Comparative Analysis of Various Image Splicing Algorithms

Daily millions of images are uploaded and download to the web, as a result the data is available in the paperless form in the computer system for organization. Nowadays, with the help of powerful computer software such as Photoshop and Corel Draw, it is very easy to alter the contents of the authenticated image without leaving any clues. This led to a big problem due to the negative impact of image splicing. It is highly recommended to develop image tampering detection technique to recognize the authentic and temper images. In this paper, we propose an enhanced technique for blind images splicing by combing Discrete Cosine Transform Domain (DTC) and Markov feature in the spatial domain. Moreover, Principal Component Analysis (PCA) is used to select the most significant features. Finally, Support Vector Machine (SVM) is applied to classify the image as being tempered or genuine on the publicly available dataset using ten-fold cross-validation. By applying different statistical techniques, the results showed that the proposed technique performs better than other available detection techniques in the literature.