A method for embedding and extracting watermark for multi-format electronic chart data

Abstract

This invention proposes a method for watermark embedding and extraction of multi-format electronic nautical chart data. The watermark embedding process includes: preprocessing the electronic nautical chart data, extracting visual features using a convolutional neural network, reducing the dimensionality of the visual features using principal component analysis, embedding the watermark using discrete cosine transform, and finally reconstructing the electronic nautical chart data using an inverse convolutional neural network. This invention can solve the shortcomings in anti-theft and security management under various electronic nautical chart data formats and can effectively protect the copyright and security of electronic nautical chart data. It is applicable to various electronic nautical chart data formats.

Description

Technical Field

[0001] This invention specifically relates to the field of watermarking technology for electronic nautical chart data, and more specifically to a method for embedding and extracting watermarks for multi-format electronic nautical chart data. Background Technology

[0002] As a core tool for maritime safety, nautical charts provide crucial information on ocean topography, channels, obstacles, and navigational aids, playing an irreplaceable role in ship navigation, route planning, and maritime traffic management. With the rapid development of digital technology, the widespread adoption of electronic nautical charts has also brought a series of challenges to copyright protection. Electronic nautical charts are easily copied and disseminated, and unauthorized use and piracy pose a serious threat to the intellectual property rights and commercial interests of nautical chart data.

[0003] Watermarking technology, as an effective means of protecting digital content, identifies the copyright owner or usage conditions of data by embedding invisible information in the original data, and has been widely used in multimedia fields such as images, audio, and video. However, the diversity of electronic nautical charts, including vector data, image data, document data, and tile data, as well as the specific requirements of different formats for watermarking technology, make it difficult for existing watermarking technologies to fully adapt to all electronic nautical chart data formats, and thus cannot fully meet the needs of anti-theft and security management of electronic nautical charts.

[0004] The challenge of preventing theft and ensuring the security of electronic nautical chart data lies in providing a method that does not affect data quality, is highly robust, can effectively trace back to the responsible unit, and can adapt to various electronic nautical chart data formats. This requires watermarking technology to not only ensure that the watermark is not destroyed during data transmission and processing and does not affect the accuracy of the nautical chart data, but also to be able to be embedded in nautical chart data of different formats.

[0005] Therefore, it is particularly urgent to develop a method for preventing and managing the theft of electronic nautical chart data that does not affect data quality, can adapt to various chart data formats, and effectively protects the copyright of chart data. This watermarking method needs to be highly adaptable and robust to ensure that watermark information can be effectively embedded and extracted from various electronic nautical chart data formats, thereby protecting the copyright and security of the chart data. Summary of the Invention

[0006] I. Technical Issues

[0007] The purpose of this invention is to provide a method for embedding and extracting watermarks for multi-format electronic nautical chart data, thereby addressing the need for anti-theft and security management of various electronic nautical chart data formats in existing technologies.

[0008] II. Technical Content

[0009] To solve the above-mentioned technical problems, the technical solution of the present invention is: a method for watermark embedding and extraction of multi-format electronic nautical chart data, comprising the following steps:

[0010] Step S1: Preprocess the electronic nautical charts that need to be embedded with watermarks, and take different preprocessing measures for different types of electronic nautical chart data;

[0011] Step S2: Using a pre-trained convolutional neural network (CNN) model, process the pre-processed electronic nautical chart through the forward propagation method to extract visual feature vectors;

[0012] Step S3: Apply principal component analysis to reduce the dimensionality of the extracted high-dimensional visual feature vectors, converting them into low-dimensional feature vectors of fixed length.

[0013] Step S4: Select a watermark string with copyright information from the preset database, and encode the string into a watermark bit code using a hash function;

[0014] Step S5: In the frequency domain of the feature vector, use Discrete Cosine Transform (DCT) to select several discontinuous positions in the low-frequency region and embed the watermark bit code into these positions;

[0015] Step S6: Use an inverse CNN to reconstruct the feature vectors embedded with the watermark into electronic nautical chart data, and use an inverse processing method to convert it back to the original format.

[0016] Preferably, in step S1, for vector data, rasterization technology is used to convert it into an image format; for image data, standardization and normalization processing is performed; for document data, image elements are separated and converted into raster images; and for tile data, standardization and normalization processing is performed.

[0017] Preferably, in step S2, a pre-trained CNN model is selected as the feature extractor, the electronic nautical chart data processed in step one is input into the CNN model, a deeper intermediate layer in the model is selected as the feature extraction layer, and the high-dimensional visual feature vector X output by the selected layer is obtained through forward propagation.

[0018] Preferably, in step S3, principal component analysis is used to reduce the dimensionality of the extracted feature vector X and convert it into a fixed-length feature vector. First, the feature vector X is standardized so that the mean of each feature is 0 and the variance is 1, as shown in the formula:

[0019]

[0020] Where μ is the mean of each feature and σ is the standard deviation of each feature.

[0021] Preferably, in step S3, the standardized feature vector X is calculated. std The covariance matrix ∑:

[0022]

[0023] Where n is the number of samples.

[0024] Preferably, in step S3, the covariance matrix ∑ is decomposed into eigenvalues ​​to obtain the eigenvalues ​​λ. i and the corresponding feature vector υ i Based on the magnitude of the eigenvalues, select the eigenvectors υ1, υ2, ..., υ corresponding to the k largest eigenvalues. k These eigenvectors form a new coordinate system. Finally, the eigenvector X... std Projecting into the new coordinate system, we obtain the dimension-reduced eigenvector Y, where k is the selected fixed length.

[0025] Preferably, in step S4, a pre-stored watermark string is extracted from the database. The watermark string contains key information such as the management ID, organization name, and control level of the electronic nautical chart. The watermark string is converted into binary bit code w using a hash function so that it can be embedded into the feature vector of the electronic nautical chart.

[0026] Preferably, in step S5, the feature vector Y is subjected to discrete cosine transform to obtain its frequency domain representation Y. D Manually select bits i to i+k-1 as embedding positions. Define an embedding strength factor α and embed the watermark bit code w into bits i to i+k-1:

[0027] Y′ D [i+j]=Y D [i+j]+α×w[j]

[0028] Among them, Y′ D is the frequency domain coefficient after embedding the watermark, and w[j] is the j-th bit of the watermark bit code.

[0029] Preferably, in step S5, Y′ D Perform inverse discrete cosine transform to obtain the feature vector Y′ after watermarking.

[0030] Preferably, in step S6, an inverse CNN is used to restore the feature vectors embedded with the watermark to electronic nautical chart data. The inverse CNN is a decoder model whose structure matches the encoder part of the CNN used for feature extraction in step two. The operation of each layer can be represented as follows:

[0031] Z i+1 =ReLU(Wi *Z i +b i )

[0032] Among them, Z i and b i Let represent the output feature map and bias vector of the i-th layer, respectively. * represents the convolution operation, and ReLU represents the activation function.

[0033] Preferably, in step S6, the image can be reconstructed from the feature vector Y′ after the watermark is embedded using an inverse CNN. For vector data, the image format is converted back to vector data using inverse rasterization technology; for image data, inverse normalization is performed to restore the original size and pixel values; for document data, image elements are recombined back into document format; and for tile data, inverse normalization is performed to restore the original size and pixel values.

[0034] Another aspect of the present invention provides a watermark extraction method for multi-format electronic nautical chart data. The watermark extraction process is the reverse process of the embedding process, and the specific steps are as follows:

[0035] Step S7: Preprocess the electronic nautical charts from which watermarks need to be extracted, and take different preprocessing measures for different types of electronic nautical chart data;

[0036] Step S8: Using the pre-trained CNN model, process the pre-processed electronic nautical chart through the forward propagation method to extract visual feature vectors;

[0037] Step S9: Apply principal component analysis to reduce the dimensionality of the extracted high-dimensional visual feature vectors, converting them into low-dimensional feature vectors of fixed length.

[0038] Step S10: In the frequency domain of the feature vector, use DCT to extract the watermark bit code at a selected position in the low-frequency region;

[0039] Step S11: Decode the extracted watermark bit code into a watermark string using a reverse hash function, and compare it with the original watermark string in the database.

[0040] Preferably, in step S7, for vector data with embedded watermarks, rasterization technology is used to convert it into an image format; for image data with embedded watermarks, standardization and normalization processing are performed; for document data with embedded watermarks, image elements are separated and converted into raster images; for tile data with embedded watermarks, standardization and normalization processing are performed.

[0041] Preferably, in step S8, a pre-trained CNN model is selected as the feature extractor, and the watermarked electronic nautical chart data processed in step one is input into the CNN model. This CNN model should be consistent with the CNN model in step one of the watermark embedding process. The same intermediate layer as in step two of the watermark embedding process is selected as the feature extraction layer, and the high-dimensional visual feature vector X′ output by the selected layer is obtained through forward propagation.

[0042] Preferably, in step S9, principal component analysis is used to reduce the dimensionality of the extracted feature vector X′ and convert it into a fixed-length feature vector. The processing steps are consistent with step three of the embedding process, resulting in a dimensionality-reduced feature vector Y′ containing the watermark.

[0043] Preferably, in step S10, the feature vector Y′ containing the watermark is subjected to discrete cosine transform to obtain its frequency domain representation Y′. D The watermark embedding position is located in step five of the watermark embedding process, and the watermark information w is extracted.

[0044] Preferably, in step S11, the encoding scheme from step four of the watermark embedding process is used, and a reverse hash function is used to reconstruct the original watermark string from the extracted watermark information w. The database is queried to retrieve a matching pre-stored watermark string. Key information such as the management ID, organization name, and control level of the electronic chart is obtained. This confirms the copyright ownership of the electronic chart and the responsible party.

[0045] Preferably, the electronic chart data is classified and formatted as follows: chart data in S57 or 000 format; vector data in Shp or Dwg format; image data in Tif, Jpg, or Png format; document data in Pdf, Docx, or Xlsx format; and tile data in vector tile or raster tile format.

[0046] III. Technical Effects

[0047] Compared with existing technologies, the advantages of this invention are as follows: This invention provides a watermark embedding and extraction method for multi-format electronic nautical chart data, which can effectively process and embed watermarks into different types of electronic nautical chart data, including vector data, image data, document data, and tile data. This compatibility effectively achieves copyright protection and security management of electronic nautical charts. Simultaneously, by using a CNN model to extract visual feature vectors and combining them with principal component analysis for dimensionality reduction, this invention can embed watermark information into the low-frequency region of the feature vector without affecting the visual quality of the electronic nautical chart, thus improving the watermark's concealment. During the watermark embedding process, DCT is used to select discontinuous positions in the frequency domain of the feature vector to embed the watermark bit code. This method enhances the watermark's resistance to common image processing operations, ensuring the integrity and recoverability of the watermark information. Attached image description:

[0048] Figure 1 This is a flowchart illustrating the watermark embedding process of a method for embedding and extracting watermarks from multi-format electronic nautical chart data according to the present invention.

[0049] Figure 2 This is a flowchart illustrating the watermark extraction process of a watermark embedding and extraction method for multi-format electronic nautical chart data according to the present invention.

[0050] Figure 3 This is an experimental data diagram illustrating the watermark embedding and extraction method for multi-format electronic nautical chart data according to the present invention. Specific implementation examples:

[0051] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0052] This embodiment selects electronic nautical chart data in image format as experimental data, such as... Figure 3 As shown.

[0053] (I) The watermark embedding process is as follows Figure 1 As shown, a method for watermark embedding and extraction of multi-format electronic nautical chart data is presented, with the following specific steps:

[0054] Step S1: Preprocess the electronic nautical charts that need to be embedded with watermarks, and take different preprocessing measures for different types of electronic nautical chart data;

[0055] Step S2: Using the pre-trained CNN model, process the pre-processed electronic nautical chart through the forward propagation method to extract visual feature vectors;

[0056] Step S3: Apply principal component analysis to reduce the dimensionality of the extracted high-dimensional visual feature vectors, converting them into low-dimensional feature vectors of fixed length.

[0057] Step S4: Select a watermark string with copyright information from the preset database, and encode the string into a watermark bit code using a hash function;

[0058] Step S5: In the frequency domain of the feature vector, use DCT to select several discontinuous positions in the low-frequency region and embed the watermark bit code into these positions;

[0059] Step S6: Use an inverse CNN to reconstruct the feature vectors embedded with the watermark into electronic nautical chart data, and use an inverse processing method to convert it back to the original format.

[0060] The electronic chart data is classified and formatted as follows: chart data, format S57 or 000; vector data, format Shp or Dwg; image data, format Tif, Jpg, or Png; document data, format Pdf, Docx, or Xlsx; and tile data, format vector tiles or raster tiles.

[0061] It should be noted that this embodiment uses a JPEG format electronic nautical chart with a resolution of 1024x1024 pixels. After standardization, the pixel values ​​of the image are adjusted to the range of [0,1], with a mean of 0.5 and a standard deviation of 0.2.

[0062] The VGG16 model, pre-trained on the ImageNet dataset, was used as the feature extractor. The electronic nautical chart data processed in step S1 was input into the VGG16 model. The 10th layer was selected as the feature extraction layer. After forward propagation, a vector X containing 512 features was obtained, with each feature having a dimension of 32×32.

[0063] Principal component analysis (PCA) is used to reduce the dimensionality of the extracted feature vector X and convert it into a fixed-length feature vector. First, the feature vector X is standardized so that the mean of each feature is 0 and the variance is 1, as shown in the formula:

[0064]

[0065] Where μ is the mean of each feature and σ is the standard deviation of each feature.

[0066] Then calculate the standardized eigenvector X. std The covariance matrix ∑:

[0067]

[0068] Where n is the number of samples.

[0069] Perform eigenvalue decomposition on the covariance matrix ∑ again to obtain the eigenvalues ​​λ. i and the corresponding feature vector υ i Based on the magnitude of the eigenvalues, select the eigenvectors υ1, υ2, ..., υ corresponding to the 128 largest eigenvalues. 128 These eigenvectors form a new coordinate system. Finally, the eigenvector X... std Projecting into the new coordinate system yields the dimensionality-reduced eigenvector Y. Here, 128 is the selected fixed length.

[0070] Extract a watermark string from the database containing the management ID "CHART12345", the organization name "NAVY", and the control level "SECRET". Encode this string into a 128-bit watermark bit code w using the MD5 hash function.

[0071] Performing a discrete cosine transform on the eigenvector Y yields its frequency domain representation Y. D We manually selected bits 10 to 137 as embedding positions. Defining the embedding strength factor α as 0.15, we embedded the watermark bit code w into bits 10 to 137, obtaining the frequency domain coefficients Y′ after watermark embedding. D :

[0072] Y′ D [i+j]=Y D [i+j]+α×w[j]

[0073] Where w[j] is the j-th bit of the watermark bit code.

[0074] Y′ D Perform an inverse discrete cosine transform to obtain the feature vector Y′ embedded with the watermark. Step Six: Reconstruct the image from the watermark-embedded feature vector Y′ using an inverse CNN. The inverse CNN is a decoder model whose structure matches the encoder part of the VGG16 model used for feature extraction in Step Two. The operation of each layer can be represented as follows:

[0075] Z i+1 =ReLU(W i *Z i +b i )

[0076] Among them, Z i and b i Let represent the output feature map and bias vector of the i-th layer, respectively. * represents the convolution operation, and ReLU represents the activation function.

[0077] Finally, inverse normalization is performed to restore the electronic chart data to its original size and pixel values.

[0078] (ii) The watermark extraction process is the reverse of the embedding process, such as... Figure 2 As shown, a watermark extraction method for multi-format electronic nautical chart data is described, with the following specific steps:

[0079] Step S7: Preprocess the electronic nautical charts from which watermarks need to be extracted, and take different preprocessing measures for different types of electronic nautical chart data;

[0080] Step S8: Using the pre-trained CNN model, process the pre-processed electronic nautical chart through the forward propagation method to extract visual feature vectors;

[0081] Step S9: Apply principal component analysis to reduce the dimensionality of the extracted high-dimensional visual feature vectors, converting them into low-dimensional feature vectors of fixed length.

[0082] Step S10: In the frequency domain of the feature vector, use DCT to extract the watermark bit code at a selected position in the low-frequency region;

[0083] Step S11: Decode the extracted watermark bit code into a watermark string using a reverse hash function, and compare it with the original watermark string in the database.

[0084] It should be noted that the extraction step uses JPEG format electronic nautical chart data with watermarks already embedded during the embedding process. After standardization, the pixel values ​​of the image are adjusted to the range of [0,1], with a mean of 0.5 and a standard deviation of 0.2.

[0085] The same pre-trained VGG16 model as used in the embedding process was employed, and the same 10th layer was selected as the feature extraction layer to process the standardized image. After forward propagation, a vector X′ containing the watermark and 512 features was obtained, with each feature having a dimension of 32×32.

[0086] The feature vector X′ underwent the same principal component analysis as in the embedding process, resulting in a dimensionality-reduced feature vector Y′ containing the watermark. The watermarked feature vector Y′ was then subjected to a discrete cosine transform to obtain its frequency domain representation Y′. D The watermark information w is extracted by locating the embedding positions of the 10th to 137th bits during the watermark embedding process.

[0087] The extracted watermark information w was reconstructed into the original watermark string using a reverse hash function. By querying the database, results matching the pre-stored watermark string were retrieved, confirming that the electronic chart's management ID was "CHART12345", the organization name was "NAVY", and the control level was "SECRET".

[0088] The present invention and its embodiments have been described above. This description is not restrictive. The accompanying drawings are only one embodiment of the present invention. In short, if a person skilled in the art is inspired by this description and designs a similar structure and embodiment without departing from the spirit of the present invention, such design should fall within the scope of protection of the invention.

Claims

1. A method of embedding a watermark into multi-format electronic chart data, characterized by, The method comprises the following steps: S1: preprocessing the electronic chart that needs to be embedded with a watermark, and taking different preprocessing measures for different types of electronic chart data; S2: using a pre-trained convolutional neural network model to process the preprocessed electronic chart through a forward propagation method to extract a visual feature vector; S3: applying a principal component analysis method to reduce the dimension of the extracted high-dimensional visual feature vector, and converting the high-dimensional feature vector into a fixed-length low-dimensional feature vector; S4: selecting a watermark string with copyright information from a pre-set database, and encoding the string into a watermark bit code through a hash function; S5: in the frequency domain of the feature vector, using discrete cosine transform to select several discontinuous positions in the low-frequency area, and embedding the watermark bit code into these positions; S6: using reverse CNN to reconstruct the feature vector embedded with the watermark into electronic chart data, and converting it into the original format using the reverse processing method; In step S2, a pre-trained CNN model is selected as a feature extractor, the processed electronic chart data in step S1 is input into the CNN model, a deeper intermediate layer in the model is selected as a feature extraction layer, and a high-dimensional visual feature vector X output by the selected layer is obtained through forward propagation; In step S3, the extracted feature vector X is reduced in dimension and converted into a fixed-length feature vector using the principal component analysis method. First, the feature vector X is standardized so that the mean of each feature is 0 and the variance is 1, and the formula is: Where μ is the mean of each feature, and σ is the standard deviation of each feature. In the step S3, the covariance matrix ∑ of the normalized feature vector X std is calculated. Where n is the number of samples. The step S3, eigenvalue decomposition is made to the covariance matrix ∑, to obtain eigenvalues λ i and corresponding eigenvectors υ i According to the size of the eigenvalue, the eigenvectors υ1, υ2, …, υ k corresponding to the first k largest eigenvalues are selected, which constitute a new coordinate system, and finally, the eigenvector X std is projected in the new coordinate system to obtain the reduced eigenvector Y, wherein k is the selected fixed length; In step S4, a pre-stored watermark string is extracted from the database. The watermark string contains the management ID, organization naming, and control level key information of the electronic chart, and is converted into a binary bit code w through a hash function, so as to be embedded into the feature vector of the electronic chart.

2. The method for embedding a watermark into multi-format electronic chart data according to claim 1, wherein: In step S1, for vector data, rasterization technology is used to convert it into image format, for image data, standardization and normalization processing is performed, for document data, image elements are separated and converted into raster images, and for tile data, standardization and normalization processing is performed.

3. The method for embedding a watermark into multi-format electronic chart data according to claim 1, wherein: In the step S5, the eigenvector Y is subjected to a discrete cosine transform to obtain its representation in the frequency domain Y D , manually select the i-th to the i+k-1-th bits as embedding positions, define an embedding strength factor a, and embed the watermark bit code w into the i-th to the i+k-1-th bits: Y' D [i + j] = Y D [i + j] + a x w[j] where Y' = Y + w[j] * W D is the frequency domain coefficient after embedding the watermark, and w[j] is the jth bit of the watermark bit code. In the step S5, Y′ D The inverse discrete cosine transform is performed to obtain the characteristic vector Y′ after embedding the watermark.

4. The method for embedding a watermark into multi-format electronic chart data according to claim 3, wherein: In step S6, the feature vector embedded with the watermark is restored to electronic chart data using reverse CNN. The reverse CNN is a decoder model, the structure of which matches the encoder part of the CNN used for feature extraction in step two, and the operation of each layer can be represented as: Z i+1 = ReLU(W i * Z i + b i ) wherein Z i and b i respectively represent the output feature map and bias vector of the i-th layer, * represents a convolution operation, and ReLU represents an activation function. In step S6, the image can be reconstructed from the feature vector Y' embedded with the watermark through reverse CNN. For vector data, inverse rasterization technology is used to convert the image format back to vector data, for image data, inverse standardization processing is performed to restore the original size and pixel value, for document data, image elements are recombined into document format, and for tile data, inverse standardization processing is performed to restore the original size and pixel value.

5. A watermark extraction method of performing the watermark embedding method for multi-format electronic chart data according to claim 1, characterized by, The method comprises the following steps: S7: preprocessing the electronic chart that needs to extract a watermark, and taking different preprocessing measures for different types of electronic chart data; S8: using a pre-trained convolutional neural network model, the pre-processed electronic chart is processed by a forward propagation method to extract a visual feature vector; S9: applying principal component analysis to the extracted high-dimensional visual feature vector for dimension reduction processing, converting the high-dimensional feature vector into a fixed-length low-dimensional feature vector; S10: in the frequency domain of the feature vector, using discrete cosine transform to extract watermark bit code at selected positions in the low-frequency area; S11: the extracted watermark bit code is decoded into a watermark string by an inverse hash function, and compared with the original watermark string in the database.

6. The method of claim 5, wherein the method is characterized by: In step S7, for the vector data embedded with the watermark, rasterization technology is used to convert it into image format, for the image data embedded with the watermark, standardization and normalization processing is performed, for the document data embedded with the watermark, image elements are separated and converted into raster images, for the tile data embedded with the watermark, standardization and normalization processing is performed; In step S8, a pre-trained CNN model is selected as a feature extractor, and the processed electronic chart data embedded with the watermark in step one is input into the CNN model. The CNN model should be consistent with the CNN model in step one of the watermark embedding process, and the same intermediate layer is selected as the feature extraction layer. The high-dimensional visual feature vector X' output by the selected layer is obtained by forward propagation. In step S9, principal component analysis is used to reduce the dimension of the extracted feature vector X' and convert it into a fixed-length feature vector. The processing steps are consistent with step three of the embedding process, and the reduced feature vector Y' containing the watermark is obtained. In the step S10, the feature vector Y' containing the watermark is subjected to a discrete cosine transform to obtain its representation in the frequency domain Y' D , positioning the embedding position in the step five of the watermark embedding process, and extracting the watermark information w; In step S11, the encoding scheme in step four of the watermark embedding process is used to reconstruct the extracted watermark information w into the original watermark string using the inverse hash function. The database is queried to retrieve the matching pre-stored watermark string, and the management ID, organizational naming, control level and other key information of the electronic chart are obtained to confirm the copyright ownership and accountability unit of the electronic chart.

7. The method for embedding a watermark into multi-format electronic chart data according to claim 1, wherein: The classification and format of the electronic chart data include: chart data, format S57 or 000; vector data, format Shp, Dwg; image data, format Tif, Jpg, Png; document data, format Pdf, Docx, Xlsx; tile data, format vector tile, raster tile.

Images