基于新闻文本图像的鲁棒水印算法OACSTPCD

In news text images,the existing watermarking algorithms fail to distinguish the regions of expressing text from the other background regions.In addition,they only embed binary watermark images in a single channel,which results in low robustness.In view of this,a robust watermarking algorithm based on news text image is proposed.The news text image is segmented by Otsu threshold segmentation(OSTU)to distinguish the text from the background regions,and the text region is selected for embedding to deepen the copyright protection of important information.The text regions of the Cr and Cb channels of the text image are subjected to discrete wavelet transform(DWT),and principal component analysis(PCA)is used to concentrate energy and calculate the contribution rates of each principal component.The embedded principal components are selected by comparing the contribution rates.The singular value decomposition(SVD)is performed on the principal component and watermark image to complete watermark embedding.After experiments,the NC value of the embedded watermark image in the face of most conventional attacks such as filtering is above 0.99,which indicates that the algorithm has strong robustness.At the same time,the average PSNR of the image after embedding the watermark is 45.66 dB,which ensures imperceptibility.