16-04-2012, 02:44 PM
EXPOSING DIGITAL IMAGE FORGERIES BY DETECTING DISCREPANCIES IN MOTION BLUR
EXPOSING DIGITAL IMAGE FORGERIES BY DETECTING DISCREPANCIES 1.ppt (Size: 2.04 MB / Downloads: 110)
INTRODUCTION
Image forgery
- process of creating fake images.
It deals with digital images.
Availability of photo manipulation software made it easy to manipulate images for malicious purposes.
IMAGE SPLICING DETECTION TECHNIQUES
Detection using
Traces of resampling
Biseptual analysis
Inconsistencies in lighting
Chromatic aberration
Image interpolation from digital image
Consistencies of defocus blur
DISCREPANCIES IN MOTION BLUR
an original forgery detection approach employing motion blur estimation via spectral characteristics of image gradients,which can detect small inconsistencies in motion blur;
a novel blur estimate measure designed especially to deal with very little motion blur;
a no-reference perceptual blur metric extended to directional motion blur;
BLOCK LEVEL ANALYSIS
The image is divided Mb X Nb into overlapping blocks bm,n
m=1 to Mb ,n=1 to Nb and the motion blur estimate m,n for each block is calculated.
m,n is a two-dimensional vector consisting of the motion blur estimate magnitudes and directions
The image subdivision has two major benefits:
Motion blur can be estimated at a number of points, as opposed to just a single estimate for the entire image, giving improved resolution,
space-invariance of motion blur can be assumed over each block, allowing for simpler calculations.
CONCLUSION
The suspected image is divided into overlapping blocks and the motion blur for each block is estimated.
Post processing is done by smoothing the blur estimates and unsampling.
Regions of the image which show inconsistent blur are then segmented from the image and displayed to the user.
BEM is calculated to provide robust segmentation in the case of little perceptible blur