Please use this identifier to cite or link to this item:
Title: Image forensics through detection of imaging regularities
Authors: Sai, Choong Han.
Keywords: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
DRNTU::Engineering::Computer science and engineering::Computing methodologies::Pattern recognition
Issue Date: 2009
Abstract: Every image has its unique pattern characteristic and regularities and one of image forensic methods is through pattern recognition. By detecting image regularities, an image can be identified if it is altered by digital manipulation or otherwise. In this project, the image regularities refer to an image tampering method called artificial blurring. Two methods are proposed for detecting artificial blurring which are bispectrum analysis and noise level detection. For bispectrum analysis, it is an improvement from [1] whereby the detection of a 1-D blur model based on zero crossings is taken a step further by analyzing a 2-D blur function. This is done by transforming a 2-D problem into a 1-D problem through line segmentation with edge detection using the Sobel operator. The second method is by noise level detection where an image is divided into many smaller segments and PSNR values of these segments are calculated. It is believed that noise level pattern in an image is consistent throughout an image and if any region is tampered, it will disturb its localized noise level. These two methods are tested based on synthetic and actual images where obtained results from the two methods shows that the algorithm works fine with certain limitations. Since the development of the bispectrum analysis and noise level detection is still in its early stage, further improvements are still needed.
Rights: Nanyang Technological University
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:EEE Student Reports (FYP/IA/PA/PI)

Files in This Item:
File Description SizeFormat 
  Restricted Access
4.22 MBAdobe PDFView/Open

Page view(s)

checked on Oct 1, 2020


checked on Oct 1, 2020

Google ScholarTM


Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.