Image classifier based digital image forensic detection-a review and simulations

From the early days, images are being generally accepted as a proof of occurrence of the past events. The availability of low cost hardware and software tools, makes easy to create and manipulate digital images with no obvious traces. This has led to the situation where one can no longer take the integrity and authenticity of the digital images for granted. In many cases, the images are copied from one image to another image to form the composite image called splicing. This is done by various image editing softwares like photoshop, GIMP etc. So, the spliced images are under scan and need to be tested for its authenticity. In this regard, image feature extraction based classifiers like K Nearest Neigbor(KNN), Fuzzy Logic and Support Vector Machine(SVM) are used for identification of spliced images. In the first step, features are extracted from the Gray Level Co-Occurrence Matrix (GLCM). The extracted features are fed to the KNN, Fuzzy and SVM classifiers. Based on the Accuracy parameter obtained from each of the classifier, the comparative analysis of the classifiers is performed.