CN1885341A - Reversible watermark method for image certification - Google Patents

Reversible watermark method for image certification Download PDF

Info

Publication number
CN1885341A
CN1885341A CN 200610036388 CN200610036388A CN1885341A CN 1885341 A CN1885341 A CN 1885341A CN 200610036388 CN200610036388 CN 200610036388 CN 200610036388 A CN200610036388 A CN 200610036388A CN 1885341 A CN1885341 A CN 1885341A
Authority
CN
China
Prior art keywords
image
value
band
watermark
bit
Prior art date
Application number
CN 200610036388
Other languages
Chinese (zh)
Other versions
CN100394443C (en
Inventor
梁小萍
黄继武
Original Assignee
中山大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中山大学 filed Critical 中山大学
Priority to CN 200610036388 priority Critical patent/CN100394443C/en
Publication of CN1885341A publication Critical patent/CN1885341A/en
Application granted granted Critical
Publication of CN100394443C publication Critical patent/CN100394443C/en

Links

Abstract

The invention relates to a reversible watermark method used in image identification, belonging to multimedia information safety technique, wherein the invention modifies high-frequency sub-band histogram to embed the SHA-256 Haxi value of original image and the data used to recover image into high-frequency sub-band; via replacing the most low bit plane of low-frequency sub-band to embed the useful watermark into low-frequency sub-band; based on extracted watermark or the differential value picture between extracted watermark and former watermark to process falsification check and positioning; comparing extracted Haxi value and the Haxi value of returned image to process content completeness check. The invention can resist false impact, check and position falsification, to return original image without aberration.

Description

一种用于图像认证的可逆水印方法 Reversible watermarking for image authentication method

技术领域 FIELD

本发明属于多媒体信号处理技术领域,具体涉及一种嵌入容量大、可抵抗伪造攻击、具有篡改定位功能的用于图像认证的可逆水印方法。 The present invention belongs to the technical field of multimedia signal processing, in particular to a embedding capacity, resistant to forgery attack, reversible watermarking for image authentication method tamper positioning function.

背景技术 Background technique

数字图像获得了广泛的使用,但是在传输和存储的过程中很容易被篡改甚至是伪造,对数字图像内容的真实性和完整性进行认证成为了信息社会的需要,特别是军事、医学、新闻出版等敏感、珍贵图像应用场合,图像认证必不可少。 Digital images were obtained widely used, but in the transmission and storage process is vulnerable to tampering or even falsified, the authenticity and integrity of digital image content authentication has become the need of the information society, in particular the military, medicine, news publishing sensitive and valuable image application, image authentication is essential.

数字水印技术可用于图像认证,并获得了研究者们极大的关注。 Digital watermarking can be used for image authentication, and access to the researchers great concern. 通过检索,相关的可对比授权专利有如下2项:[1]JMBarton,“Method and apparatus for embedding authenticationinformation within digital data,”In:USPatent(1997)5,646,997;[2]CWHonsinger,P.Jones,M.Rabbani,and JCStoffel,“Lossless recovery of an original image containing embedded data,”In:US Patent(2001)6,278,791。 By retrieving relevant patents can be compared with the following two: [1] JMBarton, "Method and apparatus for embedding authenticationinformation within digital data," In: USPatent (1997) 5,646,997; [2] CWHonsinger, P.Jones, M. Rabbani, and JCStoffel, "Lossless recovery of an original image containing embedded data," In: US Patent (2001) 6,278,791.

相关的技术文献有5篇,如下:[3]J.Fridrich,M.Goljan,and R.Du,“Invertible Authentication,”In:Proc.of SPIE,vol.3971(Jan.2001)197-208;[4]J.Fridrich,M.Goljan,and R.Du,“Invertible AuthenticationWatermark for JPEG Images,”In:Proc.of IEEE ITCC(Apr.2001)223-227;[5]J.Tian,“Wavelet-based Reversible Watermarking forAuthentication,”In:Proc.of SPIE,vol.4675(2002)679-690;[6]MUCelik,G.Sharma,E.Saber,and TAMurat,“LocalizedLossless Authentication Watermark,”In:Proc.of SPIE,vol.5002(2003)689-698;[7]MUCelik,G.Sharma,AMTekalp,“Lossless watermarkingfor image authentication:A new framework and an implementation.”In:IEEE Trans.Image Processing,vol.15,no.4(Apr.2006)1042-1049。 Related technical literature has five, as follows: [3] J.Fridrich, M.Goljan, and R.Du, "Invertible Authentication," In: Proc.of SPIE, vol.3971 (Jan.2001) 197-208; [4] J.Fridrich, M.Goljan, and R.Du, "invertible AuthenticationWatermark for JPEG Images," In: Proc.of IEEE ITCC (Apr.2001) 223-227; [5] J.Tian, ​​"Wavelet- based Reversible Watermarking forAuthentication, "In: Proc.of SPIE, vol.4675 (2002) 679-690; [6] MUCelik, G.Sharma, E.Saber, and TAMurat," LocalizedLossless Authentication Watermark, "In: Proc.of SPIE, vol.5002 (2003) 689-698; [7] MUCelik, G.Sharma, AMTekalp, "Lossless watermarkingfor image authentication: A new framework and an implementation." In: IEEE Trans.Image Processing, vol.15, no .4 (Apr.2006) 1042-1049.

传统的数字水印方法将水印嵌入到图像中,造成原始图像的永久失真,在接收端只能获得嵌有了水印的水印图像而无法获得原始图像,这在不允许有任何失真的敏感图像场合如军事、医学、新闻出版等是不适合的。 The traditional method of digital watermarking watermark is embedded into the image, causing permanent distortion of the original image at the receiving end can only be obtained with the watermark embedded watermark image and the original image can not be obtained, which does not allow any distortion of image-sensitive applications such as military, medicine, journalism, publishing, etc. are not suitable. 可逆水印技术则可以解决这一问题。 Reversible watermarking technology can solve this problem. 当水印图像通过认证后,可获得无失真的原始图像。 When the original image is watermarked image through certification obtained without distortion. 现有的用于图像认证的可逆水印方法主要有三类,第一类方法纯粹使用无损压缩技术,如专利[1]和文献[3]、[6]、[7]中的方法(文献[6]、[7]是一个方法,此处称为Celik方法),即采用无损压缩技术直接压缩图像比特序列,因无损压缩而腾出来的“空间”用于嵌入认证等信息。 Conventional reversible watermarking for image authentication methods are mainly three types, the first type method purely lossless compression technique, as disclosed in [1] and [3], [6], [7] in ([6 ], [7] a method, referred to herein as Celik method), i.e., lossless compression techniques directly compressed image bit sequence, by the lossless compression freed "space" for embedding authentication information. 第二类方法使用特殊的方法并结合无损压缩技术,如文献[3]和[5]中的方法,其中文献[3]先导出量化JPEG系数的偏移比特流,然后对这些偏移比特流进行无损压缩以腾出“空间”给认证等信息,而文献[5]使用了差值扩展和通用最低比特嵌入的方法先嵌入认证等信息,为了确定信息嵌入的位置,需要将位置比特图进行无损压缩并一同嵌入。 The second method uses a special class of methods in conjunction with lossless compression techniques, such as literature and methods [5] [3], wherein [3] Pilot JPEG bitstream quantized offset coefficient, and these offsets bitstreams lossless compression to make "room" for determining the position of the embedded information, it is necessary for the position of the bitmap to the authentication information, and [5] using the lowest bit difference expansion and general method to embed embedding authentication information, lossless compression and fitted together. 第二类方法完全不借助无损压缩技术,如专利[2]中的方法,该方法在空域上对象素进行模256加法运算,利用运算结果的循环性可在接收端恢复原始图像。 The second class of methods without the use of completely lossless compression techniques, as described in patent [2], the method of adding pixels modulo 256 in the spatial domain, using the calculation results may be circulating at the receiving end to restore the original image. 第一、第二类的方法因为采用无损压缩技术,因此运算量大,系统实现较复杂,对于纹理复杂的图片,无损压缩技术无法提供足够的嵌入容量,因此往往难以实用。 First, because the method of the second type of lossless compression technology, large amount of computation to achieve more complex systems, complex texture for images, lossless compression techniques do not provide sufficient embedding capacity, it is often difficult practical. 而第三类方法中专利[2]的方法则导致待认证图像视觉质量差,因为边缘象素值的变化太大。 And the third type of method of Patent [2] is to be authenticated image results in poor visual quality, since the edge pixel values ​​change too. 对于图像认证方法、技术来说,篡改定位是非常重要的内容和功能,可惜在我们所知道的现有文献和专利中,除了Celik方法能够提供篡改定位功能,其他的方法都无法提供篡改定位功能。 For image authentication methods, techniques, the tamper localization is a very important content and functionality, but in the existing literature and patents we know, in addition to Celik method can provide tamper positioning function, other methods are unable to provide tamper positioning function . Celik方法采用基于内容、自适应、的无损图像压缩编解码方法,嵌入容量较小,对于纹理复杂的图片,难以应用无损压缩技术,不适用于纹理复杂的图片。 Method Celik content based, adaptive, lossless image compression encoding and decoding method, embedding capacity is small, complex texture for images, lossless compression technique is difficult to apply, does not apply to complex texture images.

发明内容 SUMMARY

为了克服现有主要技术采用无损压缩、嵌入容量小、难以提供篡改定位功能的不足,本发明的目的是提供一种嵌入容量大、可抵抗伪造攻击、具有篡改定位功能的用于图像认证的可逆水印方法。 In order to overcome the prior art mainly use lossless compression, embedding a small capacity, it is difficult to provide tamper insufficient positioning function, object of the present invention is to provide an embedding capacity, resistant to forgery attack, reversible positioning function with tamper image authentication watermarking method.

本发明所采用的技术方案如下:本发明方法是将用于内容完整性检验的原始图像的SHA-256哈希值和用于图像恢复的数据通过修改高频子带直方图嵌入到高频子带中,将用于篡改检测与定位的有意义的水印通过替换低频子带的最低比特平面嵌入到低频子带中,验证时用提取的水印或者水印的差值图像进行篡改检测与定位,用提取出来的SHA-256哈希值和恢复的图像的SHA-256哈希值进行比较以验证内容完整性。 Used in the present invention, the following technical solution: a method of the present invention is a SHA-256 hash value of the original image and the content integrity check data for image restoration by modifying the high frequency sub-band to the high frequency sub-histograms fitted band, meaningful for tamper detection and location of the watermark embedded in the lower sub-band bit-planes by replacing the lowest frequency sub band, verifying the extracted watermark with the watermark or the difference image for tamper detection and location, with extracted from the SHA-256 hash value SHA-256 hash value and restore images were compared to verify the integrity of the content.

该方法具体分为数据嵌入和图像认证两个过程,其中数据嵌入的步骤为:1)计算原始图像XMxN的SHA-256哈希值;2)直方图修正:即将XMxN直方图的范围从[0,255]修正为[G,255-G],G的初始默认取值为0或根据XMxN直方图设定,并记录被改动象素为PM,包括原值、坐标和G,修正后的图像记为X'MxN;3)整数小波分解:即对X'MxN进行整数小波分解,分解级数取K=3,获得一个低频子带LLK和高频子带集合Ck,l,根据各个高频子带系数直方图的最高点值获得各个高频子带的嵌入容量,估计待嵌入容量并按照人类视觉系统(HVS)特性选取用于嵌入数据的高频子带,记录被选取的高频子带的最高点值对应的系数值为 The particular method and image authentication data embedding into two processes, wherein the data embedding steps of: 1) calculating an original image XMxN the SHA-256 hash value; 2) Histogram Modification: XMxN coming from the histogram range [0 , 255] is corrected to [G, G-255], the initial default value of 0 or G is set according XMxN histogram, and changes the pixel is recorded as PM, including the original value, and the coordinates of the image G, the correction referred to as X'MxN; 3) integer wavelet decomposition: X'MxN performed on integer wavelet i.e. the decomposition stages taking K = 3, to obtain a lower sub-band and the high frequency subband sets LLK Ck, l, according to the respective frequency subband coefficients histogram highest point value obtained embedding capacity of each frequency sub-band, and the capacity is estimated to be embedded in accordance with the human visual system (the HVS) characteristics selected for the high frequency sub-band of the embedded data, record the selected frequency sub with the highest point value of the coefficient corresponds to a value 4)低频系数修改:将经过构造的、大小与LLK相同的有意义二值水印图像W用密钥K加密后以替换LLK系数最低比特平面的方式嵌入LLK,用密钥K选择LLK系数次低比特位,将 4) low-frequency coefficients Review: After the configuration, the same size as the meaningful LLK binary watermark image encryption with key K W to replace planar manner LLK LLK embedded lowest bit coefficient, the coefficient selection key K times lower LLK bit, the 以替换被选择的次低比特位的方式嵌入LLK得到LL'K,并记录被替换的LSB和次低比特为Ori_bits;5)高频系数修改:将SHA-256哈希值、PM和Ori_bits组成比特流并用密钥K加密,通过修改被选取的高频子带系数直方图将比特流嵌入Ck,l,记为C'k,l;6)整数小波重构,将LL'K和C'k,l进行三级整数小波重构;7)象素值溢出判断:如果重构后的象素值超出[0,255]的范围,判为溢出,则重返步骤2),增大G的取值并重复上述步骤直至没有象素值溢出,G的步长默认取5或自行设定,否则得到待认证图像X”MxN。 As to the second lowest bit of the selected alternative embedded LLK obtained LL'K, and recording the replaced bits LSB and the next lowest Ori_bits; 5) to modify the high frequency coefficients: the SHA-256 hash value, PM, and composition Ori_bits and the bit stream with an encryption key K, by modifying the selected frequency subband coefficients histogram bitstream embedding Ck, l, referred to as C'k, l; 6) integer wavelet reconstruction, the LL'K and C ' k, l for three integer wavelet reconstruction; 7) determines the pixel value is overflowed: if the reconstructed pixel values ​​outside the range of [0,255], the overflow is judged, then return to step 2), increases G values ​​and the above steps are repeated until there is no pixel value overflows, step G takes a default set their own or 5, or the image to be authenticated to obtain X "MxN.

3、根据权利要求书2所述的用于图像认证的可逆水印方法,其特征是所述图像认证的步骤为:1)整数小波分解:即对待认证图像X”MxN进行三级整数小波分解,获得一个低频子带LL”K和高频子带集合C”k,l;2)水印提取与第一步认证:提取LL”K的最低比特平面获得水印W*',观察W*'或者将W*'与W进行比较,如果W*'被篡改,表明X”MxN是不可信的,图像不能通过认证,此时根据W*'与X”MxN的位置对应关系对X”MxN进行篡改定位,如果W*'未被篡改,X”MxN进入如下步骤的第二步认证;3)提取并恢复高频系数和低频系数:由密钥K获得LL”K中 3, the reversible watermarking for image authentication method according to claims 2, wherein said image authentication the steps of: 1) Integer Wavelet Decomposition: i.e., to treat the authentication image X "MxN for three integer wavelet decomposition, to obtain a low frequency sub-band LL "K and higher sub-band set C" k, l; 2) Step watermark extraction and authentication: the extracted LL "K lowest bit-planes to obtain the watermark W * ', observed W *' or W * 'is compared with W, if W *' has been tampered with, show X "MxN is not credible, the image can not be authenticated, according to the case W * 'and X" correspondence between X "MxN tamper localization position of MxN , if W * 'is not tampered, X "MxN step into the second step following authentication; 3) to extract and recover high-frequency coefficients and low frequency coefficients: LL obtained by the key K" K in 嵌入的次低比特位置并提取 Embedded and extracting the next lower bit positions by 从C”k,l中提取比特流并同时恢复C”k,l得到Ck,l,用密钥K解密并分析比特流得到SHA-256哈希值、PM和Ori_bits,用Ori_bits替换LL”K的最低比特平面和用于嵌入 "Extracted bit stream k, l and simultaneously recover C" from C k, l to obtain Ck, l, and decryption key K to obtain a bit stream analyzing SHA-256 hash value, PM, and Ori_bits, replacing LL by Ori_bits "Low bit-planes for embedding and K 的次低比特位从而恢复LL”K得到LLK;4)整数小波重构:将LLK和Ck,l进行三级整数小波重构,得到图像信号XMxN;5)直方图恢复:由PM将XMxN的直方图[G,255-G]恢复为[0,255],得到恢复的图像;6)比较提取的SHA-256哈希值和恢复的图像的SHA-256哈希值,若两者相等,表明图像是可信的,图像通过认证,此时恢复的图像便是无失真的原始图像,否则表明图像是伪造的或者图像已经被篡改但却不能定位篡改,图像不能通过认证。 The next lower bit thereby recovering LL "K obtained LLK; 4) Integer Wavelet Reconstruction: The LLK and Ck, l for three integer wavelet reconstruction, an image signal obtained XMxN; 5) Histogram FIG recovery: the XMxN histogram by the PM [G, 255-G] recovery of [0,255], to obtain a restored image; SHA- 6) comparing the extracted SHA-256 hash value of the image and the restored 256 hash, if the two are equal, indicating that the image is credible, certified image, this time to restore the image of the original image is distortion-free, otherwise the show is fake image or images have been tampered with but can not locate tampered with the images can not be authenticated.

同现有技术相比,本发明具有以下优点:1)嵌入容量大,视觉效果好,具有可逆性。 Compared with the prior art, the present invention has the following advantages: 1) embedding capacity, good visual effects are reversible. 现有的用于图像认证的可逆水印方法或多或少都采取无损压缩技术将载体信号或经处理后的载体信号的部分比特流进行无损压缩,设压缩前的比特流长度为L,无损压缩后长度为Ls,则可嵌入的比特长度不超过L-Ls。 Conventional reversible watermarking for image authentication method of lossless compression techniques are more or less taking the partial bitstream carrier signal after the carrier signal or the lossless compression processing, the bit stream length before compression set as L, lossless compression after a length of Ls, it may be embedded bit length is L-Ls. 由于自然灰度图像的纹理一般比较复杂,特别是较低的比特平面,具有类似随机噪声的特性,现有的可逆认证方法主要是压缩较低的比特平面,因而无损压缩的效果不理想,嵌入容量很小,虽然压缩较高的比特平面可增加嵌入容量,但是却会导致水印图像较差的视觉效果。 Since the natural texture effect grayscale image is generally more complicated, especially lower bit-plane, has a characteristic similar to random noise, the conventional authentication method is mainly reversible compressed low bit-plane, thus not ideal lossless compression, embedded capacity is small, although higher compression bit-plane can increase the embedding capacity, but results in poor visual effect watermark. 对于纹理复杂的图像来说,无损压缩的结果往往是数据长度的增加而不是减少,即嵌入容量为负。 For complex texture images, lossless compression often results in an increase rather than decrease data length, i.e., the embedding capacity is negative. 本发明通过修改高频子带系数直方图将认证信息和用于恢复的数据嵌入到高频子带中,取得较大的嵌入容量。 The present invention is a high frequency subband coefficients by modifying the histogram data embedding authentication information for restoration to a high frequency subband to obtain a larger embedding capacity. 由于本发明只是对小波系数作轻微的修改,因此具有较好的视觉效果。 Since the present invention is only to make minor modifications to the wavelet coefficients, thus having a better visual effect. 当水印图像没有受到任何篡改时可无失真恢复原始图像。 When the watermarked image has not been any tampering can restore the original image without distortion. 下表1显示了一系列测试图像的性能。 Table 1 below shows the performance of a series of test images.

表1 Table 1

2)精确的篡改定位功能。 2) Accurate positioning function tampering. 现有的用于图像认证的可逆水印方法难以提供准确的篡改定位功能,而本发明通过在整数小波的低频子带嵌入有意义的水印,认证时通过提取的水印或提取的水印与原始水印的差值图来确定篡改的区域。 Conventional reversible watermarking for image authentication method is difficult to provide accurate positioning function tampering, by the present invention with a watermark embedded by the low frequency sub-integer wavelet transform, watermark authentication or by extracting the extracted watermark with the original watermark FIG area difference determined tampering.

3)具有较好的安全性。 3) have better security. 本发明两步认证的验证不但可以实现严格的完全级的内容认证,而且一定程度上能够抵抗伪造攻击,即水印不变但图像内容改变的攻击,具有较好的安全性。 Two-step authentication of the present invention can be achieved not only strict verification of full-level content authentication, and can resist forgery attack to some extent, but that is the same watermark image content changes of the attack, with better security. 如果攻击者实现了伪造攻击,即水印没有篡改,则该伪造图像能通过了第一步认证,但是由于重新计算的哈希值与提取出来的不同,因此该伪造不能通过第二步认证,因而伪造图像无法骗过认证系统。 If an attacker realizes forgery attack, namely the watermark is not tampered with, forged the image through the first step of certification, but due to different recalculate the hash value extracted, so the second step can not be forged certification, thus counterfeit image can not fool the authentication system. 另外,本发明使用密钥对水印、比特流加密,并使用密钥选择信息P的嵌入位置,这三个密钥的选取可以相同也可以不同,视具体系统的实现而定,三重密钥的使用增加了攻击者破解的难度,增强了系统的安全性。 In addition, the present invention is a watermark key, encrypted bitstream, key selection and use of the embedded position information P, select the three keys may be the same or different, depending on the particular system and can not be implemented, the triple key use increases the difficulty of the attacker to crack, and enhance the security of the system.

附图说明 BRIEF DESCRIPTION

图1是本发明方法的实现原理框图;图2为在高频子带中嵌入数据时直方图修改的示意图;图3为本发明第一步认证的实现框图;图4为本发明第二步认证的实现框图;图5为本发明对大小为512×512×8的bmp图像Lena的效果图;图6为经过篡改的待认证图像及其水印差值图;图7为经过篡改的待认证图像及其水印差值图。 FIG. 1 is a block diagram of the method of the present invention is implemented; schematic histogram modification of Figure 2 the data embedded in the high frequency sub-band; FIG. 3 is a first step towards the present invention, a block diagram of authentication; FIG. 4 of the present invention, the second step authentication realization diagram; FIG. 5 FIG effect bmp image Lena size of 512 × 512 × 8 of the present invention; FIG. 6 is a tampered image and the watermark to be authenticated difference map; FIG. 7 is to be authenticated through tampered a difference image and the watermark FIG.

具体实施方式 Detailed ways

本发明可应用在一般的图像和敏感图片比如军事图片、医学图片与新闻出版等的图片中。 The present invention can be applied to general image picture and sensitive picture images such as military, medical images and news publishing in. 这里以具有代表性的大小为512×512的灰度图像lena.bmp为例,见图5所示,其中(a)为原始图像Lena,(b)为嵌入了4407比特的待认证图像,PSNR为46.24dB。 Here a size representative of grayscale image lena.bmp 512 × 512 for example, shown in Figure 5, wherein (a) is the original image Lena, (b) to be embedded in the authentication image of 4407 bits, the PSNR as 46.24dB. 下面结合附图描述本发明的实施方法。 BRIEF described below in connection with the method of the present invention.

图1所示为本发明数据隐藏的实现原理框图如,其中X′NxM表示经过直方图修正后的图像,Ck,l、LLK分别表示整数小波分解后的九个高频子带的集合和低频子带,C'k,l、LL'K分别表示嵌入了数据后的九个高频子带的集合和低频子带。 The present invention is shown in FIG. 1 a block diagram of a data hiding implementation principle as wherein represents X'NxM image histogram after correction, Ck, l, LLK represent a collection of nine high-frequency and low-frequency sub-band after wavelet decomposition of an integer subband, C'k, l, LL'K respectively embedded nine sets and lower sub-band data after the high frequency sub band. 如图1所示,在发送端,首先计算原始图像的SHA-256哈希值,然后对图像进行直方图修正。 As shown, the transmitting side, the original image is calculated first SHA-256 hash value of 1, and then the image histogram modification. 直方图修正过程为:先分析图像的象素直方图,lena图像位于[0,255]边缘的象素很少,因为本发明的数据隐藏方案对整数小波系数的改动很小,因此可以在一开始的时候取G=10或经过2次循环得到;再对象素直方图进行修改,[0,10]范围内没有象素,[245,255]范围内有一个象素(247;274,117),即象素值为247,平面坐标为(274,117),将象素值改为237;将(247;274,117)保存为PM。 Histogram modification procedure is: first histogram analysis of the image pixels, image pixels located Lena [0,255] edge rarely, because the data hiding scheme of the present invention changes to the integer wavelet coefficients is small, it is possible in a beginning to take G = 10 or obtained after 2 cycles; then modify the pixel histogram, the pixel is not within the range [0,10], a pixel (in the range of 247 [245,255]; 274,117 ), i.e., the pixel value of 247, the plane coordinates (274,117), the pixel value to 237; and (247; 274,117) save as PM.

接着对经过直方图修正后的图像使用提升格式如下进行整数小波分解,sl(0)=x2ldl(0)=x2l+1---dl(1)=dl(0)+Int(α(sl(0)+sl+1(0)))sl(1)=sl(0)+Int(β(dl(1)+dl-1(0)))---dl2=dl(1)+Int(γ(sl(1)+sl+1(1)))sl(2)=sl(1)+Int(δ(dl(2)+dl-12))]]>式组(1)dl(3)=dl(2)+Int((ζ-ζ2)sl(2))sl(3)=sl(2)+Int((-1/ζ)dl(3))---dl(4)=dl(3)+Int((ζ-1)sl(3))sl(4)=sl(3)+dl(4)---sl=sl(4)dl=dl(4)]]>式组(2)其中Int(·)函数是取整函数,5个参数的值分别为α=-1.586134342,β=-0.05298011854,γ=0.8829110762,δ=0.4435068522,ζ=1.149604398。 Next, after the image using the corrected histogram format for lifting integer wavelet decomposition, sl (0) = x2ldl (0) = x2l + 1 --- dl (1) = dl (0) + Int (& alpha; (sl (0) + sl + 1 (0))) sl (1) = sl (0) + Int (& beta; (dl (1) + dl-1 (0))) --- dl2 = dl (1) + int (& gamma; (sl (1) + sl + 1 (1))) sl (2) = sl (1) + int (& delta; (dl (2) + dl-12))]]> formula group (1 ) dl (3) = dl (2) + Int ((& zeta; - & zeta; 2) sl (2)) sl (3) = sl (2) + Int ((- 1 / & zeta;) dl (3)) --- dl (4) = dl (3) + Int ((& zeta; -1) sl (3)) sl (4) = sl (3) + dl (4) --- sl = sl (4) dl = dl (4)]]> group of formula (2) where Int (·) function is a rounding function, each parameter value of 5 is α = -1.586134342, β = -0.05298011854, γ = 0.8829110762, δ = 0.4435068522, ζ = 1.149604398. 本方案采用三级整数小波分解,这样得到9个高频子带和1个低频子带,记录低频子带的最低比特平面为Ori_bits的一部分,该部分的大小为64×64比特。 The program uses three integer wavelet decomposition, 9 thus obtained and a higher sub-band low frequency sub-band, the bit-plane recording lowest frequency subband is part Ori_bits, the size of the portion was 64 × 64 bits. 接着检测9个高频子带系数直方图的最高点对应的系数值及其嵌入容量,记录如下表2所示:表2 9 then detects a histogram of high frequency subband coefficients of the coefficient value corresponding to the highest point of its embedding capacity, recorded in Table 2 below: Table 2

然后根据待嵌入高频子带比特流的长度和结构,即哈希值的比特长度、PM的比特长度、LLK最低比特平面大小、九个高频子带各自系数直方图最高点对应的系数值的比特长度,估计待嵌入容量的最大值为256+36+4096+72=4460bits。 The coefficient value is then embedded in the length and structure to be higher sub-band bit-stream, i.e. the bit length of the hash value, the PM bit length, the bit-plane LLK minimum size, nine higher sub-band coefficients each corresponding to the highest point of the histogram bit length, the maximum value of the estimated capacity to be embedded is 256 + 36 + 4096 + 72 = 4460bits. 再然后按照人类视觉系统(HVS)特性选取用于嵌入数据的高频子带HH1,该高频子带的嵌入容量为11243比特,因此选取一个高频子带足够了,并记录HH1系数直方图的最高点对应的系数值-3为P。 And then according to the sub-frequency human visual system (the HVS) characteristics for the selected band HH1 embedded data, embedded in the higher sub-band capacity is 11,243 bits, select a higher sub-band is sufficient, and the coefficient histogram recording HH1 the coefficient values ​​corresponding to the highest point of -3 is P. 接着用经过加密后的有意义水印替换LLK的最低比特平面,将P替换用密钥K选中的LLK次比特位,被替换掉的原始比特记为Ori_bits。 After meaningful followed by replacing the encrypted watermark LLK lowest bit-plane, will be replaced with the P key K bits selected LLK times, be replaced is referred to as the original bit Ori_bits. 我们将SHA-256哈希值、PM、Ori_bits组成比特流,加密后加上头部表明比特流及各组成部分的长度,最后的比特流总长度为4407比特。 We SHA-256 hash value, PM, Ori_bits composition bitstream, together with the encrypted header indicates that the bitstream length and each portion of the composition, the total length of the last bit stream of 4407 bits.

修改HH1的直方图,将上述比特流嵌入HH1。 HH1 modified histogram, the bit stream is embedded HH1. 比特流嵌入HH1的示意图见图2,图中假设待嵌入的比特序列为“10 11 01 00 10”。 HH1 embedded bitstream schematic shown in Figure 2, the figure is assumed to be embedded in the bit sequence "1011010010." 具体做法为:首先,按照从上到下、从左到右的顺序对HH1扫描,当遇到大于-3的系数值时,给该系数值加1;然后再次按照同样的顺序扫描,当遇到值为-3的系数时,检测待嵌入的比特,如果待嵌入的是比特“1”,则将最高点系数值-3加1而成为-2,如果待嵌入的是比特“0”,则保持最高点系数值-3不变,直到将所有的数据比特嵌入。 Specifically: first, from top to bottom, left to right scan of HH1, encountered when the coefficient value is greater than -3, to the coefficient value plus 1; and again in the same scan order, when the case of when the value of the coefficient of -3, the bit to be embedded is detected, if use is to be bit "1", the highest point of the coefficient values ​​will be adding 1 -2 -3, it is to be embedded if the bit "0", the highest point is maintained constant coefficient values ​​-3, until all the data bit is embedded.

最后,按照上述式组(1)和式组(2)的反过程联合嵌入了数据的高频子带和低频子带进行三级整数小波重构得到重构的图像,检查重构的图像是否发生象素溢出,若有溢出,则增加G值(一般步长设为5),按照上述步骤依次重新开始;若没有溢出,则得到待认证图像,其视觉效果见图5(b),PSNR为46.24dB。 Finally, according to the above group of formula (1) and a group of formula (2) embedded joint inverse of the sub-band and low-frequency sub-band data of three integer wavelet reconstruction to obtain a reconstructed image, checking whether the reconstructed image pixel overflow occurs, if overflow, G value is increased (typically to step 5), according to the above steps are sequentially restarted; if there is no overflow, the resulting image to be authenticated, the visual effect is shown in Figure 5 (b), PSNR as 46.24dB.

本发明图像认证部分分为两步,只有当待认证图像通过第一步认证后,待认证图像才能进入第二步认证,其实现框图分别如图3和图4所示。 The image authentication section two steps of the present invention, only when the image to be authenticated through the authentication after the first, the second step to enter the authentication image to be authenticated, a block diagram of its implementation are shown in Figures 3 and 4. 在第一步认证中,首先对接收到的待认证图像进行三级整数小波分解,获得9个高频子带和1个低频子带,提取低频子带的最低比特平面获得水印,比较原水印与提取水印获得水印差值图像。 In the first step authentication, the first image to be authenticated the received three integer wavelet decomposition, sub-band to obtain frequency 9 and a lower sub-band, the bit-plane extraction lowest frequency subband to obtain a watermark, the watermark of the original comparison obtain and extract the watermark image watermark difference. 如果水印差值图像有白点,则水印图像已经被篡改,白点表示篡改的区域;如果水印差值图像无白点,则进入下一步验证,即第二步认证。 If the watermark difference image with white spots, the watermarked image has been tampered with, the white dots indicate tampering region; if no white spots watermarked image difference, the process proceeds to the next authentication, i.e., authentication second step.

在第二步认证中,先从低频子带的次比特平面提取 In the second authentication, the start times lower sub-band bit-plane extraction (本例中 (In this case ), ), 取的位置由密钥获得。 The position taken by the key is obtained. 然后根据 Then according to 从高频子带中提取比特流,具体做法为:首先,按照从上到下、从左到右的顺序对HH1扫描,当遇到值为-3的系数时,提取比特“0”,当遇到值为-2的系数时,提取比特“1”,同时将系数值恢复为-3,如此直到将所有的数据比特提取完毕;然后,再次按照同样的顺序对HH1扫描,当遇到大于-3的系数值时,给该系数值减1,直到整个HH1扫描完毕,这样就恢复了原来的高频子带HH1。 Extracting a bit stream from the high frequency sub-band, the specific approach: first, from top to bottom, left to right scan of HH1, encountered when the coefficient value of -3 extracted bit "0", when -2 coefficient value is encountered, the extraction bit "1", while the coefficient values ​​restored -3, and so on until all the data bits extraction is completed; and then, again in the same order of scanning HH1, when encountering greater than when the coefficient values ​​-3, to the coefficient value is decremented by 1, until the entire scan is completed HH1, so to restore the original high frequency sub-band HH1. 解密并分析比特流获得SHA-256哈希值、PM、Ori_bits。 Decrypts and analyzes the bit stream obtained SHA-256 hash value, PM, Ori_bits. 用提取出来的Ori_bits替换低频子带的最低比特平面和选作嵌 Alternatively lower sub-band with a minimum Ori_bits extracted and selected as the embedded bit-planes 的次比特位,从而恢复原来的低频子带。 Secondary bits, thereby restoring the original frequency subband. 联合已经恢复了的高频子带和低频子带进行三级整数小波重构得到重构的图像,再根据PM恢复被改动过的象素值,获得恢复的图像,最后计算其SHA-256值。 United has restored high frequency subband and low frequency subband to obtain three integer wavelet reconstruction for image reconstruction, and then to restore the pixel value is altered based on PM, a restored image is obtained, the last calculated value SHA-256 . 比较提取的SHA-256值和恢复图像的SHA-256值,若两者匹配则表明待认证图像是可信的,图像通过认证,恢复的图像就是无失真的原始图像;若两者不匹配表明待认证图像是不可信的,图像是伪造的或已经被篡改但却无法定位篡改,也即通过了第一步认证却无法通过第二步认证。 Comparing the extracted value SHA-256 and SHA-256 value of the restored image, if the two match indicates that the authentication image is to be trusted, authentication by image, the restored image is undistorted original image; show if the two do not match authentication image is to be credible, the image is fake or has been tampered with but can not locate tampered with, that it can not pass the first step of certification by the second step certification.

图6、7为本发明的两个应用实例。 6 and 7 two application examples of the present invention. 图6为经过篡改的待认证图像及其水印差值图。 6 is tampered watermark to be authenticated and the difference image of FIG. 其中图6(a)为经过画笔修改的待认证图像,图6(c)为经过粘贴修改的待认证图像,图6(b)和图6(d)是相应的水印差值图,白点表示篡改的区域。 Wherein FIG. 6 (a) to be authenticated through image brush modification, FIG. 6 (c) is attached via a modified image to be authenticated, FIG. 6 (b) and 6 (d) is a corresponding watermark difference map, white spots It represents a region of tampering. 图7为经过篡改的待认证图像及其水印差值图。 7 is to be authenticated through image and the watermark difference tampering FIG. 其中图7(a)为一有意义的水印,图7(b)为将图7(a)嵌入在空域中央的待认证图像,图7(c)为水印差值图,白点表示篡改的区域。 Wherein FIG. 7 (a) is a meaningful watermark, FIG. 7 (b) to FIG 7 (a) to be embedded in the authentication center airspace image, FIG. 7 (c) represents a difference FIG tampered watermark, white dots region. 被篡改的待认证图像图7(b)在视觉效果上与原待认证图像图5(b)没有任何差别,但篡改却被精确检测并定位。 Been tampered image to be authenticated in FIG. 7 (b) in the visual effects to be authenticated as the original image of FIG. 5 (b) there is no difference, but the tampering was accurately detected and located.

Claims (3)

1.一种可抵抗伪造攻击、检测篡改并定位的用于图像认证的可逆水印方法,其特征是将用于内容完整性检验的原始图像的SHA-256哈希值和用于图像恢复的数据通过修改高频子带直方图嵌入到高频子带中,将用于篡改检测与定位的有意义的水印通过替换低频子带的最低比特平面嵌入到低频子带中,验证时用提取的水印或者水印的差值图像进行篡改检测与定位,用提取出来的SHA-256哈希值和恢复的图像的SHA-256哈希值进行比较以验证内容完整性。 SHA-256 hash value of the original image 1. A resist forgery attack, reversible watermarking method for image authentication and tamper detection location, characterized in that the content for the integrity verification data and image restoration by modifying the high frequency subband to higher frequency subband histograms fitted, the tamper detection for meaningful watermark embedding and positioned in the lower sub-band bit-planes by replacing the lowest frequency sub band, using the extracted watermark verification difference image or watermark tampering detection and localization, extracted with SHA-256 hash SHA-256 hash value and the value of the recovered image are compared to verify the integrity.
2.根据权利要求书1所述的用于图像认证的可逆水印方法,其特征是该方法具体分为数据嵌入和图像认证两个过程,其中数据嵌入的步骤为:1)计算原始图像XMxN的SHA-256哈希值;2)直方图修正:即将XMxN直方图的范围从[0,255]修正为[G,255-G],G的初始默认取值为0或根据XMxN直方图设定,并记录被改动象素为PM,包括原值、坐标和G,修正后的图像记为X'MxN;3)整数小波分解:即对X'MxN进行整数小波分解,分解级数取K=3,获得一个低频子带LLK和高频子带集合Ck,l,根据各个高频子带系数直方图的最高点值获得各个高频子带的嵌入容量,估计待嵌入容量并按照人类视觉系统(HVS)特性选取用于嵌入数据的高频子带,记录被选取的高频子带的最高点值对应的系数值为 The book reversible watermarking for image authentication method according to claim 1, characterized in that the particular method and image authentication data embedding into two processes, wherein the data embedding steps of: 1) calculating the original image XMxN SHA-256 hash value; 2) histogram Modification: XMxN coming from the range of the histogram [0,255] is corrected to [G, 255-G], G initial default value is 0 or according to the histogram XMxN and recording the pixel has been altered as PM, including the original value, and the coordinates G, referred to as the corrected image X'MxN; 3) integer wavelet decomposition: X'MxN performed on integer wavelet i.e. the decomposition stages take K = 3, to obtain a lower sub-band and the high frequency subband sets LLK Ck, l, embedding capacity is obtained according to the respective frequency sub-band peak frequency subband coefficients of each histogram value, and the capacity is estimated to be embedded in accordance with the human visual system (the HVS) characteristics for embedding the selected frequency sub-band data, record the selected frequency sub-band corresponding to the highest value of the coefficient value 4)低频系数修改:将经过构造的、大小与LLK相同的有意义二值水印图像W用密钥K加密后以替换LLK系数最低比特平面的方式嵌入LLK,用密钥K选择LLK系数次低比特位,将 4) low-frequency coefficients Review: After the configuration, the same size as the meaningful LLK binary watermark image encryption with key K W to replace planar manner LLK LLK embedded lowest bit coefficient, the coefficient selection key K times lower LLK bit, the 以替换被选择的次低比特位的方式嵌入LLK得到LL'K,并记录被替换的LSB和次低比特为Ori_bits;5)高频系数修改:将SHA-256哈希值、PM和Ori_bits组成比特流并用密钥K加密,通过修改被选取的高频子带系数直方图将比特流嵌入Ck,l,记为C'k,l;6)整数小波重构,将LL'K和C'k,l进行三级整数小波重构;7)象素值溢出判断:如果重构后的象素值超出[0,255]的范围,判为溢出,则重返步骤2),增大G的取值并重复上述步骤直至没有象素值溢出,G的步长默认取5或自行设定,否则得到待认证图像X”MxN。 As to the second lowest bit of the selected alternative embedded LLK obtained LL'K, and recording the replaced bits LSB and the next lowest Ori_bits; 5) to modify the high frequency coefficients: the SHA-256 hash value, PM, and composition Ori_bits and the bit stream with an encryption key K, by modifying the selected frequency subband coefficients histogram bitstream embedding Ck, l, referred to as C'k, l; 6) integer wavelet reconstruction, the LL'K and C ' k, l for three integer wavelet reconstruction; 7) determines the pixel value is overflowed: if the reconstructed pixel values ​​outside the range of [0,255], the overflow is judged, then return to step 2), increases G values ​​and the above steps are repeated until there is no pixel value overflows, step G takes a default set their own or 5, or the image to be authenticated to obtain X "MxN.
3.根据权利要求书2所述的用于图像认证的可逆水印方法,其特征是所述图像认证的步骤为:1)整数小波分解:即对待认证图像X”MxN进行三级整数小波分解,获得一个低频子带LL”K和高频子带集合C”k,l;2)水印提取与第一步认证:提取LL”K的最低比特平面获得水印W*',观察W*'或者将W*'与W进行比较,如果W*'被篡改,表明X”MxN是不可信的,图像不能通过认证,此时根据W*'与X”MxN的位置对应关系对X”MxN进行篡改定位,如果W*'未被篡改,X”MxN进入如下步骤的第二步认证;3)提取并恢复高频系数和低频系数:由密钥K获得LL”K中 The book reversible watermarking for image authentication method according to claim 2, wherein said image authentication the steps of: 1) Integer Wavelet Decomposition: i.e., to treat the authentication image X "MxN for three integer wavelet decomposition, to obtain a low frequency sub-band LL "K and higher sub-band set C" k, l; 2) Step watermark extraction and authentication: the extracted LL "K lowest bit-planes to obtain the watermark W * ', observed W *' or W * 'is compared with W, if W *' has been tampered with, show X "MxN is not credible, the image can not be authenticated, according to the case W * 'and X" correspondence between X "MxN tamper localization position of MxN , if W * 'is not tampered, X "MxN step into the second step following authentication; 3) to extract and recover high-frequency coefficients and low frequency coefficients: LL obtained by the key K" K in 嵌入的次低比特位置并提取 Embedded and extracting the next lower bit positions by 从C”k,l中提取比特流并同时恢复C”k,l得到Ck,l,用密钥K解密并分析比特流得到SHA-256哈希值、PM和Ori_bits,用Ori_bits替换LL”K的最低比特平面和用于嵌入 "Extracted bit stream k, l and simultaneously recover C" from C k, l to obtain Ck, l, and decryption key K to obtain a bit stream analyzing SHA-256 hash value, PM, and Ori_bits, replacing LL by Ori_bits "Low bit-planes for embedding and K 的次低比特位从而恢复LL”K得到LLK;4)整数小波重构:将LLK和Ck,l进行三级整数小波重构,得到图像信号XMxN;5)直方图恢复:由PM将X”MxN的直方图[G,255-G]恢复为[0,255],得到恢复的图像;6)比较提取的SHA-256哈希值和恢复的图像的SHA-256哈希值,若两者相等,表明图像是可信的,图像通过认证,此时恢复的图像便是无失真的原始图像,否则表明图像是伪造的或者图像已经被篡改但却不能定位篡改,图像不能通过认证。 The next lower bit thereby recovering LL "K obtained LLK; 4) Integer Wavelet Reconstruction: The LLK and Ck, l for three integer wavelet reconstruction, an image signal obtained XMxN; 5) Histogram FIG recovery: by the PM to X "MxN histogram [G, 255-G] recovery of [0,255], to obtain a restored image; SHA- 6) comparing the extracted SHA-256 hash value of the image and the restored 256 hash, if the two are equal, indicating that the image is credible, certified image, this time to restore the image of the original image is distortion-free, otherwise the show is fake image or images have been tampered with but can not locate tampered with the images can not be authenticated.
CN 200610036388 2006-07-07 2006-07-07 Reversible watermark method for image certification CN100394443C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200610036388 CN100394443C (en) 2006-07-07 2006-07-07 Reversible watermark method for image certification

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200610036388 CN100394443C (en) 2006-07-07 2006-07-07 Reversible watermark method for image certification

Publications (2)

Publication Number Publication Date
CN1885341A true CN1885341A (en) 2006-12-27
CN100394443C CN100394443C (en) 2008-06-11

Family

ID=37583481

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200610036388 CN100394443C (en) 2006-07-07 2006-07-07 Reversible watermark method for image certification

Country Status (1)

Country Link
CN (1) CN100394443C (en)

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN100580700C (en) * 2007-03-26 2010-01-13 浙江大学 Implicit writing analysis method based on pivot characteristic in implicit writing analysis system
CN101727572A (en) * 2008-10-20 2010-06-09 美国银行公司 Method for ensuring image integrity by using file characteristics
CN101527830B (en) 2008-03-07 2010-10-20 华为技术有限公司;中山大学 Method and device for embedding watermarking information and method and device for authenticating watermarking information
CN101582106B (en) 2009-06-19 2011-07-06 重庆邮电大学 Integrity inspection method for fine-grained data
CN101635855B (en) 2009-08-27 2011-08-03 北京国铁华晨通信信息技术有限公司 Method and device for embedding and blindly picking up video watermark
CN101572819B (en) 2009-06-03 2011-09-14 北京航空航天大学 Reversible image watermark method based on quantized DCT coefficient zero values index
CN102497597A (en) * 2011-12-05 2012-06-13 中国华录集团有限公司 Method for carrying out integrity checkout on HD (high-definition) video files
CN101661605B (en) 2008-08-26 2012-07-04 浙江大学 Embedding and positioning tampering methods of digital watermark and device thereof
US8639062B2 (en) 2007-10-09 2014-01-28 Bank Of America Corporation Ensuring image integrity using document characteristics
CN106067157A (en) * 2016-05-27 2016-11-02 陕西师范大学 Direction changing difference expansion and synchronous embedding reversible watermark embedding and extraction method
US9823958B2 (en) 2016-02-08 2017-11-21 Bank Of America Corporation System for processing data using different processing channels based on source error probability
US9952942B2 (en) 2016-02-12 2018-04-24 Bank Of America Corporation System for distributed data processing with auto-recovery
US10067869B2 (en) 2016-02-12 2018-09-04 Bank Of America Corporation System for distributed data processing with automatic caching at various system levels

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6064764A (en) 1998-03-30 2000-05-16 Seiko Epson Corporation Fragile watermarks for detecting tampering in images
US6714683B1 (en) 2000-08-24 2004-03-30 Digimarc Corporation Wavelet based feature modulation watermarks and related applications
CN1141639C (en) 2002-05-09 2004-03-10 宣国荣 Digital watermark method based on integer wavelet without damage to image

Cited By (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN100580700C (en) * 2007-03-26 2010-01-13 浙江大学 Implicit writing analysis method based on pivot characteristic in implicit writing analysis system
US8639062B2 (en) 2007-10-09 2014-01-28 Bank Of America Corporation Ensuring image integrity using document characteristics
CN101527830B (en) 2008-03-07 2010-10-20 华为技术有限公司;中山大学 Method and device for embedding watermarking information and method and device for authenticating watermarking information
CN101661605B (en) 2008-08-26 2012-07-04 浙江大学 Embedding and positioning tampering methods of digital watermark and device thereof
CN101727572A (en) * 2008-10-20 2010-06-09 美国银行公司 Method for ensuring image integrity by using file characteristics
CN101572819B (en) 2009-06-03 2011-09-14 北京航空航天大学 Reversible image watermark method based on quantized DCT coefficient zero values index
CN101582106B (en) 2009-06-19 2011-07-06 重庆邮电大学 Integrity inspection method for fine-grained data
CN101635855B (en) 2009-08-27 2011-08-03 北京国铁华晨通信信息技术有限公司 Method and device for embedding and blindly picking up video watermark
CN102497597A (en) * 2011-12-05 2012-06-13 中国华录集团有限公司 Method for carrying out integrity checkout on HD (high-definition) video files
CN102497597B (en) 2011-12-05 2013-11-06 中国华录集团有限公司 Method for carrying out integrity checkout on HD (high-definition) video files
US9823958B2 (en) 2016-02-08 2017-11-21 Bank Of America Corporation System for processing data using different processing channels based on source error probability
US9952942B2 (en) 2016-02-12 2018-04-24 Bank Of America Corporation System for distributed data processing with auto-recovery
US10067869B2 (en) 2016-02-12 2018-09-04 Bank Of America Corporation System for distributed data processing with automatic caching at various system levels
CN106067157A (en) * 2016-05-27 2016-11-02 陕西师范大学 Direction changing difference expansion and synchronous embedding reversible watermark embedding and extraction method
CN106067157B (en) * 2016-05-27 2019-06-11 陕西师范大学 Difference expansion of changing direction is embedded in the reversible water mark of synchronous insertion and extracting method

Also Published As

Publication number Publication date
CN100394443C (en) 2008-06-11

Similar Documents

Publication Publication Date Title
Fridrich et al. Invertible authentication
Zhu et al. When seeing isn't believing [multimedia authentication technologies]
Lin et al. Semifragile watermarking for authenticating JPEG visual content
JP3719347B2 (en) Digital image or watermark insertion method into compressed digital image, the digital image capturing device, a computer system, and tamper detection methods watermarked digital image
US7389420B2 (en) Content authentication and recovery using digital watermarks
Sun et al. A SVD-and quantization based semi-fragile watermarking technique for image authentication
Zhang et al. Fragile watermarking with error-free restoration capability
Lu et al. Cocktail watermarking for digital image protection
CN101449278B (en) System and method for robust reversible data hiding and data recovery in the spatial domain
Zhang et al. Statistical fragile watermarking capable of locating individual tampered pixels
US7822225B2 (en) Authentication of physical and electronic media objects using digital watermarks
US7970166B2 (en) Steganographic encoding methods and apparatus
US7424131B2 (en) Authentication of physical and electronic media objects using digital watermarks
US20010016058A1 (en) Method for extracting multiresolution watermark images to determine rightful ownership
US7360093B2 (en) System and method for authentication of JPEG image data
Wang et al. A majority-voting based watermarking scheme for color image tamper detection and recovery
Chang et al. Four-scanning attack on hierarchical digital watermarking method for image tamper detection and recovery
Qin et al. Self-embedding fragile watermarking with restoration capability based on adaptive bit allocation mechanism
Celik et al. Lossless watermarking for image authentication: a new framework and an implementation
JP2004507177A (en) How to embed a digital watermark to the digital image data
Zhu et al. A new semi-fragile image watermarking with robust tampering restoration using irregular sampling
Liu et al. The survey of digital watermarking-based image authentication techniques
US7313696B2 (en) Method for authentication of JPEG image data
Zhang et al. Multipurpose watermarking based on multiscale curvelet transform
Chamlawi et al. Digital image authentication and recovery: Employing integer transform based information embedding and extraction

Legal Events

Date Code Title Description
C06 Publication
C10 Request of examination as to substance
C14 Granted
C17 Cessation of patent right