Blind image clustering based on the Normalized Cuts criterion for camera identification

Amerini, Irene; Caldelli, Roberto; Crescenzi, Pierluigi; Mastio, Andrea Del; Marino, Andrea

doi:10.1016/j.image.2014.07.003

Cited by 42 publications

(30 citation statements)

References 28 publications

(24 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Reduce computational time and maintaining accuracy, with the number of cameras and images are increased digest-based descriptor is taken into account [13], [14]. Open set scenario is depicted in [15], [16] using enhanced version of PRNU to distinguish among images taken by unknown digital cameras. Extracting robust and characterizing features to distinguish among various classes of originating devices like scanned images, photos, and computer.…”

Section: ©Ijraset (Ugc Approved Journal): All Rights Are Reservedmentioning

confidence: 99%

Forensic Technique for Social Network Provenance Classification of Images

Dhanyaja¹

2018

IJRASET

View full text Add to dashboard Cite

The unprecedented popularity of various online social networks (OSNs) makes a rapid development to share digital contents online. However, misusing and dissemination of online contents widely happens. Under this circumstance, the identification of the origin and the propagation path of an online image are crucial for many forensic applications. It is a tough task to trace out background details and pre processing of digital image by forensic scientists. Strategic way to address a problem is to simply find image's history: knowing acquisition device and model of camera etc. This paper enlightens image classification based on originating social network. Since image has distinctive traces by social network. Manipulation process will be unique for each social network. Social network provenance based image classification is done through resorting at a trained multi-SVM classifier. Experimental results administrated are distinction attainable on numerous image datasets and in varied operative conditions. Additionally, technique is to go back to the initial JPEG quality image had before being uploaded on a social network.

show abstract

Section: ©Ijraset (Ugc Approved Journal): All Rights Are Reservedmentioning

confidence: 99%

Forensic Technique for Social Network Provenance Classification of Images

Dhanyaja¹

2018

IJRASET

View full text Add to dashboard Cite

show abstract

“…To overcome the infeasibility of the manner of determining the optimal cluster number in [31], Amerini et al [45] proposed a blind SPN clustering algorithm based on normalized cut criterion [46]. Similar to [31], SPNs are considered as the vertices in a graph and the weight of each edge measures the similarity between the two vertices connected by the edge.…”

Section: The Challenges Of Source-oriented Image Clustering and Relatmentioning

confidence: 99%

“…To illustrate the advantages of our proposed algorithm, we compared it with other five clustering methods: (1) the multi-class spectral clustering (SC) method [31], (2) the hierarchical clustering (HC) method [34], (3) the shared nearest neighbor clustering (SNNC) method [43], (4) the normalized cut-based clustering (NCUT) method [45], and (5) the Markov random field based clustering (MRF) method [49]. We did not include Boly's algorithm [44] and Marra's algorithm [47], because both algorithms retain the fingerprints in the RAM for updating the centroids of clusters, which makes them unsuitable for relatively large datasets.…”

Section: Comparisons and Analysesmentioning

confidence: 99%

“…We set κ to the same value as our proposed algorithm, i.e., κ = 15, and set Eps and MinPts to 2 and 10, respectively. For NCUT, we set the aggregation threshold T h to 0.0037 rather than the 0.037 in [45], which results in many singleton clusters. For MRF, to avoid going into infinite iterations when the algorithm does not converge, we set the maximum number of iterations to 50.…”

Section: Comparisons and Analysesmentioning

confidence: 99%

See 1 more Smart Citation

A fast source-oriented image clustering method for digital forensics

Lin

2017

J Image Video Proc.

View full text Add to dashboard Cite

We present in this paper an algorithm that is capable of clustering images taken by an unknown number of unknown digital cameras into groups, such that each contains only images taken by the same source camera. It first extracts a sensor pattern noise (SPN) from each image, which serves as the fingerprint of the camera that has taken the image. The image clustering is performed based on the pairwise correlations between camera fingerprints extracted from images. During this process, each SPN is treated as a random variable and a Markov random field (MRF) approach is employed to iteratively assign a class label to each SPN (i.e., random variable). The clustering process requires no a priori knowledge about the dataset from the user. A concise yet effective cost function is formulated to allow different "neighbors" different voting power in determining the class label of the image in question depending on their similarities. Comparative experiments were carried out on the Dresden image database to demonstrate the advantages of the proposed clustering algorithm.

show abstract

“…Another interesting scenario that has received some attention in recent literature is using the PRNU to classify a set of images according to the device that acquired them [5]- [7]. However, the proposed technologies have not been applied to very large databases so far [8].…”

Section: Introductionmentioning

confidence: 99%

Large-Scale Image Retrieval Based on Compressed Camera Identification

Valsesia

Coluccia

Bianchi

et al. 2015

IEEE Trans. Multimedia

View full text Add to dashboard Cite

Retrieving pictures from large collections according to a specific criterion is an increasingly relevant task. An important, but so far overlooked, such criterion is the retrieval of pictures acquired by a specific camera. Instead of relying on metadata, which can be absent or easily manipulated, a forensic tool is exploited, namely the photo response non-uniformity (PRNU) of the camera sensor. Recent works showed that random projections can be used to significantly compress the PRNU, enabling operation on very large scales, previously impossible due to the size of the PRNU and to the complexity of the matching operations. In this paper, we propose efficient techniques for management and retrieval of images employing the PRNU, and test them on a database of 1174 cameras and half a million pictures downloaded from the Internet.Index Terms-Image forensics, image search and retrieval, photo response non-uniformity (PRNU), random projections.

show abstract

Blind image clustering based on the Normalized Cuts criterion for camera identification

Cited by 42 publications

References 28 publications

Forensic Technique for Social Network Provenance Classification of Images

Forensic Technique for Social Network Provenance Classification of Images

A fast source-oriented image clustering method for digital forensics

Large-Scale Image Retrieval Based on Compressed Camera Identification

Contact Info

Product

Resources

About