CN110930288A - Method and device for adding and extracting watermark in fundus image - Google Patents
Method and device for adding and extracting watermark in fundus image Download PDFInfo
- Publication number
- CN110930288A CN110930288A CN201911149547.7A CN201911149547A CN110930288A CN 110930288 A CN110930288 A CN 110930288A CN 201911149547 A CN201911149547 A CN 201911149547A CN 110930288 A CN110930288 A CN 110930288A
- Authority
- CN
- China
- Prior art keywords
- image
- watermark
- frequency domain
- fundus
- frequency
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/0021—Image watermarking
- G06T1/005—Robust watermarking, e.g. average attack or collusion attack resistant
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2201/00—General purpose image data processing
- G06T2201/005—Image watermarking
- G06T2201/0052—Embedding of the watermark in the frequency domain
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2201/00—General purpose image data processing
- G06T2201/005—Image watermarking
- G06T2201/0203—Image watermarking whereby the image with embedded watermark is reverted to the original condition before embedding, e.g. lossless, distortion-free or invertible watermarking
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Editing Of Facsimile Originals (AREA)
- Eye Examination Apparatus (AREA)
Abstract
The invention provides a method and a device for adding and extracting a watermark in an eyeground image, wherein the method for adding the watermark comprises the following steps: acquiring a watermark image and an eyeground image; processing the watermark image to hide content represented by the watermark image; performing wavelet transformation on the fundus image to obtain a frequency domain image; adding an image hiding the expression content to a low-frequency part of the frequency domain image according to a preset embedding coefficient, wherein the preset embedding coefficient is used for controlling the visibility of the image; and performing wavelet inverse transformation on the frequency domain image added with the image to obtain the fundus image containing the watermark information.
Description
Technical Field
The invention relates to the field of image processing, in particular to a method and equipment for adding and extracting watermarks in fundus images.
Background
In the field of medical imaging, many eye diseases can be reflected in fundus images, and the eye of a person can be preliminarily judged by the fundus image as shown in fig. 2. Meanwhile, because the fundus map is the only image which can directly display the blood vessels of the human body, the fundus map can be used for judging whether the human body has some related diseases such as diabetes, hypertension, hyperlipidemia and the like.
The fundus image is a valuable asset obtained by photographing with a dedicated camera, and therefore, a related copyright protection measure needs to be added. Digital Watermark (Digital Watermark) is a kind of protection information embedded into carrier file by using computer algorithm, specifically, some identification information is directly embedded into Digital carrier or indirectly expressed, and does not affect the use value of original carrier, and is not easy to be detected and modified again by others, but can be identified and recognized by producer. The information hidden in the carrier can achieve the purposes of confirming content creators and purchasers, transmitting secret information, judging whether the carrier is tampered or not and the like.
The problems exist when the watermark is added through the prior art, the fundus image has high resolution, the image contains much detail information, much interested contents need to be embodied through the details of the image, and the added watermark can cause misunderstanding for the understanding of the image; meanwhile, the integral tone of the fundus image is consistent, the contrast is not obvious as compared with a common image, so that the detail coefficient is small corresponding to the high-frequency energy of the image, and the added watermark is easy to directly detect. The prior art is therefore not suitable for watermarking an image of the fundus.
Disclosure of Invention
In view of the above, the present invention provides a method of adding a watermark to a fundus image, including:
acquiring a watermark image and an eyeground image;
processing the watermark image to hide content represented by the watermark image;
performing wavelet transformation on the fundus image to obtain a frequency domain image;
adding an image hiding the expression content to a low-frequency part of the frequency domain image according to a preset embedding coefficient, wherein the preset embedding coefficient is used for controlling the visibility of the image;
and performing wavelet inverse transformation on the frequency domain image added with the image to obtain the fundus image containing the watermark information.
Optionally, processing the watermark image to hide content expressed by the watermark image includes:
acquiring random sequence information;
and changing the spatial distribution of the watermark image according to the random sequence information.
Optionally, the fundus image is a multi-channel image, and performing wavelet transform on the fundus image to obtain a frequency domain image includes:
extracting data of each channel of the fundus images to obtain a plurality of single-channel fundus images;
and respectively carrying out wavelet transformation on each single-channel fundus image to obtain a plurality of corresponding frequency domain images.
Optionally, adding an image hiding the representation content to a low-frequency portion of the frequency domain image according to a preset embedding coefficient, including:
determining low frequency portions of the plurality of frequency domain images, respectively;
and adding the image hiding the expression content to the low-frequency parts of the plurality of frequency domain images according to preset embedding coefficients respectively.
Optionally, adding a watermark image hiding the representation content to a low-frequency portion of the frequency domain image according to a preset embedding coefficient, including:
dividing an image hiding the expression content into a plurality of partial images;
determining low frequency portions of the plurality of frequency domain images, respectively;
and correspondingly adding a plurality of partial images to the low-frequency parts of the plurality of frequency domain images according to preset embedding coefficients.
Optionally, in the step of adding the image of the hidden expression content to the low-frequency portion of the frequency domain image according to the preset embedding coefficient, processing the pixel value of the image of the hidden expression content according to the preset embedding coefficient to obtain an element to be added, and replacing the value of the corresponding position in the low-frequency portion of the frequency domain image with the element to be added, where the element to be added is to be addedValue of element(s)freq＝pixelwatermark*alpha，pixelwatermarkIn order to hide the pixel value of the image expressing the content, alpha is a preset embedding coefficient, and 0 < alpha < 1.
Optionally, 0.3 ≦ alpha ≦ 0.7.
Accordingly, the present invention provides an apparatus for adding a watermark in a fundus image, comprising: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the one processor to cause the at least one processor to perform the above method of watermarking a fundus image.
According to the watermark adding method and the equipment provided by the invention, firstly, the watermark image containing the semantics is processed to hide the expressed content, and simultaneously, the wavelet transformation is used for obtaining the frequency domain image of the fundus image; and then adding the image hiding the expression content to the low-frequency part of the frequency domain image according to the preset embedding coefficient. Because the low-frequency part is the rough information and the whole expression of the image, the influence of changing the data of the part on the image is small, and the low-frequency part has certain robustness on the compression, noise, rotation and the like of the image;
and the scheme introduces the embedding coefficient to process the added image, further reduces the influence on the image and reduces the visibility. In addition, the added image hides the expression content, and even if the added image can be found, the meaning of the added image cannot be identified, so that the secrecy of the watermark information is further improved.
The invention also provides a method for extracting the watermark in the fundus image, which comprises the following steps:
acquiring an eyeground image containing watermark information;
performing wavelet transformation on the fundus image to obtain a frequency domain image;
extracting an image hiding the expression content from a low-frequency part of the frequency domain image according to a preset extraction coefficient, wherein the preset extraction coefficient is used for controlling the visibility of the extracted image;
and restoring the extracted image of the hidden expression content to obtain a watermark image.
Optionally, the fundus image is a multi-channel image, and performing wavelet transform on the fundus image to obtain a frequency domain image includes:
extracting data of each channel of the fundus images to obtain a plurality of single-channel fundus images;
and respectively carrying out wavelet transformation on each single-channel fundus image to obtain a plurality of corresponding frequency domain images.
Optionally, extracting an image hiding the expression content from a low-frequency portion of the frequency domain image according to a preset extraction coefficient includes:
determining low frequency portions of the plurality of frequency domain images, respectively;
obtaining elements from low-frequency parts of the plurality of frequency domain images according to preset extraction coefficients respectively;
and averaging the obtained values of the elements to obtain an image hiding the expression content.
Optionally, extracting an image hiding the expression content from a low-frequency portion of the frequency domain image according to a preset extraction coefficient includes:
determining low frequency portions of the plurality of frequency domain images, respectively;
obtaining elements from low-frequency parts of the plurality of frequency domain images according to preset extraction coefficients respectively;
and combining the obtained elements in a spatial domain to obtain an image with hidden expression content.
Optionally, in the step of extracting an image of hidden representation content from a low-frequency portion of the frequency domain image according to preset extraction coefficients, the element pixels for composing the image of hidden representation content are obtained using the following mannerwatermark：pixelwatermark＝valuefreq*alpha，valuefreqAlpha is a preset extraction coefficient, and alpha is more than 1 and less than 2.
Alternatively, 1.2 ≦ alpha ≦ 1.8.
Optionally, the restoring the extracted image of the hidden representation content to obtain a watermark image includes:
acquiring random sequence information;
and changing the spatial distribution of the image of the hidden expression content according to the random sequence information to obtain a watermark image.
Accordingly, the present invention also provides an apparatus for extracting a watermark in a fundus image, comprising: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the one processor to cause the at least one processor to perform the above method of extracting a watermark in a fundus image.
According to the watermark extraction method and the watermark extraction equipment provided by the embodiment of the invention, firstly, the frequency domain image of the fundus image is obtained through wavelet transformation, then elements are extracted from the low-frequency part of the frequency domain image by using the preset extraction coefficient, and the visibility of the extracted image is improved, so that the image with hidden expression content is obtained; finally, the obtained image is restored to obtain an actual watermark image, even if the fundus image is compressed, added with noise and rotated, the content of the watermark image can be restored, and the scheme has strong robustness.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of a watermarking method in an embodiment of the present invention;
FIG. 2 is a fundus image;
FIG. 3 is a diagram illustrating a wavelet transform in an embodiment of the present invention;
FIG. 4 is a frequency domain image of a fundus image in an embodiment of the present invention;
fig. 5 is a flowchart of a watermark extraction method in an embodiment of the present invention.
Detailed Description
The technical solutions of the present invention will be described clearly and completely with reference to the accompanying drawings, and it should be understood that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The technical features mentioned in the different embodiments of the invention described below can be combined with each other as long as they do not conflict with each other.
Embodiments of the present invention provide a method for adding a watermark to an eye fundus image, which may be executed by an electronic device such as a computer or a server. As shown in fig. 1, the method comprises the following steps:
S1A, a watermark image and fundus image are acquired. The watermark image contains several semantic information, such as characters, patterns, etc., as the identifier of the producer. The watermark image can be a single channel image, such as bitmap, gray scale map, etc., or a multi-channel image, such as an RGB image.
Fig. 2 shows a fundus image taken by a fundus camera. The fundus image is human tissue, the whole image has consistent tone, and the contrast is not obvious as a common image.
S2A, the watermark image is processed to hide the content represented by the watermark image. The specific processing method includes various ways, such as changing color values, changing the position relationship of pixel points, and the like. In order to strengthen the hiding effect, all pixel points of the watermark image can be obtained, and each pixel point is processed respectively so as to change the content of the image; in order to improve the processing efficiency, part of pixel points of the watermark image can be selected for processing, and the number of the selected pixel points can achieve the purpose of hiding the original content.
And S3A, performing wavelet transformation on the fundus image to obtain a frequency domain image. Fig. 3 is a schematic diagram of wavelet transformation, in which an original image is decomposed into four small images, the upper left corner represents a low frequency, which is a rough information and overall expression of the image, and the energy is the largest and is not easily affected by various factors such as compression loss; the high-frequency part of the image is arranged at the lower right, the middle-frequency part of the image is arranged in the diagonal direction, and the frequency bands represent the detail part of the image and are easily influenced by various factors.
If the low-frequency part is wavelet-transformed, the result of the second-level wavelet transform can be obtained, the image is expressed in a higher level, and is not easily influenced by various factors, and simultaneously, the amount of watermark information which can be borne is reduced due to the further reduction of the resolution, so that the times of the wavelet transform can be set according to the actual image condition, and the execution of one or more times of wavelet transform processing is feasible.
S4A, adding the image hiding the expressive content to the low-frequency part of the frequency-domain image according to the preset embedding coefficient. Because many focuses in the eye background image are small and belong to detailed information, in order to prevent the added watermarks from influencing the information of the eye background image and simultaneously keep robustness to compression loss, noise and the like, the invention selects a low-frequency part as a channel for adding the watermarks.
Fig. 4 shows a frequency domain image of a fundus image, which is a low-frequency portion (corresponding to the upper left portion of fig. 3), two intermediate-frequency portions (corresponding to the lower left and upper right portions of fig. 3), and a high-frequency portion (corresponding to the lower right portion of fig. 3) in this order from left to right. Therefore, the frequency domain of the fundus image can be found to be different from the frequency domain of a common image, any information added to the medium-frequency or high-frequency part of the fundus image can be easily detected, and certain confidentiality can be achieved only by adding information to the low-frequency part.
Even if the low-frequency part adding information is selected, directly overlapping or replacing elements therein easily causes the image to be changed significantly, so the invention introduces the preset embedding coefficient to control the visibility of the added image, for example, the visibility of the image to be added can be weakened by an appropriate value (not easily observed by vision), or the invention is understood to improve the transparency of the image to be added. There are several specific algorithms, and the value ranges of the coefficients are different in different algorithms.
And S5A, performing wavelet inverse transformation on the frequency domain image added with the image to obtain the fundus image containing the watermark information.
According to the watermark adding method provided by the embodiment of the invention, firstly, a watermark image containing semantics is processed to hide the expressed content of the watermark image, and meanwhile, a frequency domain image of an eye fundus image is obtained by using wavelet transformation; and then adding the image hiding the expression content to the low-frequency part of the frequency domain image according to the preset embedding coefficient. Because the low-frequency part is the rough information and the whole expression of the image, the influence of changing the data of the part on the image is small, and the low-frequency part has certain robustness on the compression, noise, rotation and the like of the image;
and the scheme introduces the embedding coefficient to process the added image, further reduces the influence on the image and reduces the visibility. In addition, the added image hides the expression content, and even if the added image can be found, the meaning of the added image cannot be identified, so that the secrecy of the watermark information is further improved.
The size of the watermark image is related to the information amount of the watermark, and different adding modes can be adopted for different sizes. The present invention provides two alternative embodiments, first performing wavelet transform in step S3A using the following method:
data of each channel of the fundus image is extracted to obtain a plurality of single-channel fundus images, for example, an original image is a color three-channel RGB image, and an R-channel image, a G-channel image and a B-channel image are extracted and obtained here.
And respectively carrying out wavelet transformation on each single-channel fundus image to obtain a plurality of corresponding frequency domain images. Thereby obtaining a first frequency domain image corresponding to the R channel image, a second frequency domain image corresponding to the G channel image and a third frequency domain image corresponding to the B channel image.
As a first alternative, the step S4A may include the following steps:
low frequency portions of the plurality of frequency domain images are determined separately. According to the above example, the low frequency part of the first frequency domain image, the low frequency part of the second frequency domain image, and the low frequency part of the third frequency domain image are determined, respectively.
And adding the image hiding the expression content to the low-frequency parts of the plurality of frequency domain images according to preset embedding coefficients respectively. That is, the image obtained in step S2A is added to the low frequency parts of the above three frequency domain images at the same time.
The embodiment is suitable for the watermark image with small information quantity, the same watermark information is stored by utilizing the 3 channels of the original image, and the watermarks extracted by the 3 channels are subjected to average calculation when the watermark information is recovered, so that the recovery definition of the watermarks is further improved under the condition of keeping the same invisibility to naked eyes.
As a second alternative, the step S4A may include the following steps:
the image hiding the expressive content is divided into a plurality of partial images. Specifically, the division is performed according to the number of channels of the original image, and the image obtained in step S2A is divided into 3 partial images according to the above example. The fundus image resolution collected for different fundus cameras varies mostly from 1000 x 1000 to 2000 x 2000. The original image is generally a color three-channel RGB image, if the watermarks are all added to one channel, the minimum watermark image of resize is about 128 × 128 under the condition of keeping the recovery definition, if the watermark image is smaller, the recovery will be unclear, and if the size is larger, the watermark trace will be more obvious.
Low frequency portions of the plurality of frequency domain images are determined separately. According to the above example, the low frequency part of the first frequency domain image, the low frequency part of the second frequency domain image, and the low frequency part of the third frequency domain image are determined, respectively.
And correspondingly adding the partial images to the low-frequency parts of the frequency domain images according to the preset embedding coefficient. This method is suitable for a large-sized watermark image, and in order to fully utilize the original image, the present embodiment divides watermark information into three parts, reduces the watermark map of each part of information to 64 × 64, and embeds the information in each channel of the original image, thereby further reducing the traces of the watermark when the visibility is restored.
A specific application of the preset embedding coefficient is described below. For convenience of explanation, it is assumed that only one frequency domain image is obtained in step S4A. Firstly, the image obtained in step S2A is processedProcessing in pixel modewatermarkAlpha, wherein pixelwatermarkIs the value of the pixel point of the image obtained in step S2A, and alpha is a preset embedding coefficient, thereby obtaining valuefreqAs elements added to the frequency domain image. Then use valuefreqAnd replacing the value of the corresponding position in the low-frequency part of the frequency domain image, wherein the value range of alpha is 0 < alpha < 1, and the value range is used for reducing the visibility.
Further, experiments on a large number of fundus images show that the embedding coefficient alpha of the fundus images is less easily observed visually when the embedding coefficient alpha is between 0.3 and 0.7, and the watermark is extracted in the reverse direction clearly, and the alpha is actually selected to be 0.5 as the optimal value.
When a plurality of frequency domain images are obtained in step S4A, the watermark image or the divided partial watermark image is added to each frequency domain image by the above-mentioned processing method.
Describing a specific manner of expressing contents in hidden layer images, in an alternative embodiment, the step S2A includes the following steps:
and acquiring random sequence information. In order to prevent the watermark from being decrypted by others, a key can be preset, and a random sequence is generated according to the key value.
And changing the spatial distribution of the watermark image according to the random sequence information. I.e. the content of the watermark image is scrambled in the order of the random sequence.
The watermarking scheme provided by the invention is particularly suitable for processing sample images in training data. For example, when a neural network for classifying fundus lesions or segmenting a focus region needs to be trained, a large number of fundus images of known types need to be acquired as sample images, after a producer adds watermarks by using the scheme, other people cannot find watermark information in the sample images easily or cannot find the watermark information in the sample images through vision, and when other people train the neural network by using the sample images added with the watermarks, the performance of the neural network is seriously influenced, so that an actually usable image classification or segmentation model cannot be obtained, and the aim of indirectly protecting training data is fulfilled.
The embodiment of the invention also provides equipment for adding the watermark in the fundus image, which comprises at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the one processor to cause the at least one processor to perform the above method of watermarking a fundus image.
When a producer wants to verify whether a certain fundus image is from the source, if the watermark image can be obtained through reverse processing, the fundus image can be proved to be from the producer. Accordingly, an embodiment of the present invention provides a method for extracting a watermark in a fundus image, for processing a fundus image to which watermark information is added in the above manner, which may be executed by an electronic device such as a computer or a server. As shown in fig. 5, the method comprises the following steps:
S1B, a fundus image including watermark information is acquired, specifically, the result obtained in the above-described step S5A.
And S2B, performing wavelet transformation on the fundus image to obtain a frequency domain image. Specifically, reference may be made to the step S3A, which is not described herein again.
S3B, extracting an image of the hidden representation content from the low frequency part of the frequency domain image according to a preset extraction coefficient. The frequency domain image obtained here contains extra information, and as can be seen from step S4A, the contained information is a watermark image with hidden expressive content, and is processed by presetting embedding coefficients.
The preset extraction coefficient is used to control the visibility of the extracted image. If elements in the frequency domain image are directly extracted to form an image, the visibility is poor, and a watermark image cannot be restored. In order to improve the visibility, the invention improves the strength of the extracted content through proper numerical values, and the specific algorithms are various, and the value ranges of the coefficients are different in different algorithms.
And S4B, restoring the extracted image of the hidden expression content to obtain a watermark image. This step is the reverse of step S2A described above.
According to the watermark extraction method provided by the embodiment of the invention, firstly, a frequency domain image of a fundus image is obtained through wavelet transformation, then elements are extracted from a low-frequency part of the frequency domain image by using a preset extraction coefficient, and the visibility of the extracted image is improved, so that an image with hidden expression content is obtained; finally, the obtained image is restored to obtain an actual watermark image, even if the fundus image is compressed, added with noise and rotated, the content of the watermark image can be restored, and the scheme has strong robustness.
For different ways of adding watermarks, the invention provides two corresponding alternative extraction ways, firstly in step S2B, the following way is used for wavelet transform:
data of each channel of the fundus image is extracted to obtain a plurality of single-channel fundus images, for example, an original image is a color three-channel RGB image, and an R-channel image, a G-channel image and a B-channel image are extracted and obtained here.
And respectively carrying out wavelet transformation on each single-channel fundus image to obtain a plurality of corresponding frequency domain images. Thereby obtaining a first frequency domain image corresponding to the R channel image, a second frequency domain image corresponding to the G channel image and a third frequency domain image corresponding to the B channel image.
As a first alternative, the step S3B may include the following steps:
low frequency portions of the plurality of frequency domain images are determined separately. According to the above example, the low frequency part of the first frequency domain image, the low frequency part of the second frequency domain image, and the low frequency part of the third frequency domain image are determined, respectively.
Obtaining elements from low-frequency parts of the plurality of frequency domain images according to preset extraction coefficients respectively;
and averaging the obtained values of the elements to obtain an image hiding the expression content.
The method is used for processing the condition that the same watermark information is stored in the frequency domain images of all the channels, and when the watermark information is recovered, the watermarks extracted from 3 channels are subjected to average calculation, so that the recovery definition of the watermarks is further improved under the condition that the same invisibility to naked eyes is kept.
As a second alternative, the step S3B may include the following steps:
determining low-frequency parts of a plurality of frequency domain images respectively;
obtaining elements from low-frequency parts of the plurality of frequency domain images according to preset extraction coefficients respectively;
and combining the obtained elements in a spatial domain to obtain an image with hidden expression content.
The above method is used for processing the condition that watermark information of different parts is stored in the frequency domain image of each channel, and the extracted parts are spliced in space to obtain a complete image.
The following describes a specific application of the preset extraction coefficient. For convenience of explanation, it is assumed that only one frequency domain image is obtained in step S3B. Obtaining element pixels for composing an image of hidden expressive content using the following mannerwatermark：pixelwatermark＝valuefreq*alpha，valuefreqAlpha is a preset extraction coefficient, and alpha is more than 1 and less than 2.
The above processing manner can improve the visibility of the extracted image, and the extraction coefficient used in the present embodiment is larger than the embedding coefficient used in the addition process.
Further, experiments on a large number of fundus images show that the recovery watermark is relatively clear when the extraction coefficient alpha of the fundus images is between 1.2 and 1.8, and the alpha is actually selected to be 1.5 as the optimal value.
When a plurality of frequency domain images are obtained in step S3A, the watermark image or each partial watermark image is extracted by using the above processing method for each frequency domain image.
A specific way to restore the content of the hidden layer image representation is described below, and in an alternative embodiment, the step S4B includes the following steps:
and acquiring random sequence information. Specifically, a preset key in the adding process is obtained, and a random sequence is generated according to the key value.
And changing the spatial distribution of the image of the hidden expression content according to the random sequence information to obtain the watermark image. That is, the watermark image can be restored by rearranging the images according to the sequence of the random sequence.
The invention also provides a device for extracting a watermark in an eyeground image, comprising at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the one processor to cause the at least one processor to perform the above method of extracting a watermark in a fundus image.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It should be understood that the above examples are only for clarity of illustration and are not intended to limit the embodiments. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. And obvious variations or modifications therefrom are within the scope of the invention.
Claims (16)
1. A method of watermarking a fundus image, comprising:
acquiring a watermark image and an eyeground image;
processing the watermark image to hide content represented by the watermark image;
performing wavelet transformation on the fundus image to obtain a frequency domain image;
adding an image hiding the expression content to a low-frequency part of the frequency domain image according to a preset embedding coefficient, wherein the preset embedding coefficient is used for controlling the visibility of the image;
and performing wavelet inverse transformation on the frequency domain image added with the image to obtain the fundus image containing the watermark information.
2. The method of claim 1, wherein processing the watermark image to conceal the content represented by the watermark image comprises:
acquiring random sequence information;
and changing the spatial distribution of the watermark image according to the random sequence information.
3. The method of claim 1, wherein the fundus image is a multi-channel image, and wherein performing a wavelet transform on the fundus image to obtain a frequency domain image comprises:
extracting data of each channel of the fundus images to obtain a plurality of single-channel fundus images;
and respectively carrying out wavelet transformation on each single-channel fundus image to obtain a plurality of corresponding frequency domain images.
4. The method of claim 3, wherein adding the image hiding the representation content to the low frequency part of the frequency domain image according to preset embedding coefficients comprises:
determining low frequency portions of the plurality of frequency domain images, respectively;
and adding the image hiding the expression content to the low-frequency parts of the plurality of frequency domain images according to preset embedding coefficients respectively.
5. The method according to claim 3, wherein adding a watermark image hiding the representation content to the low frequency part of the frequency domain image according to preset embedding coefficients comprises:
dividing an image hiding the expression content into a plurality of partial images;
determining low frequency portions of the plurality of frequency domain images, respectively;
and correspondingly adding a plurality of partial images to the low-frequency parts of the plurality of frequency domain images according to preset embedding coefficients.
6. The method according to any one of claims 1 to 5, wherein in the step of adding the image of the hidden representation content to the low frequency part of the frequency domain image according to the preset embedding coefficient, the pixel values of the image of the hidden representation content are processed according to the preset embedding coefficient to obtain an element to be added, and the element to be added is used for replacing the element to be added in the low frequency part of the frequency domain imageValue of position, wherein value of element to be addedfreq＝pixelwatermark*alpha，pixelwatermarkAlpha is a predetermined embedding coefficient, 0, for hiding pixel values of an image representing content<alpha<1。
7. The method of claim 6, wherein 0.3 ≦ alpha ≦ 0.7.
8. A method of extracting a watermark in a fundus image, comprising:
acquiring an eyeground image containing watermark information;
performing wavelet transformation on the fundus image to obtain a frequency domain image;
extracting an image hiding the expression content from a low-frequency part of the frequency domain image according to a preset extraction coefficient, wherein the preset extraction coefficient is used for controlling the visibility of the extracted image;
and restoring the extracted image of the hidden expression content to obtain a watermark image.
9. The method of claim 8, wherein the fundus image is a multi-channel image, and wherein performing a wavelet transform on the fundus image to obtain a frequency domain image comprises:
extracting data of each channel of the fundus images to obtain a plurality of single-channel fundus images;
and respectively carrying out wavelet transformation on each single-channel fundus image to obtain a plurality of corresponding frequency domain images.
10. The method according to claim 9, wherein extracting the image of the hidden representation content from the low frequency part of the frequency domain image according to preset extraction coefficients comprises:
determining low frequency portions of the plurality of frequency domain images, respectively;
obtaining elements from low-frequency parts of the plurality of frequency domain images according to preset extraction coefficients respectively;
and averaging the obtained values of the elements to obtain an image hiding the expression content.
11. The method according to claim 9, wherein extracting the image of the hidden representation content from the low frequency part of the frequency domain image according to preset extraction coefficients comprises:
determining low frequency portions of the plurality of frequency domain images, respectively;
obtaining elements from low-frequency parts of the plurality of frequency domain images according to preset extraction coefficients respectively;
and combining the obtained elements in a spatial domain to obtain an image with hidden expression content.
12. Method according to any one of claims 8 to 11, characterized in that, in the step of extracting an image of hidden representation from the low-frequency part of the frequency-domain image according to preset extraction coefficients, the element pixels used to compose the image of hidden representation are obtained usingwatermark：pixelwatermark＝valuefreq*alpha，valuefreqFor the values of the frequency domain image, alpha is a preset extraction coefficient, 1<alpha<2。
13. The method of claim 12, wherein 1.2 ≦ alpha ≦ 1.8.
14. The method of claim 8, wherein the step of restoring the extracted image of the hidden representation content to obtain a watermark image comprises:
acquiring random sequence information;
and changing the spatial distribution of the image of the hidden expression content according to the random sequence information to obtain a watermark image.
15. An apparatus for adding a watermark to a fundus image, comprising: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the one processor to cause the at least one processor to perform the method of watermarking a fundus image according to any one of claims 1 to 7.
16. An apparatus for extracting a watermark in a fundus image, comprising: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the one processor to cause the at least one processor to perform the method of extracting a watermark in a fundus image of a eye as claimed in any one of claims 8 to 14.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911149547.7A CN110930288B (en) | 2019-11-21 | 2019-11-21 | Method and apparatus for adding and extracting watermark in fundus image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911149547.7A CN110930288B (en) | 2019-11-21 | 2019-11-21 | Method and apparatus for adding and extracting watermark in fundus image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110930288A true CN110930288A (en) | 2020-03-27 |
| CN110930288B CN110930288B (en) | 2023-09-08 |
Family
ID=69850629
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911149547.7A Active CN110930288B (en) | 2019-11-21 | 2019-11-21 | Method and apparatus for adding and extracting watermark in fundus image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110930288B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118152996A (en) * | 2024-03-13 | 2024-06-07 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Image copyright protection method based on diffusion model |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040091050A1 (en) * | 2000-09-09 | 2004-05-13 | Jong-Uk Choi | Digital image watermarking apparatus and method |
| US20040258243A1 (en) * | 2003-04-25 | 2004-12-23 | Dong-Hwan Shin | Method for embedding watermark into an image and digital video recorder using said method |
| CN102496135A (en) * | 2011-12-06 | 2012-06-13 | 银江股份有限公司 | Deadweight tonnage (DWT) domain-based digital watermark method and system |
| US20120163652A1 (en) * | 2009-09-03 | 2012-06-28 | Zte Corporation | Method and System for Embedding and Extracting Image Digital Watermark |
| CN104680470A (en) * | 2013-11-26 | 2015-06-03 | 上海宝钢包装股份有限公司 | Method for embedding and extracting strong robustness hologram watermark on the basis of DWT-SVD (Discrete Wavelet Transform-Singular Value Decomposition) |
| CN106126978A (en) * | 2016-06-30 | 2016-11-16 | 中南大学 | The lossless digital copyright management method of eye fundus image based on fingerprint Yu zero watermarking mixed architecture |
| CN106570813A (en) * | 2016-10-10 | 2017-04-19 | 湖南正晨节能科技有限公司 | Holographic digital watermark embedding method, extraction method and device |
| CN109447889A (en) * | 2018-11-22 | 2019-03-08 | 央视国际网络无锡有限公司 | A method of realizing concealed video digital watermark |
| CN110264391A (en) * | 2019-06-26 | 2019-09-20 | 上海鹰瞳医疗科技有限公司 | Training data processing method and equipment |
-
2019
- 2019-11-21 CN CN201911149547.7A patent/CN110930288B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040091050A1 (en) * | 2000-09-09 | 2004-05-13 | Jong-Uk Choi | Digital image watermarking apparatus and method |
| US20040258243A1 (en) * | 2003-04-25 | 2004-12-23 | Dong-Hwan Shin | Method for embedding watermark into an image and digital video recorder using said method |
| US20120163652A1 (en) * | 2009-09-03 | 2012-06-28 | Zte Corporation | Method and System for Embedding and Extracting Image Digital Watermark |
| CN102496135A (en) * | 2011-12-06 | 2012-06-13 | 银江股份有限公司 | Deadweight tonnage (DWT) domain-based digital watermark method and system |
| CN104680470A (en) * | 2013-11-26 | 2015-06-03 | 上海宝钢包装股份有限公司 | Method for embedding and extracting strong robustness hologram watermark on the basis of DWT-SVD (Discrete Wavelet Transform-Singular Value Decomposition) |
| CN106126978A (en) * | 2016-06-30 | 2016-11-16 | 中南大学 | The lossless digital copyright management method of eye fundus image based on fingerprint Yu zero watermarking mixed architecture |
| CN106570813A (en) * | 2016-10-10 | 2017-04-19 | 湖南正晨节能科技有限公司 | Holographic digital watermark embedding method, extraction method and device |
| CN109447889A (en) * | 2018-11-22 | 2019-03-08 | 央视国际网络无锡有限公司 | A method of realizing concealed video digital watermark |
| CN110264391A (en) * | 2019-06-26 | 2019-09-20 | 上海鹰瞳医疗科技有限公司 | Training data processing method and equipment |
Non-Patent Citations (2)
| Title |
|---|
| 兀旦晖;牟曦媛;王鑫;李慧贞;杨萍;: "一种基于人眼视觉系统的小波水印系统研究" * |
| 张丽林;李蕊;: "一种基于人眼视觉特性的小波域图像水印方法" * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118152996A (en) * | 2024-03-13 | 2024-06-07 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Image copyright protection method based on diffusion model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110930288B (en) | 2023-09-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Fan et al. | JPEG anti-forensics with improved tradeoff between forensic undetectability and image quality | |
| CN111680273B (en) | Watermark embedding method, watermark embedding device, electronic equipment and readable storage medium | |
| US8213674B2 (en) | Perceptual modeling of media signals for data hiding | |
| Abdullatif et al. | Properties of digital image watermarking | |
| EP0901274B1 (en) | Method and system for embedding information into contents | |
| US20030097568A1 (en) | Watermaking system and method for protecting a digital image from forgery or alteration | |
| Rathi et al. | Medical images authentication through watermarking preserving ROI | |
| CN112700363B (en) | Self-adaptive visual watermark embedding method and device based on region selection | |
| Kim et al. | Region-based tampering detection and recovery using homogeneity analysis in quality-sensitive imaging | |
| EP1041815B1 (en) | Image processing apparatus, image processing method, and storage medium | |
| CN106454367A (en) | Method and system for identifying multimedia file | |
| CN105898324A (en) | Video watermark hidden insertion method and device | |
| CN110930288B (en) | Method and apparatus for adding and extracting watermark in fundus image | |
| Wang et al. | Robust image watermarking via perceptual structural regularity-based JND model | |
| JP2016072706A (en) | Strength setting method for watermark embedding and computer program | |
| Singh et al. | A blind & fragile watermarking scheme for tamper detection of medical images preserving ROI | |
| EP2165453B1 (en) | Volume marking with low-frequency | |
| CN113393363B (en) | Watermark embedding and extracting method, storage medium and electronic device | |
| Singh et al. | A reversible data hiding scheme for efficient management of tele-ophthalmological data | |
| JP4167372B2 (en) | Digital watermark embedding method, extraction method, invisibility method, visualization method, and embedding device | |
| Batskos et al. | Visualizing landmark-based face morphing traces on digital images | |
| Zaheer et al. | A literature survey on various approaches of data hiding in images | |
| Knowles et al. | Image tamper detection and classification using support vector machines | |
| Pakala et al. | Forgery Detection in Medical Image and Enhancement using Modified CLAHE Method | |
| KR20130065277A (en) | Video water marking inserting/detecting apparatus using modulation of time horizon |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |