Forgery Detection in Digital Images Using Computational Intelligence

Abstract

Due to availability of powerful image editing tools images are open to several manipulations. Therefore, their authenticity is becoming questionable especially when images have influential power, for example in a court of law, news reports, insurance claims, criminal investigation, medical imaging etc. The most common image tampering often for malicious purposes is to copy a region and paste to hide some other region of the same image is known as copy-move forgery (CMF). As both regions usually have same texture properties, therefore, this artifact is invisible for the viewers and credibility of the image becomes questionable in proof centered applications. Hence, means are required to validate the integrity of the image and identify the tampered regions. Image forensic techniques determine the integrity of the images by applying various high-tech mechanisms developed in the literature. In this dissertation, three techniques for copy-move forgery detection (CMFD) are presented to verify the truthfulness of image contents. Thus, for efficient detection of CMF, the first approach that we have presented exploits local binary pattern variance (LBPV) over the low approximation components of the stationary wavelets. The proposed CMFD method is applied over the circular regions to address the possible post-processing operations in a better way. The proposed method is evaluated on CoMoFoD (Copy Move Forgery Detection) and KLTCI (kodak lossless true color image) datasets in the presence of translation, flipping, blurring, rotation, scaling, color reduction, brightness change and multiple forged regions in an image. Our second method, presents an algorithm that utilizes stationary wavelet transform (SWT). The method exploits low approximation sub-band for forgery detection. The algorithm divides the low approximation sub-band into the small overlapping square blocks. A reduced feature vector representation is achieved by dividing each block into four triangles. The experimental results demonstrate that the algorithm is capable of detecting duplicated blocks precisely and identify multiple CMF effectively, even when the images are contaminated by blurring and noise. In our third method of CMFD, images are first divided into overlapping square blocks and DCT components are adopted as the block representations. Due to the high dimensional nature of the feature space, Gaussian radial basis function (RBF) kernel principal component analysis (PCA) is applied to achieve the reduced dimensional feature vector representation that also improved the efficiency during the feature matching step. Extensive experiments are performed on DVMM image forensic dataset and google images to evaluate the proposed method in comparison against state-of-the-arts. The experimental results reveal that the proposed technique precisely determines the CMF even when the images are contaminated with blurring, noise, compression and can effectively detect multiple CMF. All the three techniques presented in this dissertation are compared against renounced methods of CMFD over multiple image forensic datasets. The evaluation reveals the prominence of the presented methods as compared to state-of-the-arts. Consequently, the proposed techniques can reliably be applied to detect the forged regions and the benefits can be obtained in journalism, law enforcement, judiciary, and other proof critical domains.