Source Camera Identification Using Support Vector Machines

Bo, Wang; Kong, Xiangwei; You, Xinge

doi:10.1007/978-3-642-04155-6_8

Cited by 15 publications

(6 citation statements)

References 14 publications

(25 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…SVM-based algorithms are capable of dealing with calibration issues, which fails Hadamard’s criteria for the uniqueness and stability of mathematical solutions, in addition to producing robust models for nonlinear spectral variations. − Figure B,C shows the results of PLSDA and SVMDA. The plots are an illustration of the predicted class (caliber) for each experimental spectrum.…”

Section: Resultsmentioning

confidence: 99%

Raman Spectroscopic Analysis of Gunshot Residue Offering Great Potential for Caliber Differentiation

2012

View full text Add to dashboard Cite

Near-infrared (NIR) Raman microspectroscopy combined with advanced statistics was used to differentiate gunshot residue (GSR) particles originating from different caliber ammunition. The firearm discharge process is analogous to a complex chemical reaction. The reagents of this process are represented by the chemical composition of the ammunition, firearm, and cartridge case. The specific firearm parameters determine the conditions of the reaction and thus the subsequent product, GSR. We found that Raman spectra collected from these products are characteristic for different caliber ammunition. GSR particles from 9 mm and 0.38 caliber ammunition, collected under identical discharge conditions, were used to demonstrate the capability of confocal Raman microspectroscopy for the discrimination and identification of GSR particles. The caliber differentiation algorithm is based on support vector machines (SVM) and partial least squares (PLS) discriminant analyses, validated by a leave-one-out cross-validation method. This study demonstrates for the first time that NIR Raman microspectroscopy has the potential for the reagentless differentiation of GSR based upon forensically relevant parameters, such as caliber size. When fully developed, this method should have a significant impact on the efficiency of crime scene investigations.

show abstract

Section: Resultsmentioning

confidence: 99%

Raman Spectroscopic Analysis of Gunshot Residue Offering Great Potential for Caliber Differentiation

2012

View full text Add to dashboard Cite

show abstract

“…Moreover, 1C classification is an alternative approach for conventional multiclass algorithms that classify the examples based on several pre-defined categories, which referred to as open set conditions. In particular, in several forensic tasks and security-oriented tools, the open set problem has been investigated so far in [86]. Wang et al, [86] considered SVMs multi-class and 1C to recognize different camera models.…”

Section: • Defense Layermentioning

confidence: 99%

“…In particular, in several forensic tasks and security-oriented tools, the open set problem has been investigated so far in [86]. Wang et al, [86] considered SVMs multi-class and 1C to recognize different camera models. This technique is more robust against DJPEG images.…”

Section: • Defense Layermentioning

confidence: 99%

A Survey of Machine Learning Techniques in Adversarial Image Forensics

Nowroozi¹,

Dehghantanha²,

Parizi³

et al. 2020

Preprint

View full text Add to dashboard Cite

Image forensic plays a crucial role in both criminal investigations (e.g., dissemination of fake images to spread racial hate or false narratives about specific ethnicity groups) and civil litigation (e.g., defamation). Increasingly, machine learning approaches are also utilized in image forensics. However, there are also a number of limitations and vulnerabilities associated with machine learning-based approaches, for example how to detect adversarial (image) examples, with real-world consequences (e.g., inadmissible evidence, or wrongful conviction). Therefore, with a focus on image forensics, this paper surveys techniques that can be used to enhance the robustness of machine learning-based binary manipulation detectors in various adversarial scenarios.

show abstract

“…This problem is often referred to as classification in open set conditions. Open set problems have been studied in several image forensic and securityoriented applications, such as fingerprint spoof detection [28] source device attribution [12], and camera model identification [35,5]. In [35], in particular, a combination of one-class and multi-class SVMs is used to simultaneously recognize the camera model among the models in a known set and, at the same time, identify outliers, acquired by unknown camera models.…”

Section: Prior Artmentioning

confidence: 99%

Improving the security of image manipulation detection through one-and-a-half-class multiple classification

Barni

Nowroozi

Tondi

2019

Multimed Tools Appl

View full text Add to dashboard Cite

Protecting image manipulation detectors against perfect knowledge attacks requires the adoption of detector architectures which are intrinsically difficult to attack. In this paper, we do so, by exploiting a recently proposed multiple-classifier architecture combining the improved security of 1-Class (1C) classification and the good performance ensured by conventional 2-Class (2C) classification in the absence of attacks. The architecture, also known as 1.5-Class (1.5C) classifier, consists of one 2C classifier and two 1C classifiers run in parallel followed by a final 1C classifier. In our system, the first three classifiers are implemented by means of Support Vector Machines (SVM) fed with SPAM features. The outputs of such classifiers are then processed by a final 1C SVM in charge of making the final decision. Particular care is taken to design a proper strategy to train the SVMs the 1.5C classifier relies on. This is a crucial task, due to the difficulty of training the two 1C classifiers at the front end of the system. We assessed the performance of the proposed solution with regard to three manipulation detection tasks, namely image resizing, contrast enhancement and median filtering. As a result the security improvement allowed by the 1.5C architecture with respect to a conventional 2C solution is confirmed, with a performance loss in the absence of attacks that remains at a negligible level. * The 1.5C code is avaliable on: http://github.com/EhsanNowroozi/ Accessed: 2019-09-05.

show abstract

Source Camera Identification Using Support Vector Machines

Cited by 15 publications

References 14 publications

Raman Spectroscopic Analysis of Gunshot Residue Offering Great Potential for Caliber Differentiation

Raman Spectroscopic Analysis of Gunshot Residue Offering Great Potential for Caliber Differentiation

A Survey of Machine Learning Techniques in Adversarial Image Forensics

Improving the security of image manipulation detection through one-and-a-half-class multiple classification

Contact Info

Product

Resources

About