Visual artifacts based image splicing detection in uncompressed images

Forensic analysis of digital images for tampering detection is a challenging task, if the information regarding them are not available. This scenario in which the prior original image information are not known is called blind/passive/non-reference. So, the forensic examiner have to look for trails of tampering by studying the artifacts that might have crept into the image editing processes or the image acquisition pipeline. We proposed a new method of image splicing detection based on visual artifacts in (uncompressed) images. It was tested on the Columbia Uncompressed Image Splicing Detection Evaluation Dataset having 180 (Authentic) and 183 (Spliced) images and the results are encouraging.