A Visual Perspective For User Identification Based On Camera Fingerprint

User identification is to identify the online accounts' identity, which is a critical problem in many applications. The key problem in that is to find a typical pattern from the online accounts data. Different from the previous works, we identify users based on the camera fingerprint. The underlying idea is that the accounts belonging to the same individual contain the photos taken by the same cameras. Based on that, we propose a new framework to deal with the user identification problem. Specifically, the camera feature of each image is extracted by the PRNU (Photo Response Non-Uniformity) algorithm. With the proposed hierarchical clustering approach, the camera features of different camera source can be achieved. Extensive experiments on a collected photo dataset show that the proposed framework is effective for identifying users, especially in the scenarios that users have multiple cameras and photos forwarding behavior.