Image forgery detection through motion blur estimates

Images can be easily manipulated for malicious purposes due to broad availability of photo exploitation software. One such form of tampering is image splicing. To detect splicing in images by searching discrepancies in motion blur is one type of method for forgery detection. The motion blur estimation is using image gradients to detect inconsistencies between the spliced region and the rest of the image. The gradient based PBM method gives better results for a wide range of magnitude values as compared to Cepstral method. A Blur Estimate Measure (BEM) is used to aid in inconsistent region segmentation in images that contain small amounts of motion blur and a no-reference Perceptual Blur Metric (PBM) has been used to detect directional motion blurs in images. Based on these measures the regions of the images are identified with consistent and inconsistent blurs. The effect of motion blur inconsistencies and region separation is achieved using K-means clustering algorithm.