원문정보
초록
영어
In order to solve such problems as poor robustness and low detection accuracy in the present copying & pasting forgery detection algorithm, an image copying & moving forgery detection algorithm based on non-sampling wavelet transform (undecimated wavelet transform, UWT) coupling with Zernike moments is proposed in this article. Firstly, the non-sampling wavelet transform is adopted to decompose the input image into approximation coefficient LL and detail coefficient HH to obtain the similarity and the diversity of the image blocks; secondly, the overlapped blocks are divided to calculate the sub-block interval of the image; thirdly, the sub-blocks are ordered and classified according to LL similarity and HH diversity; fourthly, Zernike moments are introduced to construct the distance matrix for image feature matching and image forgery detection. The experiment shows: the algorithm proposed thereby has good translation invariance and amplitude rotation invariance, good post-processing resistance performance, high detection accuracy, low false negative rate and low false positive rate.
목차
1. Introduction
2. Mathematical Model for Copying & Moving Forgery
3. Forgery Detection Algorithm
3.1. Non Sampling Wavelet Transform
3.2. Zernike Moments
3.3. Specific Implementation
4. Experimental Result
4.1. Copying & Pasting Forgery Detection of the Tampered Zone without Rotation
4.2. Test under Noise andJPEGCompression Factor
5. Conclusion
Reference