Dissertation
Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

The Image Copy-Move Forgery Detection Based on DWT and Phase Correlation

Author HouDongMei
Tutor BaiZhengYao
School Yunnan University
Course Signal and Information Processing
Keywords Copy and paste forgery Wavelet Transform Cross sub- block method Phase correlation Pixel matching
CLC TP391.41
Type Master's thesis
Year 2011
Downloads 70
Quotes 0
Download Dissertation

The rapid development of powerful image processing software , ordinary users without leaving obvious traces easier for digital image processing . If these false picture abuse will lead to the emergence of some relevant issues related to the stability of our society , the law of forensics and personal privacy . For digital image tampering with digital image forensics has become a very important research topic . Existing digital image forensics technology , in addition to the active forensic methods based on digital signatures and digital watermarking , passive blind forensics methods do not require any additional information that is added to the image . These methods have to copy and paste detect JPEG double compression detection , resampling detection and blur estimation detection . This paper studies a specific digital image forgery - Regional blind copy-paste forgery detection algorithm . The area to copy paste forgery is a common image forgery techniques , it is a part of the region in the image copy , paste it into disjoint another area of the same image , in order to remove the image of an important purpose . In this paper, the tampering technology , to improve the existing algorithms based on , and thus presents a new effective copy and paste tamper detection algorithm . Firstly discrete wavelet transform scale reduction in the measured image , after reduction and scale images with Cross Shadow chunking method is divided into nine image sub-block . Then calculated for each of the two image sub-block phase correlation and do pulse determination , based on the correlation coefficient with the pulse obtained a copy - paste regional displacement . Then according to the demand displacement pixel matching to locate a copy - paste area . Finally, the use of the morphological filtering technology to further improve positioning effect . The algorithm is simple to understand and easy to implement. Experimental results show that the algorithm can not only small areas of location detection , and multiple tampered block detection , and in ensuring the accuracy and expand the scope of its application .

Related Dissertations
More Dissertations