Identifying Image Manipulation Software from Image Features.

摘要

As technology steadily increases in the field of image manipulation, determining which software was used to manipulate an image becomes increasingly complex for law enforcement and intelligence agencies. To combat this difficult problem, new techniques that examine the artifacts left behind by a specific manipulation are converted to features for classification. This research implemented four preexisting image manipulation detection techniques into a framework of modules: Two-Dimensional Second Derivative, One-Dimensional Zero Crossings, Quantization Matrices Identification, and File Metadata analysis. The intent is the creation of a framework to develop a capability to determine which specific image manipulation software program manipulated an image. The determination is based on each image manipulation software program having implemented the manipulation algorithms differently. These differences in the implementation will leave behind different artifacts in the resultant image. Experimental results demonstrate the framework's ability to identify from the 48 combinations of image manipulation software programs, scaling, and the algorithm used with a true positive rate of 0.54, false positive rate of 0.01, and a Kappa statistic of 0.53 for Joint Photographic Experts Group (JPEG). The results for Tagged Image File Format (TIFF) images were a true positive rate of 0.53, false positive rate of 0.01, and a Kappa statistic of 0.52.