The Image Copy-Move Forgery Detection Based On DWT And Phase Correlation

Rapid advancement in imaging technology has made it remarkably easy to process digital image contents without leaving any obvious traces of tampering for almost anyone. If the forgery images are abused, it may lead to potentially serious social, legal and/or private consequences. Therefore, research on digital image forensics has become an important topic.Now, in the digital image authentication, there are some passive-blind authentication algorithms emerging which need not to add additive information besides active authentication which based on digital signature and digital watermarking. The methods are copy-move forgery detection,JPEG re-compressing detection,re-sampling detection,blur estimation detection and so on. In this paper, we concentrate on detection of a specific category of digital image forgery—copy-move forgery. Copy-move forgery is a common image forgery technique, in which a portion of the image itself is copied and pasted into another portion of the same image. This manipulation always in order to make an object "disappear" from the image by covering it with a segment copied from another portion of the image.By improving the existing algorithms, this paper describes a new and efficient algorithm for detecting this type of malicious tampering. First, discrete wavelet transform was applied to the input image to yield a reduced dimension representation. Then the image was divided into nine sub-images by using "crossing shadow" division. Next, phase correlation was calculated and pulse was determined to estimate the spatial offset between the copied part and the pasted part. By virtue of pixel matching, the copy-move regions can be easily located. Finally, the mathematical morphological operations were used to improve the detection. This algorithm is easy understood and carried out. The experimental results show that our method is not only achieved an effective and accurate location for small tampered areas but also apply to multiple image blocks detection. By which, algorithm's accurate was guaranteed and application scope of the algorithm was expanded simultaneously.