Pixel and Edge Based Illuminant Color Estimation for Image Forgery Detection

AbstractImages are one of the powerful media for communication. Image security is a main issue in the fields that using digital images. By the development of high resolution cameras, personal computers and photo-editing software's, the manipulation of images is becoming common. This paper mainly focuses on common form of image manipulation such as image splicing. The process of analysis is done with the help of inconsistencies in illuminant color of images. Illumination inconsistencies detection is a powerful way for image forgery detection. Inconsistency detection among different images can be identified with the help of pixel and edge based illuminant color estimation on image regions. From these illuminant estimators, extract shape and color features, which is then provided to a classifier for making decision. Classification using SVM and its performance is evaluated using distinct testing process. The main contribution of this method is, how illuminant color estimation on various constraints can be exploited as a forgery detection method and how these are provided for decision-making with minimal user interaction