Source Camera Identification Issues

Statistical image features play an important role in forensic identification. Current source camera identification schemes select image features mainly based on classification accuracy and computational efficiency. For forensic investigation purposes; however, these selection criteria are not enough. Consider most real-world photos may have undergone common image processing due to various reasons, source camera classifiers must have the capability to deal with those processed photos. In this work, the authors first build a sample camera classifier using a combination of popular image features, and then reveal its deficiency. Based on the experiments, suggestions for the design of robust camera classifiers are given. Copyright © 2011, IGI Global.