Adaptive Equalization Watermarking Algorithm Based On An Image-Block

The traditional watermarking algorithm uses the transform domain to embed the watermark,which is not conducive to the synchronization of the watermark embedding and extracting.At the same time,the strength parameters need to be determined artificially,and the workload is large and random.It can not play the advantage of the embedding coefficient to realize the equalization of the invisibility and robustness of the watermark system.In this paper,a kind of problem is proposed.Image block adaptive equalization watermarking algorithm.Based on the scale invariant feature transform(SIFT),the robust feature points in the original image are extracted from the original image,and the watermark is embedded in the same size as the watermark.It is divided into four image blocks with equal size and no overlap,and the singular value decomposition(SVD)is made to the image block,and the singular value matrix will be obtained.With the superposition of the subbands generated by the first level discrete wavelet transform,the watermark encryption subblock is generated,the watermark matrix is obtained and the watermark image is obtained by reducing the pixel points after reducing the dimension.The embedding strength affects the performance of the watermarking algorithm.According to the fitness function of the fruit fly optimization algorithm(FOA),the intensity parameters of the encrypted watermark are iteratively determined,and the adaptive embedding of the watermark image is constructed to balance the invisibility and robustness of the watermark.The watermark detection can directly affect the image after the attack,without the need for correction.Multi group simulation experiments are carried out on the standard gray scale image,and the similarity between the original watermark and the watermark is evaluated by using the peak signal to noise ratio(PSNR)to measure the invisibility and normalized correlation coefficient(NC)of the watermark.In this paper,the PSNR value of the watermark carrier is above 43 dB,and the NC value of the watermark extracted by noise,compression,shear,rotation and other attacks are over 0.9.The SIFT algorithm realizes the local embedding of the image block,and the extracted feature points are stable.In combination with the SVD theory,the watermark embedding performance is good.The embedded parameters are determined by the FOA algorithm.The embedding effect of the watermark image is best,and the invisibility and robustness of the watermark algorithm is finally balanced.