块级篡改定位的JPEG图像脆弱水印

 JPEG是一种常见的图像格式，在JPEG图像中进行准确的篡改定位具有重要意义. 本文提出一种新的JPEG图像脆弱水印方案，将每个小块主要内容的Hash比特重新分组，并将每组的模2和作为水印信息. 也就是说每一个小块都对应多个水印比特，每个水印比特也对应多个小块. 载体图像的每个小块中仅嵌入于1比特水印，保证了良好的隐蔽性. 认证时依据图像内容与水印比特的整体匹配情况估计篡改率，再根据每个小块对应的水印信息的被破坏程度判别该小块是否曾被篡改. 理论分析和实验结果标明该方法可以在篡改区域小于1%的情况下准确地找到所有篡改小块.

Fragile Watermarking Capable of Locating Tampered Blocks in JPEG Images

It is significantly meaningful to exactly locating the tampered areas in digital images. This paper proposes a novel fragile watermarking scheme for JPEG covers, in which the hash bits of all blocks with a size of 8×8 are pseudo-randomly reorganized as a series of subsets, and their sums with modulus 2 are regarded as watermark bits. This way, each block maps a number of watermark bits, and each watermark bit corresponds to a number of blocks. Then, a selected non-zero DCT coefficient in each block is used to accommodate one watermark bit. At authentication side, after estimating the tampering rate, the inconsistency between the calculated and extracted watermark bits of each block is used to judge it as “tampered” or “reserved”. Theoretical analysis and experimental result show that the proposed method is capable of locating all tampered blocks when the tampering rate is less than 1%.