Video Camera Identification from Sensor Pattern Noise with a Constrained ConvNet

Abstract

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 source camera of a video based on camera specific noise patterns that we extract from video frames. For the extraction of noise pattern features, we propose an extended version of a constrained convolutional layer capable of processing color inputs. Our system is designed to classify individual video frames which are in turn combined by a majority vote to identify the source camera. We evaluated this approach on the benchmark VISION data set consisting of 1539 videos from 28 different cameras. To the best of our knowledge, this is the first work that addresses the challenge of video camera identification on a device level. The experiments show that our approach is very promising, achieving up to 93.1% accuracy while being robust to the WhatsApp and YouTube compression techniques. This work is part of the EU-funded project 4NSEEK focused on forensics against child sexual abuse. c https://orcid.org/0000-0003-2081-774X d https://orcid.org/0000-0001-6552-2596 * Derrick Timmerman and Guru Swaroop Bennabhaktula are both first authors.

PRNU-Net: a Deep Learning Approach for Source Camera Model Identification based on Videos Taken with Smartphone

“…References [20] and [21] proposed a deep learning method (MISLnet architecture) for source camera identification using video frames to train the network. They extended a version of a constrained convolutional layer introduced in [19] as mentioned in Section I.…”

“…The key difference between the two methods relates to the size of images and type of color modes. [20] and [21] used RGB and gray scale modes, respectively. Patches used in former is 480 while latter fed patched with 256.…”

“…We divide the experiments into different scenarios showing the influence of some conditions on the results, which related to the position and repetition of the layer during training. Our proposed network is compared with MISLnet architecture [19] that has been used in several recent studies [20], [26], [21]. To show the impact of separating content from noise, MISLnet is considered with and without the constrained layer.…”

“…To show the impact of separating content from noise, MISLnet is considered with and without the constrained layer. Our implementation of the architecture is based more on [20], which was considered for identifying the source camera. The Stochastic gradient descent (SGD) is considered to train the model.…”

“…(1) kj denotes the jth kernel of the kth filter in the first layer of the network. Despite promising results of the method [20], [21], because the degree of sensitivity in the field, an improvement in the field is always essential.…”