公开(公告)号：EP1052612A1

公开(公告)日：2000-11-15

申请号：EP98911172.9

申请日：1998-04-01

申请人： Kowa Co., Ltd. ,  Matsui, Kineo

发明人： MATSUI, Kineo ,  IWAKIRI, Munetoshi

IPC分类号： G09C5/00 ,  H03M7/30 ,  G10L9/16

CPC分类号： G11B20/00086 ,  G10L19/018 ,  G10L21/00 ,  G10L25/00 ,  G11B20/00913 ,  H04B1/707

摘要： In the process of embedding watermark information in one-dimensional source data, first spread one-dimensional data x(t) is generated by subjecting one-dimensional source data s(t) to spectrum-spreading operation using a PN sequence g(t). This first spread one-dimensional data x(t) then undergoes a modified discrete cosine transformation. A specific frequency component value of the MDCT coefficients is modulated to provide association with the watermark information. The modulated frequency component then undergoes inverse transformation to generate second spread one-dimensional data x'(t). Then, using the same PN sequence in the spreading, the second spread one-dimensional data is subjected to spectrum-despreading to produce digitally watermarked one-dimensional data in which watermark data has been embedded.

摘要翻译： 在一维源数据中嵌入水印信息的过程中，通过使用PN序列g（t）对一维源数据s（t）进行频谱扩展操作来生成第一扩展一维数据x（t） 。 首先扩展的一维数据x（t）然后经历修正的离散余弦变换。 调制MDCT系数的特定频率分量值以提供与水印信息的关联。 调制频率分量然后进行逆变换以产生第二扩展一维数据x'（t）。 然后，在扩展中使用相同的PN序列，对第二扩展的一维数据进行频谱解扩，以产生嵌入了水印数据的数字水印的一维数据。

公开(公告)号：EP1052612B1

公开(公告)日：2006-06-07

申请号：EP98911172.9

申请日：1998-04-01

申请人： Kowa Company. Ltd. ,  Matsui, Kineo

发明人： MATSUI, Kineo ,  IWAKIRI, Munetoshi

IPC分类号： G09C5/00 ,  H03M7/30 ,  G10L19/00 ,  G11B20/00

CPC分类号： G11B20/00086 ,  G10L19/018 ,  G10L21/00 ,  G10L25/00 ,  G11B20/00913 ,  H04B1/707

摘要： In order to bury watermark information in an original one-dimensional data, first a 1st spread one-dimensional data x(t) is generated by spreading the spectrum of the original one-dimensional data s(t) by using a PN series g(t). Then the 1st spread one-dimensional data is subjected to modified discrete cosine transformation. The specific frequency component value of the MDCT coefficient is so adjusted as to correspond to the watermark information. By subjecting the adjusted frequency component to the reverse conversion, a 2nd spread one-dimensional data x'(t) is generated. Lastly, by the reverse spectrum-spreading using the same PN series as those used in the spectrum-spreading, watermark one-dimensional data in which the watermark information is buried is generated.

公开(公告)号：EP1139650A1

公开(公告)日：2001-10-04

申请号：EP99949436.2

申请日：1999-10-26

申请人： Kowa Co., Ltd.

发明人： MATSUI, Kineo

IPC分类号： H04N1/387

CPC分类号： G06T1/005 ,  G06T1/0028 ,  G06T2201/0065 ,  H04N1/32288 ,  H04N1/32309 ,  H04N2201/3233 ,  H04N2201/327

摘要： The present invention applies a technique, which is characteristic of color images, to embed a digital watermark in a color image and decode the embedded digital watermark. The technique adjusts a color signal transformation matrix by adding or subtracting a digital watermark matrix to or from the color signal transformation matrix in the process of transforming image data of an original color image in an RGB colorimetric system into image data in a YCbCr colorimetric system with the color signal transformation matrix. The digital watermark matrix includes an arbitrary numerical value of real number x, which represents digital watermark information, as an entry thereof and satisfies a condition that a sum of respective entries in each row and in each column is all substantially equal to zero. The technique inversely transforms the image data, which is obtained by the transformation with the adjusted color signal transformation matrix, into the image data in the RGB colorimetric system by means of an inverse color signal transformation matrix. The inversely transformed image has the numerical value of real number x embedded therein as the digital watermark. The numerical value of real number x is decoded from transformed image data, which is obtained by transforming the original image into color signals in the YCbCr colorimetric system.

摘要翻译： 本发明适用的技术，所有这些是彩色图像的特性，以嵌入在彩色图像的数字水印和解码嵌入的数字水印。 通过加上或减去一个数字水印矩阵或从与转化的原始彩色图像的图像数据在RGB表色系统为图像数据在YCbCr色度系统的过程中的彩色信号变换矩阵bestimmt一个彩色信号变换矩阵的技术 彩色信号转换矩阵。 数字水印矩阵包括在实数x，它darstellt数字水印信息的任意数值，作为一个入口和其外资企业SATIS的状态在每一行中，并在每列中没有respectivement条目的总和是所有基本上等于零。 该技术逆变换由所述变换与调节的彩色信号变换矩阵由逆颜色信号变换矩阵来获得，到RGB色度系统中的图象数据的图象数据，全部。 逆变换的图像具有的实数的数值x。其中嵌入作为数字水印。 实数x的数值从转化的imageData解码，所有这些都通过把原始图像转换成在YCbCr色度系统中的彩色信号获得。

公开(公告)号：EP1049259B1

公开(公告)日：2008-03-26

申请号：EP98901084.8

申请日：1998-01-30

申请人： Kowa Co., Ltd.

发明人： MATSUI, Kineo ,  IWAKIRI, Munetoshi

IPC分类号： G10L19/00

CPC分类号： G10L19/12 ,  G10L19/018

摘要： An encoder which encodes a voice in accordance with LD-CELP (low delay code excitation linear prediction encoding) of the ITU-T Recommendation G.728 as a method wherein, when a vibration wave such as a voice signal is encoded by vector quantization, the code is secretly combined with another data. The encoder stores dividing key data kidx by which 128 types of representative vector data (waveform codes) yj; j = 0, 1, ..., 127 are labeled with 0 or 1 in the order from the uppermost bit. If the bit to be combined is "0", the vectors are quantized by using only the waveform codes yj corresponding to the bit "0" of the dividing key data kidx as the selection objects. If the bit to be combined is "1", the vectors are quantized by using only the waveform codes yj corresponding to the bit "1" of the dividing key data kidx as the selection objects. Thus, the outputted voice code is combined with another datum bit.

公开(公告)号：EP1148708A1

公开(公告)日：2001-10-24

申请号：EP99943227.1

申请日：1999-09-08

申请人： Kowa Co., Ltd.

发明人： FUKUOKA, Yoshihide ,  MATSUI, Kineo

IPC分类号： H04N1/387 ,  G06T1/00

CPC分类号： G06T1/005 ,  G06T2201/0052 ,  G06T2201/0063

摘要： The procedure of the present invention specifies predetermined elements included in a matrix F, which is obtained by Fourier transform of master image data P0 (step S122), and adds a minute variation ΔF of a predetermined magnitude to either a real number array FR or an imaginary number array FI of the predetermined elements (step S124 and S126). Here note that the minute variation ΔF should be subtracted from corresponding elements, in order to keep the symmetry of the Fourier spectrum. A resulting image obtained by an inverse transform of the matrix with the minute variation ΔF added thereto includes a phase difference pattern W01 that is embedded therein and corresponds to the minute variation ΔF. As long as a master image is closed to the public, this embedded pattern can not be taken out of the resulting image nor be deleted by overwriting attacks. Even if the processed data with the digital watermark embedded therein is exposed to overwriting attacks of different pieces of watermark information by the similar algorithm, the arrangement of the invention enables the digital watermark embedded in the master data to be taken out accurately. The similar series of processing may be carried out with regard to a specific area of the low frequency component obtained by wavelet transform of the mater image.

摘要翻译： 本发明的过程指定预定包括在矩阵F，所有这是由傅里叶获得主图像数据P0（步骤S122）的变换元件，并且增加了一个微小变化DELTA F A预定大小的任一实数阵列FR或 在预定的元件（步骤S124和S126）的虚数阵列FI。 这里注意，DASS模具微小变化的ΔF应该从相应元素中减去，以保持傅立叶光谱的对称性。 矩阵的通过在逆而获得的结果的图像变换用微小变化的ΔF加入其中包括相差图案W01那样被嵌入其中，并且对应于微小变化DELTA F.只要主图像被向公众开放， 此嵌入图案不能被取出所得到的图像，也不通过重写攻击被删除。 即使与嵌入其中的数字水印的处理过的数据被暴露于由类似的算法重写条不同的水印信息的攻击，本发明的布置能够嵌入在主数据中的数字水印被精确取出设定。 类似的一系列处理可以对于由小波变换获得的图像母校变换的低频分量的一特定区域来进行。

公开(公告)号：EP1049259A1

公开(公告)日：2000-11-02

申请号：EP98901084.8

申请日：1998-01-30

申请人： Kowa Co., Ltd.

发明人： MATSUI, Kineo ,  IWAKIRI, Munetoshi

IPC分类号： H03M7/30 ,  H04B14/04 ,  G10L9/14 ,  G10L9/18

CPC分类号： G10L19/12 ,  G10L19/018

摘要： An encoder which encodes a voice in accordance with LD-CELP (Low-Delay Code Excited Linear Prediction) of the ITU-T Recommendation G.728 as a method wherein, when a vibration wave such as a voice signal is encoded by vector quantization, the code is secretly combined with another data. The encoder stores dividing key data k idx by which 128 types of representative vector data (waveform codes) yj; j = 0, 1, ..., 127 are labeled with 0 or 1 in the order from the uppermost bit. If the bit to be combined is "0", the vectors are quantized by using only the waveform codes yj corresponding to the bit "0" of the dividing key data k idx as the selection objects. If the bit to be combined is "1" the vectors are quantized by using only the waveform codes yj corresponding to the bit "1" of the dividing key data k idx as the selection objects. Thus, the outputted voice code is combined with another datum bit.

摘要翻译： 根据ITU-T建议G.728的LD-CELP（低延迟码激励线性预测）对语音进行编码的编码器，作为一种方法，其中当通过矢量量化对语音信号等振动波进行编码时， 该代码与另一个数据秘密结合。 编码器存储128个代表矢量数据（波形码）yj的分割关键数据kidx， j = 0,1，...，127以最高位的顺序用0或1标记。 如果要组合的位为“0”，则仅使用与分割键数据kidx的位“0”相对应的波形代码yj作为选择对象量化矢量。 如果要组合的比特是“1”，则仅使用与划分密钥数据kidx的比特“1”相对应的波形码yj作为选择对象量化矢量。 因此，输出的语音代码与另一个数据位组合。

公开(公告)号：EP1156662B1

公开(公告)日：2009-06-03

申请号：EP00900903.6

申请日：2000-01-24

申请人： Kowa Co., Ltd.

发明人： MATSUI, Kineo

IPC分类号： G06T1/00 ,  H04N1/32

CPC分类号： G06T1/005 ,  H04N1/32154 ,  H04N1/32288 ,  H04N1/32309 ,  H04N2201/3233 ,  H04N2201/327

摘要： An original image is segmented into blocks. Each block is subjected to orthogonal transformation. The coefficients of the adjacent blocks are compared with each other, and the relationship of magnitude between the coefficients is changed depending on the bit information to be embedded. The bit information to be embedded is made to include a basic pattern formed by the density pattern method and embedded repetitively in image data. As a result since the bit information is converted to a pattern and repetitively embedded, the endurance strength is improved. Further since the original image is converted and then an electronic watermark is embedded, the original image does not necessarily need to be concealed when managed.

公开(公告)号：EP1139650B1

公开(公告)日：2004-01-07

申请号：EP99949436.2

申请日：1999-10-26

申请人： Kowa Co., Ltd.

发明人： MATSUI, Kineo

IPC分类号： G06T1/00 ,  H04N1/32

CPC分类号： G06T1/005 ,  G06T1/0028 ,  G06T2201/0065 ,  H04N1/32288 ,  H04N1/32309 ,  H04N2201/3233 ,  H04N2201/327

公开(公告)号：EP1156662A1

公开(公告)日：2001-11-21

申请号：EP00900903.6

申请日：2000-01-24

申请人： Kowa Co., Ltd.

发明人： MATSUI, Kineo

IPC分类号： H04N1/387 ,  G06T1/00

CPC分类号： G06T1/005 ,  H04N1/32154 ,  H04N1/32288 ,  H04N1/32309 ,  H04N2201/3233 ,  H04N2201/327

摘要： The technique of the present invention extracts a plurality of blocks from a master image, processes image data corresponding to each block by orthogonal transform, compares coefficients of adjoining blocks with each other, and modifies the order of magnitude of the coefficients, based on bit information to be embedded. The bit information to be embedded is constructed as a basic pattern, which is generated, for example, by the density pattern method. The basic pattern is embedded in image data iteratively. The process of iteratively embedding the basic pattern enhances the resistance against various attacks. The technique of embedding a digital watermark in a transformed master image may not require labor-consuming management to keep master images closed to the public.

摘要翻译： 本发明的技术提取从主图像块的复数，处理图像数据由正交对应于每个块变换，海誓山盟比较邻接的块的系数，并修改系数的数量级，根据比特信息 被嵌入。 要被嵌入的比特信息被构造为基本模式，所有这一切都被生成，例如，通过密度模式方法。 基本图案被嵌入在图像数据中迭代。 迭代嵌入的基本图形的工艺增强抵御各种攻击的抵抗性。 变换的主图像中嵌入数字水印可能不需要费力的管理，以保持原图像的技术，向公众开放。

公开(公告)号：EP0996278A1

公开(公告)日：2000-04-26

申请号：EP98924656.6

申请日：1998-06-15

申请人： Kowa Co., Ltd.

发明人： MATSUI, Kineo ,  ONISHI, Junji

IPC分类号： H04N1/387

CPC分类号： H04N19/63 ,  G06T1/0057 ,  G06T2201/0052 ,  H04N1/32154 ,  H04N1/3216 ,  H04N1/3217 ,  H04N1/32176 ,  H04N1/32187 ,  H04N1/32288 ,  H04N21/23892 ,  H04N21/8358 ,  H04N2201/3233 ,  H04N2201/327

摘要： An electronic watermarking method which makes it difficult to erase other information such as signature data combined with image data and which does not require the original image in extracting the information. The image signal is subjected to wavelet transform to extract a multiple resolution approximated area (MRA), which is spread to a wide-band frequency spectrum in a pseudonoise sequence (PN sequence), and signature data is incorporated into a specific part of the spread frequency spectrum according to predetermined rules. Then, the combined signals are subjected to spectrum-despreading according to the PN sequence and to wavelet inverse transform so as to generate an image signal combined with the signature data. By this method, because the signature data is combined for an MRA, It is resistant to signal processings such as filtering. Further, the use of the spread spectrum provides a high level of secrecy. Moreover, the feature of the wavelet transform can be utilized to synthesize signature data in multiple layers.

摘要翻译： 一种电子水印方法，该方法使得难以擦除其他信息：诸如与数据和图像，其不需要在提取信息的原始图象组合签名数据。 图像信号进行小波变换，以提取多分辨率近似区域（MRA），所有这一切都被扩展到一个宽带频率频谱中的伪噪声序列（PN序列），和签名数据并入扩散的特定部分 频谱gemäß预定规则。 然后，将合并的信号进行频谱解扩gemäß到PN序列和小波逆变换，从而生成与签名数据相组合的图像信号。 通过这种方法，因为签名数据被组合在MRA，它是耐信号处理：如过滤。 此外，使用扩频提供保密的较高水平。 更上方，小波的特征变换可以利用在多个层中以合成签名数据。