Image watermarking processing method and apparatus

Abstract

The application discloses an image watermark processing method and device. The method comprises the following steps: determining a center image region and at least two symmetric image regions symmetric about a center pixel point of a target image from the target image; performing image decomposition processing on a local image indicated by each image region to obtain a first component image and a second component image of the corresponding local image; embedding preset watermark information into the first component image corresponding to each local image to obtain a first component image of each local image containing watermark information; performing image reconstruction processing on the corresponding local image based on a frequency domain signal of each first component image containing watermark information and a frequency domain signal of the second component image of the corresponding local image to obtain each local image containing watermark information; updating each local image of the target image to the corresponding local image containing watermark information to obtain a target image containing watermark information, so that the target image containing watermark information has strong attack resistance.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to an image watermarking processing method and apparatus. Background Technology

[0002] With the rapid development of internet technology, everyone can now publish their creative works online. For social media platforms with large user traffic, the need for copyright protection for user-generated content is enormous. Currently, to protect users' copyrights, users can choose to add watermarks to their published works (such as images and videos) so that if their work is misused, the copyright ownership can be verified by decoding the watermark. To avoid affecting the visual appeal of the image, the added watermark usually needs to be somewhat concealed. While traditional image processing methods can ensure a certain degree of concealment when adding watermarks, it is difficult to decode the watermark from the watermarked image after adjustments (such as brightness, color, and mirroring). In other words, images watermarked using traditional methods have low resistance to attacks. Therefore, developing an image watermarking method to ensure that watermarked images have strong resistance to attacks has become an important research topic. Summary of the Invention

[0003] This application provides an image watermarking method and apparatus that can ensure that images containing watermark information have strong anti-attack capabilities.

[0004] On one hand, embodiments of this application provide an image watermarking processing method, including:

[0005] Multiple image regions are determined from a target image, including a central image region and at least two symmetrical image regions. The central pixel of the target image is located in the central image region, and the positions of the at least two symmetrical image regions in the target image are symmetrical about the central pixel.

[0006] Each local image region is decomposed to obtain a first component image and a second component image of the corresponding local image. The frequency domain signal of the first component image corresponding to any local image is the target frequency domain signal, and the frequency domain signal of the second component image corresponding to any local image is a frequency domain signal different from the target frequency domain signal. The first component image corresponding to any local image refers to the component image after the local image is decomposed to a preset frequency band.

[0007] The preset watermark information is embedded into the first component image corresponding to each local image to obtain the first component image containing the watermark information corresponding to each local image.

[0008] Based on the frequency domain signal of the first component image containing watermark information corresponding to each local image and the frequency domain signal of the second component image corresponding to each local image, image reconstruction processing is performed on each local image to obtain the local image containing watermark information corresponding to each local image.

[0009] Each local image of the target image is updated to a corresponding local image containing watermark information, thus obtaining the target image containing watermark information.

[0010] In one implementation, the method further includes:

[0011] From the plurality of image regions, determine the target image region whose color complexity is greater than or equal to the second complexity threshold;

[0012] The step of performing image decomposition processing on the local images indicated by each image region to obtain the first component image and the second component image of the corresponding local image includes:

[0013] The local image indicated by the target image region is subjected to image decomposition processing to obtain the first component image and the second component image of the corresponding local image.

[0014] Furthermore, embodiments of this application provide an image watermarking processing apparatus, comprising:

[0015] A determining unit is configured to determine multiple image regions from a target image, the multiple image regions including a central image region and at least two symmetrical image regions, wherein the central pixel of the target image is located in the central image region, and the positions of the at least two symmetrical image regions in the target image are symmetrical about the central pixel.

[0016] The image decomposition unit is used to perform image decomposition processing on the local images indicated by each image region to obtain the first component image and the second component image of the corresponding local image. The frequency domain signal of the first component image corresponding to any local image is the target frequency domain signal, and the frequency domain signal of the second component image corresponding to any local image is a frequency domain signal different from the target frequency domain signal. The first component image corresponding to any local image refers to the component image after the local image is decomposed to a preset frequency band.

[0017] The watermark embedding unit is used to embed preset watermark information into the first component image corresponding to each local image, so as to obtain the first component image containing watermark information corresponding to each local image.

[0018] The image reconstruction unit is used to perform image reconstruction processing on each local image based on the frequency domain signal of the first component image containing watermark information corresponding to each local image and the frequency domain signal of the second component image corresponding to each local image, so as to obtain the local image containing watermark information corresponding to each local image.

[0019] The update unit is used to update each local image of the target image to a corresponding local image containing watermark information, thereby obtaining a target image containing watermark information.

[0020] In another aspect, embodiments of this application provide a computer device, including:

[0021] A processor, the processor being adapted to implement one or more computer programs;

[0022] A computer storage medium storing one or more computer programs, the one or more computer programs being adapted to be loaded by the processor and executed by the above-described image watermarking processing method.

[0023] In another aspect, embodiments of this application also provide a computer storage medium storing one or more computer programs, which are adapted to be loaded by a processor and executed by the above-described image watermarking processing method.

[0024] In this embodiment, when watermark information needs to be added to a target image, multiple image regions are determined from the target image, and watermark information is embedded in the local images indicated by each image region. These multiple image regions include a central image region and at least two symmetrical image regions that are symmetrical about the central pixel of the target image. This ensures that even after being subjected to various geometric attacks, including cropping attacks, padding attacks, and sticker attacks, the target image containing watermark information still has a high probability of retaining at least one local image containing watermark information. This guarantees that the watermark information still has a high probability of being successfully parsed after being subjected to geometric attacks, thus giving the target image containing watermark information obtained by using this embodiment a strong geometric attack defense capability. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1a This is a schematic diagram of a central image region and a symmetrical image region provided in an embodiment of this application;

[0027] Figure 1b This is a schematic diagram of another central image region and symmetrical image region provided in the embodiments of this application;

[0028] Figure 2 This is a schematic diagram illustrating the watermark information embedding effect provided in an embodiment of this application;

[0029] Figure 3 This is a schematic flowchart of an image watermarking processing method provided in an embodiment of this application;

[0030] Figure 4 This is a schematic diagram of a watermark information adding process provided in an embodiment of this application;

[0031] Figure 5 This is a schematic diagram of an image parsing process provided in an embodiment of this application;

[0032] Figure 6 This is a schematic diagram of the structure of an image watermarking processing device provided in an embodiment of this application;

[0033] Figure 7 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0034] This application provides an image watermarking method applicable to various scenarios requiring image watermarking. This method adds highly concealed and defensive watermark information to a target image. The general principle of the method is described below: When watermark information needs to be added to a target image, multiple image regions are determined from the target image. These regions include a central image region and at least two symmetrical image regions symmetrical about the central pixel of the target image. Each image region indicates a corresponding local image. For any local image indicated by an image region, image decomposition processing is performed to obtain a first component image and a second component image of the local image. A component image represents an image obtained after a local image is decomposed into a certain frequency band. The frequency band indicates the frequency range of the frequency domain signal. The frequency band corresponding to the first component image is a preset frequency band. That is, the frequency domain signal obtained after frequency domain transformation of the first component image is the target frequency domain signal in the target image that belongs to the frequency range indicated by the preset frequency band. The frequency domain signal of the second component image is a different frequency domain signal from the target frequency domain signal. For example, the target frequency domain signal may include low-frequency signals, mid-low-frequency signals, or mid-high-frequency signals. Preset watermark information is embedded into the first component image of the local image to obtain a first component image containing watermark information corresponding to that local image. Image reconstruction processing is then performed based on the frequency domain signal of the first component image containing watermark information and the frequency domain signal of the second component image to obtain a local image containing the watermark information. Based on this, watermark information can be embedded in the local image indicated by each image region, thereby obtaining a local image containing watermark information corresponding to each local image. Then, the corresponding local image in the target image is updated using each local image containing watermark information to obtain the target image containing the watermark information.

[0035] As can be seen from the above description, in this embodiment, the preset watermark information will be embedded into multiple local images of the target image, and the corresponding watermark information can be obtained by extracting the watermark from any local image containing the watermark information. In specific applications, for example, the positions of multiple image regions in the target image can be as follows: Figure 1a As shown or as Figure 1b As shown. Based on Figure 1aIt's easy to understand that these multiple local images include multiple local images symmetrical about the central pixel and local images containing the central pixel. Generally, when performing various geometric attacks such as cropping, scaling, and padding attacks on images containing watermark information, in order to ensure the aesthetics of the image obtained after the attack, the attack will be performed along a certain edge or corner of the target image containing watermark information in a certain direction. Therefore, there is at least one local image with sufficiently high similarity or even complete similarity between the image after the geometric attack and the image before the geometric attack, so that the image after the geometric attack can also detect the corresponding watermark information, effectively improving the resistance of the embedded watermark information to geometric attacks.

[0036] Furthermore, since the target image is decomposed into a first component image and a second component image, and the frequency domain signal of the first component image is the target frequency domain signal while the frequency domain signal of the second component image is another frequency domain signal, and the preset watermark information is embedded in the first component image of the local image, the embedding of the preset watermark information is only related to the target frequency domain signal. The frequency of the target frequency domain signal is within the frequency range indicated by the preset frequency band. In other words, in this embodiment, the watermark information is added to the target image from a frequency domain perspective. Typically, adding watermark information to the target image from a frequency domain perspective has a small impact on the corresponding pixels in the target image, and therefore does not affect the display effect of the target image (e.g., image clarity, image color). This ensures that the target image containing the watermark information has a good visual effect, guaranteeing the image quality of the target image containing the watermark information while achieving copyright protection, thereby improving user stickiness and user experience of related application products. In practical applications, the watermark information added in this way can be called a blind watermark, and the effect of blind watermark embedding can be as follows: Figure 2 As shown.

[0037] In one implementation, watermark information is embedded in a local portion of the target image. This results in only a small number of pixels in the target image containing the watermark having changed pixel values ​​compared to the original target image. This effectively reduces the impact of the watermark information on the target image, thus minimizing the impact of the watermark embedding on the image quality. In practical applications, to ensure good concealment of the embedded watermark information while minimizing its impact on the image quality, the computer device can refer to color complexity to determine the appropriate watermarking method. For example, color complexity can be determined by calculating the variance of each pixel value, or by other methods, such as calculating the average variance of the component values ​​of each pixel in each color channel. This application does not impose any limitations on this method. Color complexity effectively reflects the variation range of pixel values ​​in the target image. The greater the variation range, the higher the color complexity, and the easier it is to conceal the watermark information; conversely, the smaller the variation range, the more difficult it is to conceal. Specifically, for target images with high color complexity (e.g., greater than or equal to a first complexity threshold), the pixel values ​​of each pixel vary significantly. In this case, embedding a watermark in a local image, while altering the differences between pixel values ​​in that local area and those in other areas, is difficult to detect due to the high color complexity. Conversely, when the color complexity of the target image is low (e.g., less than the first complexity threshold), the pixel values ​​of each pixel vary less. Embedding watermark information into a local area of ​​the target image in this case can easily amplify the differences between pixel values ​​in that local area and those in other areas, thus reducing the concealment of the watermark information. Therefore, for target images with a color complexity greater than or equal to the first complexity threshold (e.g., 45), the computer device identifies multiple image regions from the target image to embed the preset watermark information into the local areas indicated by these regions. This ensures both a small embedding range and good concealment of the embedded watermark information. For target images with color complexity less than or equal to a first complexity threshold, the computer device can directly embed the watermark information globally within the target image. Specifically, the computer device can perform image decomposition processing on the target image to obtain a first component image and a second component image, and embed the preset watermark information into the first component image. Based on the frequency domain signals of the first component image containing the watermark information and the second component image, the computer device can perform image reconstruction processing on the target image to obtain the target image containing the watermark information.

[0038] In one implementation, the image watermarking method proposed in this application can be executed by a computer device, which may include any one or more of a terminal device and a server. The terminal device may include, but is not limited to, mobile phones, computers, in-vehicle terminals, and smartwatches, and can run corresponding image processing applications and other types of applications. The server may include, but is not limited to, independent physical servers, server clusters or distributed systems composed of multiple physical servers, and cloud servers that provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, and cloud storage.

[0039] Please see Figure 3 , Figure 3 This is a schematic flowchart illustrating an image watermarking method provided in an embodiment of this application. The method can be executed by the aforementioned computer device, and as... Figure 3 As shown, the method includes at least steps S301-S305:

[0040] S301. Determine multiple image regions from the target image. The multiple image regions include a central image region and at least two symmetrical image regions. The central pixel of the target image is located in the central image region, and the positions of the at least two symmetrical image regions in the target image are symmetrical about the central pixel.

[0041] In a specific embodiment, upon receiving an image watermarking processing request, the computer device can trigger the determination of multiple image regions from a target image. The target image can be an image for which watermarking embedding is required, carried in the image watermarking processing request, or it can be obtained by the computer device through other means, such as obtaining it from an image set in response to corresponding computer instructions, or obtaining it from video frames contained in a video file. In practical applications, there can be one or multiple target images. When there are multiple target images, the computer device can separately confirm multiple image regions for each target image.

[0042] When a computer device determines multiple image regions from a target image, various feasible methods can be employed. The primary principle for any implementation method is that: among the multiple image regions, there exists a central image region and at least two symmetrical image regions whose locations are symmetrical about the central pixel; and the central pixel of the target image must be located within the central image region. The following continues in conjunction with... Figure 1a And specific examples will be provided for illustration. Figure 1a Midpoint A is the center pixel of the target image, specifically the intersection of the midline of the long side and the midline of the wide side of the target image. In practical applications, center pixel A can overlap with the center of the central image region (e.g., ...). Figure 1aIt can also be that the region center of the image does not overlap with the region center of the central image area (e.g., Figure 1b This application does not limit the scope of the embodiments. The region location can refer to the coordinates of the center of a symmetrical image region, or it can refer to the coordinates of individual pixels within the symmetrical image region. Taking the region location as the coordinates of the region center, and the coordinates of the center pixel as (0, 0), symmetry about the center pixel means that the coordinates of the centers of the two image regions are symmetrical about the coordinate point (0, 0). For example, Figure 1a If the coordinates of the center of image region 101 are (i, j), then the coordinates of the center of image region 102, which is symmetrical about the center pixel, are (-i, -j).

[0043] In one implementation, the computer device determines multiple image regions from a target image as follows: First, the computer device obtains the coordinates of the center pixel of the target image. Based on the obtained coordinates, it determines a central image region in the target image containing these coordinates and with a region size of a first reference size. Then, based on the coordinates of the center pixel, it further determines at least two symmetrical image regions in other image regions of the target image, excluding the central image region, with a region size of a second reference size and a region position symmetrical about the coordinates of the center pixel. The first reference size can be a preset size, which may be carried in the image watermarking processing request or be built into the computer device. Alternatively, the first reference size can be a size determined by the computer device based on preset watermark information, which may include, but is not limited to, one or two of the following: watermark image and text information. Typically, watermark embedding is based on two images. That is, if the preset watermark information includes text information, the preset watermark information can be converted into a corresponding watermark image before watermark embedding. In this case, the computer device can use the image size of the watermark image as the first reference size, or, referring to the image size of the watermark image and the image size of the target image, determine an intermediate size that is slightly larger than the image size of the watermark image but smaller than the image size of the target image as the first reference size. This ensures that the complete watermark image can be embedded in the local image indicated by the central image area, thereby achieving the purpose of copyright protection.

[0044] Correspondingly, the second reference size can also be a preset size, or the same size as the first reference size. In this case, the computer device can directly use the acquired preset size or the first reference size as the second reference size. Optionally, the second reference size can also be different from the first reference size. Specifically, the second reference size can be determined after the central image region is determined in the following manner: The computer device first acquires the reference number (greater than 2) of symmetrical image regions, and determines the region size of each symmetrical image region in the reference number of symmetrical image regions based on the other image regions in the target image besides the central image region and the reference number, and then uses the determined region size as the second reference size. At this time, the region size of each symmetrical image region is the same, and the way the computer device determines the region size of each symmetrical image region can refer to the following steps (1)-(3):

[0045] (1) Obtain the side lengths of all image regions other than the central image region in the target image. Each side length refers to the length of each boundary line used to constitute the other image regions. The lengths of different boundary lines can be different, depending on the shape of the symmetrical image region. However, this application embodiment does not restrict the shape of any image region. That is, in this application embodiment, the shape of the image region can include: polygons (squares, triangles, rectangles), circles, and irregular shapes, which can enrich the fun of image watermarking to a certain extent and thus increase user stickiness. (2) Determine the side lengths of each symmetrical image region based on the minimum side length and reference number of the side lengths of the other image regions in the target image. Optionally, the maximum side length of each symmetrical image region can be less than the minimum side length of the other image regions. (3) Determine the region size of each symmetrical image region, which includes the side lengths of the corresponding symmetrical image region.

[0046] S302. Perform image decomposition processing on the local images indicated by each image region to obtain the first component image and the second component image of the corresponding local image. The first component image corresponding to any local image refers to the component image after decomposing any local image to a preset frequency band.

[0047] In a specific embodiment, image decomposition processing refers to decomposing an image into component images in different frequency bands. These component images are composed of components in different frequency domains. For example, the first component image is composed of low-frequency components, while the second component image is composed of frequency domain components other than low-frequency components. The second component image can include multiple components, such as component images composed of mid-to-low frequency components, mid-to-high frequency components, and high-frequency components. The preset frequency band indicates the frequency domain in which the watermark information needs to be embedded. The frequency domain signal of the first component image corresponding to any local image is the target frequency domain signal, and the frequency domain signal of the second component image corresponding to any local image is a frequency domain signal different from the target frequency domain signal. In other words, the preset frequency band indicates the frequency range of the target frequency domain signal. In one implementation, the preset frequency bands corresponding to different local images can be different, allowing the computer device to embed watermark information in different local images based on different frequency domains, thereby embedding watermark information into multiple frequency domain signals in the same target image.

[0048] In practical applications, to ensure the concealment of watermark information, it is required that the watermark information reside in unimportant visual information of the image. This visual information is usually determined by the high-frequency components of the image (i.e., high-frequency frequency domain signals). The robustness of the watermark information requires that it be relatively intact after various image attacks. This necessitates that the pixels with added watermark information in the image have a high probability of being retained. Generally, after compressing any image, the original high-frequency components are removed as redundant information, leaving only low-frequency components in the compressed image. In other words, the main information of the image is concentrated in the low-frequency components. Therefore, embedding the watermark information into the low-frequency components (i.e., low-frequency frequency domain signals) can ensure the robustness of the watermark information. Thus, in this embodiment, embedding the watermark information in multiple frequency domains of the target image can balance the robustness and concealment of the watermark information. Furthermore, frequency domain signals in different frequency domains have varying sensitivities to changes in pixel values. Therefore, embedding watermark information in multiple frequency domains ensures that even when pixel values ​​in an image subjected to non-geometric attacks change significantly, the target image containing watermark information can still recover a less sensitive frequency domain signal from a frequency domain with lower sensitivity to pixel value changes. This guarantees the probability of detecting watermark information based on the recovered frequency domain signal, thereby ensuring the likelihood of watermark information detection in the target image and effectively improving the watermark information's resistance to non-geometric attacks. For example, non-geometric attacks can include one or more of the following: image compression, brightness attacks, contrast attacks, saturation attacks, filter attacks, beautification attacks, and image quality enhancement attacks.

[0049] Since different local images may correspond to different preset frequency bands, it can be understood that the signal type of the target frequency domain signal of the first component image of different local images may be different.

[0050] In one implementation, the signal type of the target frequency domain signal of the first component image of any local image can be associated with the image region indicating that local image. Specifically, for any image region, the computer device can obtain the frequency domain signal type associated with the region type of the image region. The region type is used to indicate whether the image region is a central image region or a symmetrical image region. In this case, the target frequency domain signals of the first component images corresponding to image regions of the same region type are all within the same frequency band. For example, the target frequency domain signal of the first component image of the central image region can be a low-frequency type, and the target frequency domain signal of the first component image of the symmetrical image region can be a mid-low frequency type or a mid-high frequency type, and vice versa. This application embodiment does not impose any limitations on this. It should be noted that the frequency of the low-frequency type frequency domain signal is lower than the frequency of the mid-low frequency type frequency domain signal, and the frequency of the mid-low frequency type frequency domain signal is lower than the frequency of the mid-high frequency type frequency domain signal. However, there is no limitation on the frequency range of each specific type, and it can be set according to the actual application scenario.

[0051] In another implementation, the signal type of the target frequency domain signal of the first component image of any local image can be determined as follows: The computer device first obtains a reference complexity and the color complexity of the local image. When the color complexity of the local image is greater than the reference complexity, a low-to-mid frequency type or a mid-to-high frequency type is used as the signal type of the target frequency domain signal. When the color complexity of the local image is less than or equal to the reference complexity, a low-frequency type is used as the signal type of the target frequency domain signal. The reference complexity can be a preset complexity or it can be determined by the computer device based on the color complexity of each local image. For example, the computer device can use the average color complexity of each local image as the preset complexity value.

[0052] In one implementation, to reduce the workload of the computer device while ensuring the concealment and robustness of the watermark information, the computer device can embed watermark information into local images indicated by partial image regions in multiple image regions. These partial image regions can be image regions whose color complexity is greater than a second complexity threshold (e.g., 15) among the multiple image regions. In this case, a feasible implementation principle of this application embodiment can be as follows: Figure 4 As shown. Based on Figure 4As can be seen, in this embodiment, for target images with color complexity less than the first complexity threshold, steps S401 to S406 and S413 are executed to achieve global embedding of watermark information, thereby ensuring the concealment of watermark information. For target images with color complexity greater than or equal to the first complexity threshold, steps S401 to S403 and S407 to S413 are executed to achieve the division of multiple image regions and to calculate the color complexity of each segment. Watermark information is embedded only when the color complexity of a segment is greater than the second complexity threshold. This not only ensures the concealment of watermark information but also reduces the computational load of computer equipment, thereby improving the efficiency of image watermark processing to a certain extent.

[0053] In a specific implementation of this application embodiment, the first component image of a local image of any image region can be generated as follows: A frequency domain transform is performed on the local image indicated by the image region to obtain a target frequency domain signal of the same signal type. Then, an inverse frequency domain transform is performed on the target frequency domain signal to obtain the first component image of the local image indicated by the image region. Here, frequency domain transform refers to converting an image from a time-domain signal to a frequency-domain signal. The frequency domain transform of the local image can be achieved by performing at least one discrete wavelet transform on the local image using a discrete wavelet transform function (such as the dwt2 function), or by using other frequency domain transform functions. For example, in this embodiment of the application, after performing frequency domain transformation on the local image using the discrete wavelet transform function, the frequency domain signal of the local image in four frequency bands can be obtained, namely the low frequency band, the mid-low frequency band, the mid-high frequency band, and the high frequency band. Thus, after image decomposition of the local image in this embodiment of the application, four component images can be obtained. These four component images correspond to the low frequency type frequency domain signal (hereinafter referred to as low frequency signal), the mid-low frequency type frequency domain signal (mid-low frequency signal), the mid-low frequency type frequency domain signal (mid-high frequency signal), and the high frequency type frequency domain signal (high frequency signal), respectively.

[0054] S303. Embed the preset watermark information into the first component image corresponding to each local image to obtain the first component image containing the watermark information corresponding to each local image.

[0055] In specific embodiments, the preset watermark information is typically a meaningful watermark image. To ensure the effectiveness of watermark embedding, in this embodiment, the image size of the watermark image can be determined after the image size of the target image is determined. In one implementation, the length of both the long and short sides of the watermark image cannot exceed 1 / 8 of the minimum side length of the target image. Simultaneously, to ensure a fast embedding speed for the computer device, the watermark image can be converted to a black-and-white image before embedding. Furthermore, to ensure that the watermark image can be easily recognized after parsing, the size of the watermark image is preferably set to be greater than 50px * 50px. Of course, this embodiment does not impose limitations on this, and the specific size can be determined according to the application scenario.

[0056] In one implementation, the watermark image is embedded into the first component image as follows: The computer device unfolds the watermark image into a one-dimensional vector and binarizes it to obtain a one-dimensional 0-1 vector (i.e., a vector where each element has a value of 0 or 1). The threshold used for binarization can be any value between 0 and 255. For pixel values ​​greater than the threshold, the corresponding element in the one-dimensional 0-1 vector is set to 1; otherwise, it is set to 0. Furthermore, the computer device can divide the first component image into blocks, obtaining N (positive integers greater than 1) sub-images of size block-size. For example, the block-size can be 5px * 5px. The i-th sub-image in the N sub-images corresponds to the i-th pixel value in the one-dimensional 0-1 vector. For each sub-image, discrete cosine transform and singular value decomposition are performed to obtain three feature matrices U, S, and V, where S is the singular value matrix. A pre-defined algorithm is used to calculate new singular values ​​based on the first singular value in S and the pixel values ​​of the corresponding pixels in the watermark image. These new singular values ​​are then used to update the first singular value in S. Inverse singular value decomposition and inverse discrete cosine transform are then performed on the updated singular value matrix, U feature matrix, and V feature matrix to obtain the sub-image embedding the watermark information. For example, the pre-defined algorithm can be shown in Equation 1:

[0057] S1[0]=(S[0] / d1+1 / 2*watermark_bit)*d1 Formula 1

[0058] Where S1[0] represents the new singular value, S[0] represents the first value in the singular value matrix S, and d1 is the intensity parameter, for example, d1 = 30. watermark_bit represents the element value in a one-dimensional 0-1 vector corresponding to the current sub-image. It can be understood that after watermarking each sub-image in the above manner, merging the sub-images containing watermark information can obtain the first component image of the local image after watermarking.

[0059] S304. Based on the frequency domain signal of the first component image containing watermark information corresponding to each local image and the frequency domain signal of the second component image corresponding to each local image, perform image reconstruction processing on each local image to obtain the local image containing watermark information corresponding to each local image.

[0060] In this embodiment, since the first component image is composed of frequency domain signals within a preset frequency band, in order to obtain a local image in the time domain, image reconstruction processing (essentially inverse frequency domain transform) needs to be performed based on the frequency domain signals of the first component image containing watermark information and the second component image to obtain a local image containing watermark information. The image reconstruction processing can be implemented using the idwt2 function (i.e., inverse discrete wavelet transform), which will not be detailed in this embodiment.

[0061] S305. Update each local image of the target image to the corresponding local image containing watermark information to obtain the target image containing watermark information.

[0062] In a specific embodiment, for adopting Figure 3 The watermark information embedded in the method shown is further addressed in this application embodiment by providing an image parsing method to increase the probability of successfully parsing the watermark information and, to a certain extent, enhance the practical value of this application embodiment. This image parsing method specifies that when parsing a reference image containing watermark information, image verification is first performed on the reference image. Image verification may include… Figure 5 The image is processed by one or more methods indicated by 501, which then performs watermark parsing on the candidate parsed image obtained after image verification to obtain the corresponding watermark information. The reference image can be the target image containing the watermark information, or it can be an image obtained after performing image attack processing on the target image containing the watermark information. The so-called image attack processing can include one or more of mirror attacks and non-geometric attacks. For example, a mirror attack is image mirroring processing, and a non-geometric attack can be understood as chroma adjustment, saturation adjustment, and / or brightness adjustment, etc.

[0063] In this context, in one implementation, when performing watermark parsing on a target image containing watermark information (as a reference image), the computer device can directly parse the reference image to obtain the watermark information. However, when the reference image is an image that has undergone image attack processing, the watermark parsing method can be as follows: perform an image inverse attack on the reference image to obtain candidate parsing images, and then perform watermark extraction processing on the candidate parsing images to obtain the watermark information contained in the reference image. The so-called inverse attack can be understood as: attacking the image in the opposite direction to the image attack processing, with the aim of restoring the un-attacked image as much as possible, thereby increasing the probability of successful watermark parsing. For example, assuming the image attack processing is brightness increase processing, then the image inverse attack processing is brightness decrease processing. More specifically, if the brightness is increased by 50%, then the brightness should be reduced to the brightness before the increase. In a specific implementation, optionally, after obtaining the reference image from which watermark information needs to be parsed, image detection can be performed on the reference image to determine whether it is an image after image attack, thereby selecting an appropriate watermark parsing method for watermark parsing. Alternatively, in other implementations, to reduce the workload and computational complexity of the computer equipment, image detection can be omitted, and image verification can be performed directly on the reference image to perform watermark parsing on the candidate parsing image obtained after verification. That is, regardless of whether the reference image is the target image containing watermark information or the image obtained after image attack processing on the target image containing watermark information, image verification is performed on it.

[0064] Image attacks include image color attacks, and image inverse attacks can include image color inverse attacks. Image color attacks refer to the process of changing pixel values ​​by altering the colors of an image. In this case, the method for performing an image color inverse attack on a reference image can be as follows: A computer device obtains color adjustment coefficients; for each pixel in the reference image, the result of multiplying the color adjustment coefficient by the color component value of each pixel is used as the target color component value for each pixel; the color component values ​​of each pixel in the reference image are updated to the corresponding target color component values ​​to obtain a candidate parsed image. For example, assuming the image color attack is chroma enhancement, the color adjustment coefficient mainly acts on the chroma component of the reference image, i.e., the color component value is the chroma value. In practical applications, the color adjustment coefficient can include multiple values, such as 0.7, 0.8, 0.9, 1, 1.1, 1.2, and 1.3. This application embodiment does not impose specific limitations on the color adjustment coefficient. It should be noted that image color attacks can also include luminance attacks, in which case the color component value is the luminance value.

[0065] In another implementation, the image attack also includes an image mirroring attack, and the image inverse attack further includes an image mirroring inverse attack. The method of obtaining a candidate parsed image by performing an image inverse attack on the reference image can also be as follows: obtain a reference mirroring direction, and mirror the reference image according to the reference mirroring direction to obtain a candidate parsed image. The reference mirroring direction can include at least one or more of the horizontal and vertical directions. In other embodiments, the reference mirroring direction can also include a diagonal direction; this application does not impose such limitations.

[0066] In this embodiment, when watermark information needs to be added to a target image, multiple image regions are determined from the target image, and watermark information is embedded in the local images indicated by each image region. These multiple image regions include a central image region and at least two symmetrical image regions that are symmetrical about the central pixel of the target image. This ensures that even after being subjected to various geometric attacks, including cropping attacks, padding attacks, and sticker attacks, the target image containing watermark information still has a high probability of retaining at least one local image containing watermark information. This guarantees that the watermark information still has a high probability of being successfully parsed after being subjected to geometric attacks, thus giving the target image containing watermark information obtained by using this embodiment a strong geometric attack defense capability. Furthermore, when embedding watermark information in a local image, the embedding is performed on the first component image of that local image. The frequency domain signal of the first component image is the target frequency domain signal indicated by a preset frequency band. The preset frequency bands corresponding to the first component images of different local images can be different, allowing watermark information to be embedded in multiple frequency domains within the same target image, such as low-frequency domain, mid-low-frequency domain, and mid-high-frequency domain. Embedding watermark information in the low-frequency domain ensures the robustness of the watermark information, while embedding it in the mid-low-frequency domain or mid-high-frequency domain ensures its concealment. Therefore, by using the embodiments of this application to embed watermark information in multiple frequency domains, the watermark information in the target image can be guaranteed to have both high concealment and robustness.

[0067] Based on the aforementioned embodiments of the image watermarking processing method, this application also discloses an image watermarking processing apparatus. This apparatus can be a computer program (including program code) running on the aforementioned computer device. The image watermarking processing apparatus can execute... Figure 2 as well as Figure 4 For the image watermarking method shown, please refer to [link / reference]. Figure 6 The image watermarking processing device may include at least: a determining unit 601, an image decomposition unit 602, a watermark embedding unit 603, an image reconstruction unit 604, and an updating unit 605.

[0068] The determining unit 601 is configured to determine multiple image regions from a target image, the multiple image regions including a central image region and at least two symmetrical image regions, wherein the central pixel of the target image is located in the central image region, and the positions of the at least two symmetrical image regions in the target image are symmetrical about the central pixel.

[0069] The image decomposition unit 602 is used to perform image decomposition processing on the local images indicated by each image region to obtain the first component image and the second component image of the corresponding local image. The frequency domain signal of the first component image corresponding to any local image is the target frequency domain signal, and the frequency domain signal of the second component image corresponding to any local image is a frequency domain signal different from the target frequency domain signal. The first component image corresponding to any local image refers to the component image after the local image is decomposed to a preset frequency band.

[0070] The watermark embedding unit 603 is used to embed preset watermark information into the first component image corresponding to each local image, so as to obtain the first component image containing watermark information corresponding to each local image.

[0071] The image reconstruction unit 604 is used to perform image reconstruction processing on each local image based on the frequency domain signal of the first component image containing watermark information corresponding to each local image and the frequency domain signal of the second component image corresponding to each local image, so as to obtain the local image containing watermark information corresponding to each local image.

[0072] The update unit 605 is used to update each local image of the target image to a corresponding local image containing watermark information, thereby obtaining a target image containing watermark information.

[0073] In one implementation, the determining unit 601 may specifically be used to perform:

[0074] Obtain the coordinates of the center pixel of the target image within the target image;

[0075] In the target image, a central image region with a region size of a first reference size is determined, and the coordinates of the central pixel point are located within the region coordinates of the central image region;

[0076] Based on the coordinates of the center pixel, at least two symmetrical image regions with a second reference size are determined in other image regions of the target image besides the center image region.

[0077] In another embodiment, the determining unit 601 can be used to determine a second reference dimension, and the determining unit 601 can specifically perform the following:

[0078] Obtain the reference number of symmetrical image regions;

[0079] Based on the other image regions and the reference number, determine the region size of each symmetrical image region in the reference number of symmetrical image regions, wherein the reference number is at least two;

[0080] The determined area size is used as the second reference size.

[0081] In another embodiment, the determining unit 601 may further perform the following specific actions:

[0082] Obtain the side lengths of the other image regions;

[0083] Based on the minimum side length among the side lengths of the other image regions and the reference quantity, the side lengths of each symmetrical image region are determined, wherein the maximum side length among the side lengths of each symmetrical image region is less than the minimum side length corresponding to the other image regions.

[0084] Determine the region size of each symmetrical image region, wherein the region size includes the side lengths of the respective symmetrical image region.

[0085] In another embodiment, the determining unit 601 can also be used to perform:

[0086] When the color complexity of the target image is less than a first complexity threshold, the target image is decomposed to obtain a first component image and a second component image of the target image.

[0087] The preset watermark information is embedded into the first component image of the target image. Based on the frequency domain signal of the first component image of the target image containing the watermark information and the frequency domain signal of the second component image of the target image, the target image is reconstructed to obtain the target image containing the watermark information.

[0088] When the color complexity of the target image is greater than or equal to the first complexity threshold, the process of determining multiple image regions from the target image is triggered.

[0089] In yet another embodiment, the image decomposition unit 602 can be used to perform:

[0090] For any image region, obtain the frequency domain signal type associated with the region type of the image region, wherein the region type is used to indicate whether the image region is a central image region or a symmetrical image region;

[0091] Perform frequency domain transformation processing on the local image indicated by any of the image regions to obtain a target frequency domain signal with the signal type of the frequency domain signal type;

[0092] The target frequency domain signal is subjected to inverse frequency domain transformation to obtain the first component image corresponding to any image region.

[0093] In another embodiment, the image watermarking processing apparatus further includes a watermark parsing unit 606, which can be used to perform:

[0094] Obtain a reference image, wherein the reference image refers to an image obtained by performing image attack processing on the target image containing watermark information;

[0095] The reference image is subjected to an image inverse attack to obtain a candidate parsed image;

[0096] The candidate parsed image is subjected to watermark extraction processing to obtain the watermark information contained in the reference image.

[0097] In another embodiment, when the image attack includes an image color attack, and the image inverse attack includes an image color inverse attack, the watermark parsing unit 606 can also be used to perform:

[0098] Obtain the color adjustment coefficient;

[0099] For each pixel in the reference image, the result of multiplying the color adjustment coefficient with the color component value of each pixel is used as the target color component value of each pixel;

[0100] The color component values ​​of each pixel in the reference image are updated to the corresponding target color component values ​​to obtain the candidate parsing image.

[0101] In another embodiment, the image attack includes an image mirroring attack, and the image inverse attack includes an image mirroring inverse attack; the watermark parsing unit 606 can also be used to perform:

[0102] Obtain the reference mirror orientation;

[0103] The reference image is mirrored according to the reference mirror direction to obtain the candidate parsing image.

[0104] In another embodiment, the determining unit 601 can also be used to perform:

[0105] From the plurality of image regions, determine the target image region whose color complexity is greater than or equal to the second complexity threshold;

[0106] When the image decomposition unit 602 performs image decomposition processing on the local images indicated by each image region to obtain the first component image and the second component image of the corresponding local image, it is specifically used to perform:

[0107] The local image indicated by the target image region is subjected to image decomposition processing to obtain the first component image and the second component image of the corresponding local image.

[0108] According to another embodiment of this application, Figure 6 The units in the illustrated image watermarking processing apparatus are divided based on logical functions. These units can be individually or entirely merged into one or more other units, or some of the units can be further divided into multiple functionally smaller units. This achieves the same operation without affecting the technical effects of the embodiments of this application. In other embodiments of this application, the aforementioned image watermarking processing apparatus may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and multiple units can collaborate to achieve them.

[0109] According to another embodiment of this application, the following can be achieved by running on a general-purpose computing device, such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM), a device capable of performing operations such as... Figure 3 The computer program (including program code) involved in each step of the method shown is used to construct, for example... Figure 6 The image watermarking processing apparatus shown herein, and the image watermarking processing method for implementing the embodiments of this application, are described. The computer program may be recorded on, for example, a computer storage medium, loaded onto the aforementioned computer device via the computer storage medium, and run therein.

[0110] In this embodiment, when watermark information needs to be added to a target image, the image watermark processing device determines multiple image regions from the target image and embeds watermark information into the local images indicated by each image region. These multiple image regions include a central image region and at least two symmetrical image regions symmetrical about the central pixel of the target image. This ensures that even after various geometric attacks, including cropping attacks, padding attacks, and sticker attacks, the target image containing watermark information still has a high probability of retaining at least one local image containing watermark information. This guarantees that the watermark information still has a high probability of being successfully parsed after being subjected to geometric attacks, thus giving the target image containing watermark information obtained using this embodiment a strong geometric attack defense capability. Furthermore, when embedding watermark information in a local image, the embedding is performed on the first component image of that local image. The frequency domain signal of the first component image is the target frequency domain signal indicated by a preset frequency band. The preset frequency bands corresponding to the first component images of different local images can be different, allowing watermark information to be embedded in multiple frequency domains within the same target image, such as low-frequency domain, mid-low-frequency domain, and mid-high-frequency domain. Embedding watermark information in the low-frequency domain ensures the robustness of the watermark information, while embedding it in the mid-low-frequency domain or mid-high-frequency domain ensures its concealment. Therefore, by using the embodiments of this application to embed watermark information in multiple frequency domains, the watermark information in the target image can be guaranteed to have both high concealment and robustness.

[0111] Based on the descriptions of the above method and device embodiments, this application also provides a computer device. Please refer to [link to relevant documentation]. Figure 7 The computer device includes at least a processor 701 and a computer storage medium 702, and the processor 701 and the computer storage medium 702 of the computer device can be connected by a bus or other means.

[0112] The aforementioned computer storage medium 702 is a memory device in a computer device used to store programs and data. It is understood that the computer storage medium 702 can include both the built-in storage medium of the computer device and extended storage media supported by the computer device. The computer storage medium 702 provides storage space for storing the operating system of the computer device. Furthermore, this storage space also stores one or more computer programs suitable for loading and execution by the processor 701. These computer programs can be one or more program codes. It should be noted that the computer storage medium can be high-speed RAM or non-volatile memory, such as at least one disk storage device; optionally, it can also be at least one storage medium located remotely from the aforementioned processor. The processor 701 (or CPU (Central Processing Unit)) is the computing and control core of the computer device, suitable for implementing one or more computer programs, specifically suitable for loading and executing one or more computer programs to achieve corresponding method flows or corresponding functions.

[0113] In one embodiment, processor 701 may load and execute one or more computer programs stored in computer storage medium 702 to achieve the aforementioned related functions. Figure 3 The corresponding method steps in the illustrated method embodiment; in specific implementation, one or more computer programs in the computer storage medium 702 are loaded by the processor 701 and executed as follows:

[0114] Multiple image regions are determined from a target image, including a central image region and at least two symmetrical image regions. The central pixel of the target image is located in the central image region, and the positions of the at least two symmetrical image regions in the target image are symmetrical about the central pixel.

[0115] Each local image region is decomposed to obtain a first component image and a second component image of the corresponding local image. The frequency domain signal of the first component image corresponding to any local image is the target frequency domain signal, and the frequency domain signal of the second component image corresponding to any local image is a frequency domain signal different from the target frequency domain signal. The first component image corresponding to any local image refers to the component image after the local image is decomposed to a preset frequency band.

[0116] The preset watermark information is embedded into the first component image corresponding to each local image to obtain the first component image containing the watermark information corresponding to each local image.

[0117] Based on the frequency domain signal of the first component image containing watermark information corresponding to each local image and the frequency domain signal of the second component image corresponding to each local image, image reconstruction processing is performed on each local image to obtain the local image containing watermark information corresponding to each local image.

[0118] Each local image of the target image is updated to a corresponding local image containing watermark information, thus obtaining the target image containing watermark information.

[0119] In one implementation, the processor 701 may specifically be used to load and execute:

[0120] Obtain the coordinates of the center pixel of the target image within the target image;

[0121] In the target image, a central image region with a region size of a first reference size is determined, and the coordinates of the central pixel point are located within the region coordinates of the central image region;

[0122] Based on the coordinates of the center pixel, at least two symmetrical image regions with a second reference size are determined in other image regions of the target image besides the center image region.

[0123] In another embodiment, the processor 701 can be used to determine a second reference size, and the processor 701 can specifically be used to load and execute:

[0124] Obtain the reference number of symmetrical image regions;

[0125] Based on the other image regions and the reference number, determine the region size of each symmetrical image region in the reference number of symmetrical image regions, wherein the reference number is at least two;

[0126] The determined area size is used as the second reference size.

[0127] In yet another embodiment, the processor 701 may specifically be used to load and execute:

[0128] Obtain the side lengths of the other image regions;

[0129] Based on the minimum side length among the side lengths of the other image regions and the reference quantity, the side lengths of each symmetrical image region are determined, wherein the maximum side length among the side lengths of each symmetrical image region is less than the minimum side length corresponding to the other image regions.

[0130] Determine the region size of each symmetrical image region, wherein the region size includes the side lengths of the respective symmetrical image region.

[0131] In yet another embodiment, the processor 701 may specifically be used to load and execute:

[0132] When the color complexity of the target image is less than a first complexity threshold, the target image is decomposed to obtain a first component image and a second component image of the target image.

[0133] The preset watermark information is embedded into the first component image of the target image. Based on the frequency domain signal of the first component image of the target image containing the watermark information and the frequency domain signal of the second component image of the target image, the target image is reconstructed to obtain the target image containing the watermark information.

[0134] When the color complexity of the target image is greater than or equal to the first complexity threshold, the process of determining multiple image regions from the target image is triggered.

[0135] In yet another embodiment, the processor 701 may specifically be used to load and execute:

[0136] For any image region, obtain the frequency domain signal type associated with the region type of the image region, wherein the region type is used to indicate whether the image region is a central image region or a symmetrical image region;

[0137] Perform frequency domain transformation processing on the local image indicated by any of the image regions to obtain a target frequency domain signal with the signal type of the frequency domain signal type;

[0138] The target frequency domain signal is subjected to inverse frequency domain transformation to obtain the first component image corresponding to any image region.

[0139] In yet another embodiment, the processor 701 may specifically be used to load and execute:

[0140] Obtain a reference image, wherein the reference image refers to an image obtained by performing image attack processing on the target image containing watermark information;

[0141] The reference image is subjected to an image inverse attack to obtain a candidate parsed image;

[0142] The candidate parsed image is subjected to watermark extraction processing to obtain the watermark information contained in the reference image.

[0143] In another embodiment, when the image attack includes an image color attack, and the image inverse attack includes an image color inverse attack, the processor 701 can specifically be used to load and execute:

[0144] Obtain the color adjustment coefficient;

[0145] For each pixel in the reference image, the result of multiplying the color adjustment coefficient with the color component value of each pixel is used as the target color component value of each pixel;

[0146] The color component values ​​of each pixel in the reference image are updated to the corresponding target color component values ​​to obtain the candidate parsing image.

[0147] In another embodiment, the image attack includes an image mirroring attack, and the image inverse attack includes an image mirroring inverse attack; the processor 701 may specifically be used to load and execute:

[0148] Obtain the reference mirror orientation;

[0149] The reference image is mirrored according to the reference mirror direction to obtain the candidate parsing image.

[0150] In yet another embodiment, the processor 701 can also be used to load and execute:

[0151] From the plurality of image regions, determine the target image region whose color complexity is greater than or equal to the second complexity threshold;

[0152] When the processor 701 performs image decomposition processing on the local images indicated by each image region to obtain the first component image and the second component image of the corresponding local image, it specifically performs the following:

[0153] The local image indicated by the target image region is subjected to image decomposition processing to obtain the first component image and the second component image of the corresponding local image.

[0154] In this embodiment, when watermark information needs to be added to a target image, the computer device determines multiple image regions from the target image and embeds watermark information into the local images indicated by each image region. These multiple image regions include a central image region and at least two symmetrical image regions symmetrical about the central pixel of the target image. This ensures that even after various geometric attacks, including cropping attacks, padding attacks, and sticker attacks, the target image containing watermark information still has a high probability of retaining at least one local image containing watermark information. This guarantees that the watermark information still has a high probability of being successfully parsed after being subjected to geometric attacks, thus giving the target image containing watermark information obtained using this embodiment a strong geometric attack defense capability. Furthermore, when embedding watermark information in a local image, the embedding is performed on the first component image of that local image. The frequency domain signal of the first component image is the target frequency domain signal indicated by a preset frequency band. The preset frequency bands corresponding to the first component images of different local images can be different, allowing watermark information to be embedded in multiple frequency domains within the same target image, such as low-frequency domain, mid-low-frequency domain, and mid-high-frequency domain. Embedding watermark information in the low-frequency domain ensures the robustness of the watermark information, while embedding it in the mid-low-frequency domain or mid-high-frequency domain ensures its concealment. Therefore, by using the embodiments of this application to embed watermark information in multiple frequency domains, the watermark information in the target image can be guaranteed to have both high concealment and robustness.

[0155] This application also provides a computer storage medium storing one or more computer programs corresponding to the above-described image watermarking processing method. When one or more processors load and execute the one or more computer programs, the image watermarking processing method described in the embodiments can be implemented, and will not be repeated here. The beneficial effects of using the same method will also not be repeated here. It is understood that the computer program can be deployed and executed on one or more devices capable of communicating with each other.

[0156] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer storage medium, and when executed, it can include the processes of the embodiments of the image watermarking processing method described above. The computer storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0157] The above-disclosed embodiments are merely partial embodiments of this application and should not be construed as limiting the scope of this application. Those skilled in the art will understand that implementing all or part of the above embodiments and making equivalent changes in accordance with the claims of this application still fall within the scope of this invention.

Claims

1. An image watermarking processing method, characterized in that, include: Multiple image regions are determined from a target image, including a central image region and at least two symmetrical image regions. The central pixel of the target image is located in the central image region, and the positions of the at least two symmetrical image regions in the target image are symmetrical about the central pixel. Image decomposition processing is performed on the local images indicated by each image region to obtain a first component image and a second component image of the corresponding local image. The frequency domain signal of the first component image corresponding to any local image is the target frequency domain signal, and the frequency domain signal of the second component image corresponding to any local image is a frequency domain signal different from the target frequency domain signal. The first component image corresponding to any local image refers to the component image after the local image is decomposed to a preset frequency band. The process of performing image decomposition processing on the local images indicated by each image region to obtain the first component image of the corresponding local image includes: for any image region among the plurality of image regions, obtaining the frequency domain signal type associated with the region type of the any image region, where the region type indicates whether the any image region is a central image region or a symmetrical image region; performing frequency domain transformation processing on the local image indicated by the any image region to obtain a target frequency domain signal with the signal type of the frequency domain signal; and performing inverse frequency domain transformation processing on the target frequency domain signal to obtain the first component image corresponding to the corresponding local image. The frequency domain signal of the first component image corresponding to any image region is the target frequency domain signal. The preset watermark information is embedded into the first component image corresponding to each local image to obtain the first component image containing the watermark information corresponding to each local image. Based on the frequency domain signal of the first component image containing watermark information corresponding to each local image and the frequency domain signal of the second component image corresponding to each local image, image reconstruction processing is performed on each local image to obtain the local image containing watermark information corresponding to each local image. Each local image of the target image is updated to a corresponding local image containing watermark information, thus obtaining the target image containing watermark information.

2. The method according to claim 1, characterized in that, The step of determining multiple image regions from the target image includes: Obtain the coordinates of the center pixel of the target image within the target image; In the target image, a central image region with a region size of a first reference size is determined, and the coordinates of the central pixel point are located within the region coordinates of the central image region; Based on the coordinates of the center pixel, at least two symmetrical image regions with a second reference size are determined in other image regions of the target image besides the center image region.

3. The method according to claim 2, characterized in that, The second reference dimension is determined by the following methods: Obtain the reference number of symmetrical image regions; Based on the other image regions and the reference number, determine the region size of each symmetrical image region in the reference number of symmetrical image regions, wherein the reference number is at least two; The determined area size is used as the second reference size.

4. The method according to claim 3, characterized in that, The step of determining the region size of each symmetrical image region among the reference number of symmetrical image regions based on the other image regions and the reference number includes: Obtain the side lengths of the other image regions; Based on the minimum side length among the side lengths of the other image regions and the reference quantity, the side lengths of each symmetrical image region are determined, wherein the maximum side length among the side lengths of each symmetrical image region is less than the minimum side length corresponding to the other image regions. Determine the region size of each symmetrical image region, wherein the region size includes the side lengths of the respective symmetrical image region.

5. The method according to claim 1, characterized in that, The method further includes: When the color complexity of the target image is less than a first complexity threshold, the target image is decomposed to obtain a first component image and a second component image of the target image. The preset watermark information is embedded into the first component image of the target image. Based on the frequency domain signal of the first component image of the target image containing the watermark information and the frequency domain signal of the second component image of the target image, the target image is reconstructed to obtain the target image containing the watermark information. When the color complexity of the target image is greater than or equal to the first complexity threshold, the process of determining multiple image regions from the target image is triggered.

6. The method according to claim 1, characterized in that, The method further includes: Obtain a reference image, wherein the reference image refers to an image obtained by performing image attack processing on the target image containing watermark information; The reference image is subjected to an image inverse attack to obtain a candidate parsed image; The candidate parsed image is subjected to watermark extraction processing to obtain the watermark information contained in the reference image.

7. The method according to claim 6, characterized in that, When the image attack includes an image color attack, the image inverse attack includes an image color inverse attack; the step of performing image inverse attack processing on the reference image to obtain a candidate parsed image includes: Obtain the color adjustment coefficient; For each pixel in the reference image, the result of multiplying the color adjustment coefficient with the color component value of each pixel is used as the target color component value of each pixel; The color component values ​​of each pixel in the reference image are updated to the corresponding target color component values ​​to obtain the candidate parsing image.

8. The method according to claim 6, characterized in that, The image attack includes an image mirroring attack, and the image inverse attack includes an image mirroring inverse attack; the step of performing image inverse attack processing on the reference image to obtain a candidate parsed image includes: Obtain the reference mirror orientation; The reference image is mirrored according to the reference mirror direction to obtain the candidate parsing image.

9. An image watermarking processing device, characterized in that, include: A determining unit is configured to determine multiple image regions from a target image, the multiple image regions including a central image region and at least two symmetrical image regions, wherein the central pixel of the target image is located in the central image region, and the positions of the at least two symmetrical image regions in the target image are symmetrical about the central pixel. An image decomposition unit is used to perform image decomposition processing on the local images indicated by each image region to obtain a first component image and a second component image of the corresponding local image. The frequency domain signal of the first component image corresponding to any local image is a target frequency domain signal, and the frequency domain signal of the second component image corresponding to any local image is a frequency domain signal different from the target frequency domain signal. The first component image corresponding to any local image refers to the component image after the local image is decomposed to a preset frequency band. The step of performing image decomposition processing on the local images indicated by each image region to obtain the first component image of the corresponding local image includes: for any image region among the plurality of image regions, obtaining the frequency domain signal type associated with the region type of the any image region, where the region type indicates whether the any image region is a central image region or a symmetrical image region; performing frequency domain transformation processing on the local image indicated by the any image region to obtain a target frequency domain signal with the signal type of the frequency domain signal; and performing inverse frequency domain transformation processing on the target frequency domain signal to obtain the first component image corresponding to the corresponding local image. The frequency domain signal of the first component image corresponding to any image region is the target frequency domain signal. The watermark embedding unit is used to embed preset watermark information into the first component image corresponding to each local image, so as to obtain the first component image containing watermark information corresponding to each local image. The image reconstruction unit is used to perform image reconstruction processing on each local image based on the frequency domain signal of the first component image containing watermark information corresponding to each local image and the frequency domain signal of the second component image corresponding to each local image, so as to obtain the local image containing watermark information corresponding to each local image. The update unit is used to update each local image of the target image to a corresponding local image containing watermark information, thereby obtaining a target image containing watermark information.

10. A computer device, characterized in that, The computer device includes: A processor, the processor being adapted to implement one or more computer programs; A computer storage medium storing one or more computer programs, said one or more computer programs being adapted to be loaded by the processor and executed as described in any one of claims 1-8.

11. A computer storage medium, characterized in that, The computer storage medium stores one or more computer programs, which are adapted to be loaded by a processor and executed as described in any one of claims 1-8.

Images