Harnessing motion blur to uncover splicing

Abstract

Image tampering has become rampant in today's world due to availability of sophisticated image editing tools. In this paper, we deal with the problem of image splicing which is one form of tampering. We propose a passive method to detect the presence of splicing in a given image based on inconsistencies derived from motion blur. Both planar and 3D scenes are considered. The cause of blurring in the image is restricted to translation camera motion while the scene is assumed to be static. We validate our approach on synthetic as well as real examples.

Image Forgery Detection Based on Motion Blur Estimated Using Convolutional Neural Network

“…However, as directly depending on the discrepancy of MBKs of adjacent image patches, it will fail when rotation involves in, as discussed in [7] and [8]. To overcome this issue, [9], [10] estimate nonparameterized MBKs of image patches using [18], model the camera motion using transformation spread function (TSF) based on MBKs, and then identify tamper by comparing the MBKs estimated from the image patches and the deduced MBKs from TSF. The main issues of this method are that, first, the estimation of non-parameterized MBKs highly depends on the content of the image, and is inaccurate on small image patches, as discussed in [18]- [21].…”

“…However, for image tamper detection, such postprocessing procedure can not be used. Therefore, in this paper, the size of the image patch is set to 120 × 120 × 3, which is similar to [9]- [11]. As a result, two convolution/ReLU layers and two max-pooling layers are added in front of the CNN used in [23], resulting a 10-layer CNN CNN 1 , the structure is C1-M2-C3-M4-C5-M6-C7-M8-F9-S10.…”