为了改善信息隐藏后的图像质量,减少失真,并提高嵌入信息的容量,本文提出了一种新的改进的基于快速相关矢量量化(MFCVQ,modified fast correlation VQ)的信息隐藏方法.由于图像本身的相关性,快速相关矢量量化利用了当前索引的相邻矢量进行替代编码,计算编码时使用的相邻矢量所产生的失真与事先设定的门限做比较,当失真小于门限时可以进行替代编码,一步完成了编码和嵌入信息；当失真大于门限时不能进行替代编码,从而控制了当前索引的嵌入和最后生成图像的视觉质量.同时,增加了相邻矢量的数目,提高了嵌入信息的容量.实验结果表明改进后的算法能够显著改善图像失真,在门限为18时对不同复杂程度的图像其PSNR分别提高了0.018dB-2.125dB,并且有效地提高了嵌入容量,进而大幅度的提高了嵌入效率.改进算法的嵌入效率达到了Yang算法的1.967-4.683倍.%An improved information hiding scheme based-on modified fast correlation vector quantization is proposed to improve visual image quality, decrease the encoding distortion and increase the embedding capacity of the secret data in this paper. Modified fast correlation vector quantization uses the neighbor vectors to encode the current index because of the correlation of the image itself. The proposed method compares encoding distortion between the current index and its neighbor vector selected by secret information with the given threshold to determine the embedding status of the current index. If the encoding distortion is less than the threshold, the current index can be encoded by its neighbor vector and secret information can be embedded at the same time. If the encoding distortion is larger than the threshold, the current index cannot embed secret information. In addition, the method increases the number of neighboring blocks with high payload. The experimental results show that the proposed method decreases image distortion significantly, which means PSNR increasement of 0.018 dB-2. 125dB for thresholding of 18. It also increases embedding capacity effectively and then improves embedding-efficien-cy correspondingly. As a result, the proposed method obtains 1.967-4.683 times higher embedding-efficiency compared to Yang' s method.
展开▼
