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

Abstract

Recent advances in digital imaging have meant that every smartphone has a video camera that can record highquality video for free and without restrictions. In addition, rapidly developing Internet technology has contributed significantly to the widespread distribution of digital video via web-based multimedia systems and mobile applications such as YouTube, Facebook, Twitter, WhatsApp, etc. However, as the recording and distribution of digital video has become affordable nowadays, security issues have become threatening and have spread worldwide. One of the security issues is the identification of source cameras on videos. Generally, two common categories of methods are used in this area, namely Photo Response Non-Uniformity (PRNU) and Machine Learning approaches. To exploit the power of both approaches, this work adds a new PRNU-based layer to a convolutional neural network (CNN) called PRNU-Net. To explore the new layer, the main structure of the CNN is based on the MISLnet, which has been used in several studies to identify the source camera. The experimental results show that the PRNU-Net is more successful than the MISLnet and that the PRNU extracted by the layer from low features, namely edges or textures, is more useful than high and mid-level features, namely parts and objects, in classifying source camera models. On average, the network improves the results in a new database by about 4%.

Source Camera Identification with a Robust Device Fingerprint: Evolution from Image-Based to Video-Based Approaches

“…However, our proposed system achieved higher identification accuracy for the Huawei Y7 and Huawei Y9 models, despite their lower resolution of 720 × 1280 compared to other cameras with 1080 × 1920 resolution. This contrasts with the findings in [43,51] where they reported lower performance for the Huawei Y7 model. In particular, our proposed system, leveraging the new device fingerprint, demonstrated remarkable performance for the Huawei Y7 model, achieving 98.40% accuracy at the frame level and 99.20% accuracy at the video level.…”

“…This result indicates that distinguishing between individual devices within a camera model based solely on video content proved to be a more challenging task. Akbari et al [51] improved the source identification accuracy by integrating a PRNUbased layer into their CNN, thereby harnessing the strengths of both PRNU and machine learning methods. By combining the capabilities of both PRNU and machine learning techniques, they developed a novel CNN architecture called PRNU-net.…”

“…In this sub-section, we aim to assess the strength and significance of our new fingerprint in the field of video-based source identification. To achieve this, we compare the performance of our proposed system with other existing methods [43,51,52] that have demonstrated benchmark results on the QUFVD [43] through two experiments: source camera model identification (SCMI) and individual source camera identification (ISCI). Firstly, in Table 5, we compare the results of the experiment involving 10 different camera models present in the QUFVD dataset to evaluate the performance of our system in SCMI.…”

“…Firstly, in Table 5, we compare the results of the experiment involving 10 different camera models present in the QUFVD dataset to evaluate the performance of our system in SCMI. At the frame-level, it was hard for PRNU-net [51], which combines the benefits of PRNU fingerprint and deep learning, to identify the Huawei Y7 model due to the lower resolution of videos, resulting in a lower performance of 71.5% compared to other models. However, our proposed system achieved higher identification accuracy for the Huawei Y7 and Huawei Y9 models, despite their lower resolution of 720 × 1280 compared to other cameras with 1080 × 1920 resolution.…”

“…Overall, our proposed system outperformed the other existing methods in terms of overall identification accuracy at the frame-level and video level. In [43,51], the authors also faced challenges in identifying cameras such as iPhone 8 Plus, and Redmi Note 9 Pro, resulting in lower performance. However, our proposed system demonstrated robust performance, achieving more than 90% identification accuracy for these challenging camera models.…”