基于BCH编码的局部化标注水印算法

LOCALIZED CAPTION WATERMARKING SCHEME BASED ON BCH CODING

本文提出一种基于BCH编码和小波变换的局部化标注数字水印算法。本算法为公开数字水印算法,提取水印时不需要原图的任何信息。算法利用图像中稳定的特征点标示水印嵌入的位置,在每个特征点对应的局部区域中,独立嵌入预先用BCH容错编码的标注信息(字符串)。当我们只得到部分图像时仍能通过这些特征点来定位并提取水印,读出嵌入的字符串。实验表明,算法对裁剪有很强的抵抗能力,同时对压缩、噪声等也有必要的鲁棒性。此水印算法在不要求准确提取所嵌入的字符信息时,还可以作为鲁棒水印使用;在能保证数据无失真时,则可以嵌入大量标注信息。

In this paper, we propose a localized caption watermarking scheme based on BCH coding and wavelet transform, which embeds watermarks in the local regions of the image. This scheme is a public watermarking algorithm that allows us to extract the embedded watermarks without knowing the original image. We find some stable feature points in the image, and then use these points to identify the locations where we embed the BCH-coded captions. Caption watermarks are then inserted into each region corresponding to the feature point independently. When the image is cropped, the feature points in the remaining piece of the image are not change, so we can extract watermarks through the remaining feature points. Experiment results show that the scheme is robust to cropping, compressing, and noise adding. If it is not required to extract the exact caption string, this scheme can be used as robust watermarking. Furthermore, if we can guarantee there is no other distortions except for cropping before we get the entire watermarked image or only part of it, we have the ability to hide a lot amount of information in the local regions of the image.