基于熵编码的JPEG压缩域脆弱图像水印算法

为了解决JPEG图像水印算法在嵌入容量、不可感知性、实时性和窜改定位的问题，提出了一种基于熵编码的JPEG压缩域脆弱图像水印算法。首先选择JPEG熵编码阶段进行水印嵌入，这样就可避免正向和逆向的DCT和量化运算，保证算法在嵌入阶段对图像的改变量较小且具有较好的实时性；接着根据熵编码过程中某些比特不会参与哈夫曼编码这一原则，在其中选择符合条件的系数嵌入水印，将水印信息嵌入到这些系数的最低有效位中，以规避修改哈夫曼编码系数后带来的编码误差，进一步加强了算法的透明性。实验表明，该算法不仅具有较大的嵌入容量和很好的不可感知性，而且脆弱性较高且能够对恶意窜改进行准确定位。

JPEG compressed domain fragile image watermarking algorithm based on entropy coding

In order to solve the problem of embedding capacity, imperceptibility, real-time and tampering localization in JPEG image watermarking algorithm, this paper proposes a JPEG watermarking algorithm based on entropy coding. Firstly, the algorithm locates the process of JPEG entropy coding to embed the watermarking, which can avoid the forward and reverse quantization operations and ensure that during embedding image changes is small and has good real-time performance. Secondly, according to the principle that some bits in the entropy coding process do not participate in the Huffman coding, the algorithm chooses appropriate coefficients to embed watermark. Our algorithm embeds watermarking information into the least significant bits of these coefficients to avoid the coding error caused by the modification of the Huffman coding coefficient, which should further enhance the imperceptibility of the algorithm. Experiments results show that the algorithm not only has a large embedded capacity and good imperceptibility, but also has high fragility and can accurately locate malicious tampering.