Exposing Digital Image Forgeries from Near Duplicate Images

Abstract

Nowadays image modification is easy due to availability of powerful digital image editing software. Images are powerful tool for communication, after images are posted on internet, other users can copy resize and re-encode them and then repost their version by generating similar but not identical copies. The main task is to find out original image between near duplicate image is difficult. It is difficult to adjust the illumination condition when Image composition or splicing operation performed on image to create composite image. Due to these illumination inconsistencies gives clue to detect forgery. But method that operate on illuminant color prone to estimation error, further improvement can be achieved by advanced illuminant estimator as skin color of the faces automatically detected in the image. After finding out the original image relationship within set of near duplicate images is shown by Image Phylogeny Tree (IPT) to finding the structure of transformation and their parameters among near duplicate image .The Technique is applicable to images containing two or more people.