基于重要块二值特征的电子凭证自恢复水印算法

发布时间：2018-05-29 11:58

本文选题：电子凭证 + 数字图像　；参考：《西南交通大学学报》2017年01期

【摘要】：现有高性能自恢复水印算法多是针对自然图像设计,水印容量较大且不利于文字等信息的恢复.为解决这一问题,在兼顾水印嵌入容量和篡改凭证图像恢复质量的基础上,提出了一种适用于电子凭证图像真实性认证的自恢复水印算法.该算法首先根据电子凭证图像特性,将图像块划分为重要块、非重要块和空白块3类,并对其进行分类编码;其次,为保证类型码的正确性,将所有图像块类型码置乱,采用RS编码后生成部分恢复水印信息,将其平均嵌入在所有图像块中;再次,对重要块采用二值化方法生成恢复水印信息,并依次嵌入在非重要块或空白块的最低有效位;最后,通过7张具体电子凭证图像,对比给出了本文算法与现有同类变容量文献算法的性能.研究结果表明,本文算法采用分类编码并仅对重要块生成恢复水印信息,在不降低算法篡改检测和篡改恢复性能的条件下,使水印嵌入容量从对比文献的1.73和2.99降低至0.64,含水印凭证图像的峰值信噪比分别高出对比文献8 d B和23 d B.
[Abstract]:Most of the existing high performance self-restoration watermarking algorithms are designed for natural images, which have large watermark capacity and are not conducive to the restoration of information such as text. In order to solve this problem, a self-restoration watermarking algorithm suitable for authenticity authentication of electronic voucher image is proposed on the basis of both watermark embedding capacity and tampering certificate image restoration quality. According to the characteristic of electronic certificate image, the algorithm divides the image block into three classes: important block, non-important block and blank block, and classifies and encodes them. Secondly, in order to ensure the correctness of the type code, all image block type codes are scrambled. After RS coding, the watermark information is partially restored and embedded in all image blocks on average. Thirdly, the restoration watermark information is generated by binary method for important blocks, and then embedded in the lowest significant bits of non-important blocks or blank blocks in turn. Finally, the performance of the proposed algorithm is compared with that of the existing variable capacity literature algorithms through 7 specific electronic credentials images. The results show that the proposed algorithm uses classification coding and only generates restoration watermark information for important blocks, without reducing the tampering detection and tampering recovery performance of the algorithm. The watermark embedding capacity is reduced from 1.73 and 2.99 to 0.64, and the peak signal-to-noise ratio of the image with watermark certificate is higher than that of contrast document 8 dB and 23 dB, respectively.
【作者单位】：西南交通大学信息科学与技术学院;
【基金】：国家自然科学基金资助项目(61373180)
【分类号】：TP309.7

【相似文献】