计算机科学与探索 ›› 2018, Vol. 12 ›› Issue (12): 1961-1973.DOI: 10.3778/j.issn.1673-9418.1802024

• 人工智能与模式识别 • 上一篇    下一篇

图像插值空间自适应大容量无损信息隐藏算法

熊祥光,陈熙,曹永锋,欧卫华,刘彬,夏道勋   

  1. 贵州师范大学 大数据与计算机科学学院,贵阳 550001
  • 出版日期:2018-12-01 发布日期:2018-12-07

Adaptive High Capacity Lossless Data Hiding Algorithm for Image Interpolation Space

XIONG Xiangguang, CHEN Xi, CAO Yongfeng, OU Weihua, LIU Bin, XIA Daoxun   

  1. School of Big Data and Computer Science, Guizhou Normal University, Guiyang 550001, China
  • Online:2018-12-01 Published:2018-12-07

摘要:

针对基于图像插值技术的无损信息隐藏算法虽然具有秘密数据隐藏容量大,但是隐秘图像质量欠佳的问题,提出一种基于图像插值技术的自适应大容量无损信息隐藏算法。首先,提出一种改进的图像插值算法。然后,将该插值算法应用于下采样图像得到待隐藏秘密数据的载体图像和对其进行重叠分块,计算各个分块中所有基准像素的方差,根据方差大小进行升序排序并得到各个分块的索引。其次,利用得到的索引按顺序选择各个分块,分别计算各个非基准像素与分块中基准像素最大值和最小值差值的绝对值。对于各个非基准像素,自适应选择绝对值最大的差值作为它的最终差值,并据此差值计算它能隐藏的数据量大小。最后,采用自适应的数据隐藏算法来隐藏秘密数据。在NCID(never-compressed image database)和USC-SIPI两个图像数据库上的实验结果表明,该算法数据隐藏率的平均值为2.492 5 bit/pixel。与同类算法相比,该算法的平均数据隐藏容量不仅提高了约90%,而且隐秘图像质量也提高了约9%。

关键词: 无损信息隐藏, 图像插值, 自适应隐藏, 大容量

Abstract:

Although the hiding capacity of interpolation-based lossless data hiding algorithm is high, image quality of hided image is poor. For this problem, this paper proposes an adaptive high capacity lossless data hiding algorithm based on interpolation technology. Firstly, an improved image interpolation method is proposed, and then the input image is down-sampled, and a new cover image is generated by using the proposed interpolation method on down-sampled image. The cover image is divided into overlapping blocks and the variance of all reference pixels of each block is calculated. According to the variance of all blocks, the sorting operation is processed with ascending and the index of each block is obtained. Secondly, the block is selected by the obtained index and the maximum and minimum of all reference pixels of each block are obtained, and then two differences of each non-reference pixel between the maximum and minimum are calculated, respectively. The final difference of each non-reference pixel is selected adaptably by the magnitude relationship of the absolute value of two differences and the number of bits which can be hidden is computed according to its absolute value of difference. Finally, the adaptive hiding method is used to hide secret data. The proposed algorithm is tested on never-compressed image database (NCID) and USC-SIPI image database. The results show the average data hiding rate of the proposed algorithm is 2.4925 bit/pixel. Compared with similar lossless data hiding algorithms, the results show that not only the hiding capacity of the proposed algorithm is improved by about 90%, but also the image quality of hidden image is also enhanced by about 9% on average.

Key words: lossless data hiding, image interpolation, adaptive hiding, high capacity