Exposing image forgery using inconsistent reflection vanishing point

Digital image is an effective communication media due to its immediacy and the easy-to-understand its content. The rapid development of image processing techniques has made it extremely easy to alter the content of images or to create new images. So photographs, which appear in magazines, social media, political attacks, can no longer be trusted. Over the past ten years, the field of digital image forensics has emerged to help restore some trust to digital images. This paper proposes a novel and effective technique to expose image forgery using inconsistent reflection vanishing point. More specifically, we give the definition of error distance, propose sin²()-based function to normalize error distance, and employ maximum normalized error distance to detect image forgery for the first time. Effectiveness and accuracy of the proposed technique is evaluated using a data set consisting of 200 authentic and forged images. Experimental results indicate that the proposed method can detect image forgery with very high success rate and it is extremely robust to compression, filtering, and other image processing operations.