Forensic Camera Identification: Effects of Off-Nominal Exposures
Photo response non-uniformity (PRNU) is a technology that can match a digital photograph to the camera that took it. Due to its use in forensic investigations and use by forensic experts in court, it is important that error rates for this technology are reliable for a wide range of evidence image types. In particular, images with off-nominal exposures are not uncommon. This paper presents a preliminary investigation of the impact that images with different exposure types - too dark or too light - have on error rates for PRNU source camera identification. We construct a new dataset comprised of 8400 carefully collected images ranging from under-exposed (too dark) to nominally exposed to over-exposed (too bright). We first establish baseline error rates using only nominally exposed images, resulting in a true-positive rate of 100% and a true-negative rate of 99.92%. When off-nominal images are tested, we find striking results: the true-negative rate for under-exposed images is 99.46% (a false-positive rate of roughly one in two hundred, typically unacceptable in a forensic context), and for over-exposed images the true-positive rate falls to 82.90%. Our results highlight the importance of continued study of error rates for the PRNU source camera identification to assure adherence to the high standards set for admissibility of forensic evidence in court.
Code (0)
등록된 구현이 없습니다.
Methods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
Training CNNs in Presence of JPEG Compression: Multimedia Forensics vs Computer Vision
Convolutional Neural Networks (CNNs) have proved very accurate in multiple computer vision image classification tasks that required visual inspection in the past (e.g., object recognition, face detection, etc.). Motivate…
Face DetectionGeneral Classificationimage-classificationImage Classification+1First Steps Toward Camera Model Identification with Convolutional Neural Networks
Detecting the camera model used to shoot a picture enables to solve a wide series of forensic problems, from copyright infringement to ownership attribution. For this reason, the forensic community has developed a set of…
General ClassificationImage ForensicsApplication of DenseNet in Camera Model Identification and Post-processing Detection
Camera model identification has earned paramount importance in the field of image forensics with an upsurge of digitally altered images which are constantly being shared through websites, media, and social applications. …
Image ForensicsOverall - TestTransfer LearningVideo Camera Identification from Sensor Pattern Noise with a Constrained ConvNet
The identification of source cameras from videos, though it is a highly relevant forensic analysis topic, has been studied much less than its counterpart that uses images. In this work we propose a method to identify the…
Location Forensics of Media Recordings Utilizing Cascaded SVM and Pole-matching Classifiers
Information regarding the location of power distribution grid can be extracted from the power signature embedded in the multimedia signals (e.g., audio, video data) recorded near electrical activities. This implicit mech…
General Classification