Image recognition algorithm undecimated wavelet transform coupling moment

In order to solve the problem that robustness in copying and pasting forgery detection algorithm is poor and detection accuracy is insufficient, this paper puts forward to image forgery detection algorithm of coupling Zernike Moment (Zernike) by Undecimated Wavelet Transform (UWT). Firstly, it uses Undecimated Wavelet Transform to break the input image up into approximation coefficient LL and specific coefficient HH and get similarity and difference among blocks; then it divides overlapping block and calculate space between image subblock, makes order and classification according to similarity LL and difference HH and introduces Zernike Moment, builds distance matrix for image feature matching and image forgery detection. The experiment indicates that the mentioned algorithm has good translation invariance and amplitude rotation invariance and it has good anti-post processing performance, high accurate measurement, low omission factor and false positive rate.