A Machine Learning Approach to Approximate the Age of a Digital Image

In-field image sensor defects develop almost continually over a camera’s lifetime. Since these defects accumulate over time, a forensic analyst can approximate the age of an image under investigation based on the defects present. In this context, the temporal accuracy of the approximation is bounded by the different defect onset times. Thus, the approximation of the image age based on in-field sensor defects can be regarded as a multi-class classification problem. In this paper, we propose to utilize two well-known machine learning techniques (i.e. a Naive Bayes
Classier and a Support Vector Machine) to solve this problem. The accuracy of each technique is empirically evaluated by conducting several experiments, and the results are compared to the current state-of-the art in this field. In addition, the prediction results are assessed individually for each class.