source device. A typical trace exploited for source device identification is the Photo Response Non-Uniformity (PRNU), a noise pattern left by the device on the acquired images. In this paper, we devise a methodology for suppressing such a trace from natural images without significant impact on image quality. Specifically, we turn PRNU anonymization into an optimization problem in a Deep Image Prior (DIP) framework. In a nutshell, a Convolutional Neural Network (CNN) acts as generator and returns an image that is anonymized with respect to the source PRNU, still maintaining high visual quality. With respect to widely-adopted deep learning paradigms, our proposed CNN is not trained on a set of input-target pairs of images. Instead, it is optimized to reconstruct the PRNUfree image from the original image under analysis itself. This makes the approach particularly suitable in scenarios where large heterogeneous databases are analyzed and prevents any problem due to lack of generalization. Through numerical examples on publicly available datasets, we prove our methodology to be effective compared to state-of-the-art techniques.

show abstract

Section: A Datasetsmentioning

confidence: 99%

Section: B State-of-the-art Solutionsmentioning

confidence: 99%

Section: B State-of-the-art Solutionsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

DIPPAS: A Deep Image Prior PRNU Anonymization Scheme

Picetti¹,

Mandelli²,

Bestagini³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…In the case of seam carving, it was subsequently shown that when multiple seam carved images are available from the same camera, successful verification could still be done by increasing alignment between the camera fingerprint and the seam-carved images [10], [11], provided no additional operation such as scaling and cropping has been performed. In-painting [12], Patch-based desynchronization [13], and image stitching [14] are other examples of complex techniques for breaking alignment.…”

Section: Introductionmentioning

confidence: 99%

Source Camera Attribution of Multi-Format Devices

Taspinar,

Mohanty,

Memon

2019

Preprint

View full text Add to dashboard Cite

Photo Response Non-Uniformity (PRNU) based source camera attribution is an effective method to determine the origin camera of visual media (an image or a video). However, given that modern devices, especially smartphones, capture images, and videos at different resolutions using the same sensor array, PRNU attribution can become ineffective as the camera fingerprint and query visual media can be misaligned. We examine different resizing techniques such as binning, lineskipping, cropping and scaling that cameras use to downsize the raw sensor image to different media. Taking such techniques into account, this paper studies the problem of source camera attribution. We define the notion of Ratio of Alignment, which is a measure of shared sensor elements among spatially corresponding pixels within two media objects resized with different techniques. We then compute the Ratio of Alignment between the different combinations of three common resizing methods under simplified conditions and experimentally validate our analysis. Based on the insights drawn from the different techniques used by cameras and the RoA analysis, the paper proposes an algorithm for matching the source of a video with an image and vice versa. We also present an efficient search method resulting in significantly improved performance in matching as well as computation time.

show abstract