CN114339258A - Information steganography method and device based on video carrier - Google Patents
Information steganography method and device based on video carrier Download PDFInfo
- Publication number
- CN114339258A CN114339258A CN202111626840.5A CN202111626840A CN114339258A CN 114339258 A CN114339258 A CN 114339258A CN 202111626840 A CN202111626840 A CN 202111626840A CN 114339258 A CN114339258 A CN 114339258A
- Authority
- CN
- China
- Prior art keywords
- video
- carrier
- generated
- steganography
- generating
- 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
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000012549 training Methods 0.000 claims abstract description 56
- 238000012986 modification Methods 0.000 claims abstract description 44
- 230000004048 modification Effects 0.000 claims abstract description 44
- 239000011159 matrix material Substances 0.000 claims abstract description 34
- 238000010586 diagram Methods 0.000 claims abstract description 22
- 230000008569 process Effects 0.000 abstract description 5
- 238000005457 optimization Methods 0.000 abstract description 4
- 230000006870 function Effects 0.000 description 32
- 230000004913 activation Effects 0.000 description 13
- 238000012545 processing Methods 0.000 description 8
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000013461 design Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000000605 extraction Methods 0.000 description 4
- 230000003068 static effect Effects 0.000 description 4
- 230000003042 antagnostic effect Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000009826 distribution Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 239000000969 carrier Substances 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
- 238000003892 spreading Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Landscapes
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
- Image Processing (AREA)
- Editing Of Facsimile Originals (AREA)
Abstract
The invention discloses an information steganography method and a device based on a video carrier, wherein the method comprises the following steps: generating a modified probability matrix for the countermeasure network according to the steganography generated by pre-training based on the carrier video, wherein the carrier video is a pseudo video which is generated by the network and accords with natural semantics; the modified probability matrix is subjected to self-adaptive generation of an optimal embedded modified position diagram through an optimal binary embedded function; and embedding the secret information in the position corresponding to the optimal embedded modification position diagram of the carrier video to generate the secret-contained video. The invention can effectively solve the problems that the existing network can not directly process the video carrier with large data volume and the optimization modification strategy in the traditional video information hiding algorithm is difficult to construct, and realizes the information hiding by taking the video as the carrier.
Description
Technical Field
The invention relates to the technical field of information security, in particular to an information steganography method and device based on a video carrier.
Background
Steganography is a secret communication technology which hides secret information in carrier information, so that an attacker cannot know whether the carrier contains the secret information or not, and the purpose of hidden transmission is achieved. The method is widely applied to the aspects of public service, network transmission, military communication and the like, is not easy to be analyzed and detected by a malicious attacker due to the imperceptibility of the method compared with an encryption technology, is a popular subject in the field of information security at present, and has important application value in the departments of military affairs, intelligence and the like.
However, most of the existing digital steganography uses digital images as carriers, the research on the digital video steganography technology is relatively less, considering that the digital video has larger absolute data quantity compared with the images, the embedding capacity and the security of the digital video have better performance, along with the development of a 5G high-speed network, a large amount of video medium information is rapidly spread in the internet, the digital video is already a commonly used media form and is an ideal data hiding carrier, and therefore, the advantages of the video steganography in the aspects of carrier quantity and spreading security are increasingly highlighted.
The existing video steganography technology generally selects a position from a DCT (Discrete Cosine Transform) coefficient or a motion vector in a compressed video through a designed rule, and implements information embedding by modifying data of the position; and relatively few video steganography algorithms based on an antagonistic neural network.
The existing video steganography technology mainly has the following defects: 1. the existing video steganography algorithm improves the security of the steganography algorithm by enabling the statistical distortion caused after embedding to be as small as possible through a certain modification strategy, but the modification strategy is generally established through experience, and the optimal modification strategy is difficult to realize manually in design. 2. The existing steganography algorithm based on the neural network is mainly applied to image and text steganography, and cannot be directly applied to digital videos due to network construction, running speed, data processing capacity and the like. Meanwhile, the invisibility and the safety of the existing neural network-based steganography algorithm are still to be improved. Moreover, the secret information is embedded into a single image, the fact that each frame of image of a video sequence has different bottom layer characteristic information is not considered, the information cannot be reasonably distributed into the whole video sequence, and different secret information is embedded into different frame images of the video.
Therefore, how to effectively perform information steganography by using video as a carrier is a problem to be solved.
Disclosure of Invention
In view of this, the present invention provides an information steganography method based on a video carrier, which can effectively perform information steganography by using a video as a carrier.
The invention provides an information steganography method based on a video carrier, which comprises the following steps:
generating a modified probability matrix for a countermeasure network according to steganography generated by pre-training based on a carrier video, wherein the carrier video is a pseudo video which is generated by a network and accords with natural semantics;
the modification probability matrix is subjected to self-adaptive generation of an optimal embedding modification position diagram through an optimal binary embedding function;
and embedding secret information at the position corresponding to the optimal embedding modification position diagram of the carrier video to generate a secret video.
Preferably, before generating the modified probability matrix according to the pre-trained generated hidden-writing generation countermeasure network based on the carrier video, the method further includes:
and generating a carrier video according to the video generated by the pre-training.
Preferably, the generating of the countermeasure network generation carrier video according to the video generated by the pre-training comprises:
and taking noise as the input of the video generation countermeasure network generated by the pre-training to generate the carrier video consisting of the foreground, the background and the mask.
Preferably, the generating the carrier video composed of foreground, background and mask by using noise as the input of the video generation countermeasure network generated by the pre-training comprises:
and taking noise as an input of the video generation countermeasure network generated by the pre-training, generating the foreground and the mask through a foreground generator in the video generation countermeasure network generated by the pre-training, and generating the background through a background generator in the video generation countermeasure network generated by the pre-training, wherein the carrier video consisting of the foreground, the background and the mask.
Preferably, the generating, based on the carrier video, the modified probability matrix according to the pre-trained generated steganography generation countermeasure network includes:
and taking the foreground in the carrier video as the input of the pre-generated steganography generation countermeasure network, and generating a modified probability matrix through a steganography generator in the pre-generated steganography generation countermeasure network.
An information steganography apparatus based on a video carrier, comprising:
the method comprises the steps that a pair of rejection networks are generated through pre-training and are used for generating a modified probability matrix based on a carrier video, wherein the carrier video is a pseudo video conforming to natural semantics;
generating a countermeasure network by the pre-trained generated steganography, and adaptively generating an optimal embedded modification position diagram by the modification probability matrix through an optimal binary embedded function;
and generating a countermeasure network by the pre-training generated steganography, and embedding secret information at the position corresponding to the optimal embedding modification position diagram of the carrier video to generate a dense video.
Preferably, the apparatus further comprises:
and training the generated video in advance to generate a countermeasure network for generating the carrier video.
Preferably, the video generation countermeasure network generated by the pre-training is specifically configured to:
and generating the carrier video consisting of the foreground, the background and the mask by taking the noise as an input.
Preferably, the pre-training generated video generation countermeasure network includes: a foreground generator and a background generator; wherein:
the foreground generator is used for generating the foreground and the mask by taking noise as input;
and the background generator is used for generating the background by taking noise as input, wherein the carrier video is composed of the foreground, the background and a mask.
Preferably, the pre-training generated steganography generation countermeasure network comprises: a steganography generator; wherein:
the steganography generator is used for generating a modified probability matrix by taking the foreground in the carrier video as input.
In summary, the invention discloses an information steganography method based on a video carrier, when information steganography is required to be carried out by taking a video as the carrier, firstly, a modified probability matrix is generated for a countermeasure network according to steganography generated by pre-training based on the carrier video, wherein the carrier video is a pseudo video conforming to natural semantics; and then, adaptively generating an optimal embedding modification position diagram by the modification probability matrix through an optimal binary embedding function, and embedding secret information in a position corresponding to the optimal embedding modification position diagram of the carrier video to generate a secret-contained video. The invention can effectively solve the problems that the existing network can not directly process the video carrier with large data volume and the optimization modification strategy in the traditional video information hiding algorithm is difficult to construct, and realizes the information hiding by taking the video as the carrier.
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, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flowchart of a method of an embodiment of a method for steganography of information based on a video carrier according to the present disclosure;
FIG. 2 is a schematic structural diagram of an embodiment of a video generator in a video generation countermeasure network according to the present disclosure;
FIG. 3 is a schematic diagram of the separation and combination of three channels of the steganographic discriminator disclosed in the present invention;
fig. 4 is a schematic structural diagram of an embodiment of an information steganography apparatus based on a video carrier.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. 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.
As shown in fig. 1, a method for steganography of information based on a video carrier is disclosed, the method may include the following steps:
s101, generating a countermeasure network generation carrier video according to a video generated by pre-training;
when the video is required to be used as a carrier for information steganography, firstly, a countermeasure network is generated through the video to generate a carrier video, wherein the generated carrier video is a pseudo video which is generated by the network and accords with natural semantics. Wherein, the video generation countermeasure network is generated by pre-training.
Specifically, the pre-training generated video generation countermeasure network includes: and the video generator is composed of a foreground generator and a background generator. Generating a foreground and a mask by taking the noise as input through a foreground generator; and generating the background by using the noise as an input through a background generator. The generated foreground, background and mask form a carrier video.
Compared with the prior art, the prior art is a single-stream generation model, a video sample is directly generated, the double-stream generation model is adopted in the embodiment, the generated video model comprises a foreground generator and a background generator, the foreground generator generates a foreground and a space-time mask, the background generator generates a background, the foreground is motion information, and the background is static information. And synthesizing the foreground and background information into a complete video through the space-time mask information. Compared with a video generated by a single-flow model, the double-flow model is more flexible to use; in addition, in order to increase the resolution of the video and improve the visual quality and the embedding capacity, on the basis of the existing double-stream video generation network, the number of deconvolution layers of the foreground generator and the background generator is increased by one layer, the resolution is from 64 × 64 to 128 × 128, and the size of a convolution kernel is set to be 3 × 3 × 3, so that the method is more suitable for data processing.
S102, generating a modified probability matrix according to a hidden-writing generation countermeasure network generated by pre-training based on a carrier video;
and then, taking the carrier video as the input of the hidden-writen generation countermeasure network, and generating a modified probability matrix through the hidden-writen generation countermeasure network.
Specifically, the steganography generation countermeasure network includes: a steganography generator; and generating a modified probability matrix by using the foreground in the carrier video as an input through a steganography generator.
Compared with the prior art, the steganography generation countermeasure network is responsible for generating the embedded modification probability matrix and is called a steganography generator. Unlike the video steganography network, the video stream mainly contains two types of information, including temporal information in addition to spatial information. In the image spatial steganography, spatial information of a single carrier image is mostly analyzed, and then modification probability is generated for steganography. Video information is often continuous sequence information, and both a static space and a motion space contain a large amount of redundant information for embedding information, but video data has larger data volume and a feature extraction network is not easy to train compared with images. Therefore, in the framework provided by this embodiment, the foreground generated by the video generation countermeasure network is directly used as the input of the steganography generator, and the feature information of the foreground is extracted by using the convolution network. The two-dimensional convolutional layer of the traditional processing diagram is expanded into a three-dimensional convolutional layer, and the method is suitable for processing video data, and the designed convolutional kernel is 1 multiplied by 5.
S103, adaptively generating an optimal embedding modification position diagram by the modification probability matrix through an optimal binary embedding function;
after the modification probability matrix is generated, the optimal embedded modification position graph is further generated in a self-adaptive mode through the optimal binary embedded function.
And S104, embedding secret information in the position corresponding to the optimal embedding modification position diagram of the carrier video to generate a secret-containing video.
And then, modifying and embedding the secret information into the lowest bit of the pixel at the corresponding position of each channel of the video frame in the carrier video by using the optimal embedded modification position map to obtain the secret-contained video.
Further, the sender can transmit the video containing the secret to the receiver through a public channel, and transmit the embedded modified position map to the receiver through a secret channel. And after receiving the video with the secret, the receiver takes the modified position diagram transmitted by the video sender through the secret channel as an extraction reference, and extracts the secret information of the lowest bit of the pixel at the corresponding position in the video. The specific extraction process is that the modification position map is a matrix map with the same size as the video pixel, each modification position marked on the modification position map is embedded with information, and the specific position of the information is a value on the lowest bit of the marked pixel.
In particular, in the above embodiment, as shown in fig. 2, the input to the video generator is a low dimensional noise, which in most cases may be sampled from a distribution, such as a gaussian distribution. As shown in fig. 2, the foreground generator includes a four-dimensional tensor (tensor), as represented by 2 × 4(512), the time dimension is 2, the row dimension is 4, the column dimension is 4, and the number of channels is 512. The transposed three-dimensional convolutional layer is arranged between the square blocks, the size of the convolutional core is 3 multiplied by 3, S is 2, the step size of the convolutional layer is 2, the background generator can be explained by the same principle, but the two-dimensional convolutional layer is arranged between the square blocks of the background generator. At the end of the last convolutional layer of the dual-stream generator, one layer of activation function is added to output of the convolutional layer, and Tanh and Sigmoldo in the figure are two different activation functions to control the output of the whole video generator. (in order to increase the resolution of the video and improve the embedding capacity, the number of deconvolution layers of the foreground generator and the background generator can be increased by one more layer, and the resolution is from 64 × 64 to 128 × 128). the principle of the video generation flow design in the dual-stream architecture is as follows:
G(z)＝m(z)⊙f(z)+(1-m(z))⊙b(z)
wherein £ is Hadamard product, m (z) can be regarded as a spatio-temporal mask, and f (z) and b (z) represent foreground (foreground) information and background (background) information, respectively. And selecting foreground information or background information by utilizing a space-time mask according to the position and the time step of each pixel in the video, wherein the foreground represents information of motion in the video, and the background represents a static background in the video. In order to generate background information in a video time sequence, a background learns the static pixel information in the video image through two-dimensional convolution, and generates a plane image copied with time. The foreground here is a four-dimensional space-time tensor representing the space-time information for each pixel.
When training the generated video generation countermeasure network, the embodiment uses a 5-layer space-time 3D convolutional network as the discriminator of the video generation countermeasure network, with a convolution kernel of 3 × 3 × 3 and a step size of 2. Thus, the convolutional layer can learn the statistical information in the video background and also can learn the space-time relationship of the motion of the object. The discriminator designed in this embodiment is opposite to the structure of the convolutional layer in the video generator, which generates the foreground, and the three-dimensional convolution replaces the three-dimensional transpose convolution to process the video features. The input of the discriminator is a real video sample and a generated non-secret-carrying sample, and the output is a class label and a classification probability logit, when the class label is 1, the real video sample is represented; when the label is 0, the generated video sample is represented, and the classification probability logit is a numerical value between 0 and 1. Except that the sidmiod function is used after the last convolutional layer in the discriminator network, LeakyReLU is used as an activation function after the convolution operation of each layer in the first four layers for processing. loss is a loss function of the discriminator, sigmoid cross entropy is adopted for training the discriminator, and loss represents a specific performance index of the video generation network. The specific formula is as follows:
loss＝-[y*ln(p)+(1-y)ln(1-p)]
the method takes the foreground information generated by the video generation network as the input of the steganography generator in the pre-training generation countermeasure network, avoids the processing of the steganography countermeasure network on the whole video data, and effectively solves the problem of overlarge video data volume.
Specifically, in the above embodiment, a generator suitable for video information embedding is constructed in the steganography generation countermeasure network, and a two-dimensional convolution layer based on an image is expanded into a three-dimensional convolution layer to realize information embedding on a video carrier, and the generator includes a set of steganography generator and steganography discriminator to generate an optimal modification probability matrix. The steganography generator structure is shown in table 1:
TABLE 1 steganography generator structure
As shown in table 1, the input layer first converts the pixel values of the input video from [0: 255 interval is transformed to [0:1], (pixel values are generally distributed in the interval from 0 to 255, normalization processing is performed for convenience of data processing, and the whole is transformed to 0 to 1), 3DConv represents a three-dimensional convolutional layer, 3Ddeconv represents a transposed three-dimensional convolutional layer, Concat (L7) represents that the output of the layer is spliced with the output of the 7 th layer in the last dimension, for example, the ninth layer has an output structure of (n, h/2, w/2,128), and the input of the tenth layer is changed to (n, h/2, w/2,256) after Concat operation. Without the convolutional layer operated by Concat, the output of the previous layer is directly used as the input of the next layer. In the framework, a total of 16 three-dimensional convolutional layers are included, each layer of the first 8 layers comprises a three-dimensional convolutional layer with a 1 × 5 × 5 convolutional kernel, the next layer is a BN layer, the first 8 groups adopt a Leaky-ReLU activation function, the last 8 groups are a ReLU activation function and a transposed three-dimensional convolutional layer, the output of the last group uses a sigmoid activation function, the feature extracted by the convolutional network is mapped by using the sigmoid function and is converted into the probability with the value in the interval of 0 to 1, in order to prevent the safety of the embedded message from being reduced due to the overlarge modification probability, 0.5 is subtracted from each probability, and the interval is controlled in the range of 0 to 0.5. In Table 1, n, h, and w represent the frame number, height, and width of the video, respectively, h/2 represents half the height of the input video, the rest w/4, h/4 …, and so on.
When training to generate the pair reactance network, the steganography discriminator is used for analyzing the correlation and the reactance detection capability between each channel in the color steganography sample. After game countermeasure training, the steganography generator generates a secret-carrying sample corresponding to the modified probability matrix, and the capacity of resisting steganography analysis and detection is enhanced.
The embodiment designs a steganographic discriminator based on three-dimensional convolution. As shown in fig. 3, in consideration of the space-time dimension and the number of channels of the video signal, in order to use the video signal for analysis and detection of a color three-channel dense video frame, the embodiment expands the high-pass filter in the time dimension and the channel dimension, and respectively preprocesses three channel samples (a video frame image is composed of a red channel, a green channel, and a blue channel) of an image of each frame through the high-pass filter, so that feature information of each channel can be respectively processed and correlations among the three channels can be analyzed, the original high-pass filter is 6 convolution kernels of 5 × 5, and the convolution kernels composed of the original high-pass filter and the original high-pass filter are input into three channels of each frame image to be three-dimensionally convolved to preprocess data, and the convolution kernels are respectively the first to six convolution kernels from left to right in the following formula:
finally, combining the signals into a feature map with the channel number of 18 and the time dimension of 32 so as to process continuous video signals, inputting the obtained residual features into two 6-layer three-dimensional convolution networks to analyze the space-time features of the signals, wherein one network structure is responsible for extracting the accurate features of the channel correlation in the secret-carrying video and judging the difference of the channel correlation between the secret-carrying video and the original video image, so that embedding capacity is distributed in three channels of each frame of image, and the other network structure is responsible for extracting the related information between the frames of the secret-carrying video, further promoting a steganographic generator to reasonably embed the information in each frame of image of a video sequence, and judging whether secret information is embedded in the input. The size of the three-dimensional convolution kernel of the first convolution network is 1 × 5 × 5, and the size of the three-dimensional convolution kernel of the second convolution network is 5 × 5 × 5. The convolution step of the two is set to be 1, and the number of convolution kernels is as follows: the first layer is 16, the second layer is 32, the third layer is 64, the fourth layer is 128, the fifth layer is 256, and the sixth layer is 512. The convolution output result is processed by a batch normalization layer and an activation function, the activation functions of the first two layers are set as tanh activation functions, the activation functions of the third layer to the fifth layer are used, the activation functions are not adopted in the last layer, the outputs of the two convolution networks are combined together in the last dimension, and finally the space-time characteristics are mapped into the recognition probability value by using a full connection layer.
In the two stages of training, two antagonistic networks are respectively trained, the discriminators of the two antagonistic networks have difference in structural design, but the label information output by the networks is taken as the key reference information for measuring the performance of the discriminators, and the loss of the steganographic discriminators is defined as:
in the above loss function, y' is defined as an output of an activation function in the steganography discriminator, which is an identification probability of the steganography discriminator on an input sample, and a value is between 0 and 1. y isiThe steganographic arbiter classifies the carrier samples and the embedded samples into labels, wherein the label value of 1 represents the carrier samples, and the label value of 0 represents the embedded samples.
For the steganography generator's loss function, it is defined as:
lG＝-α×lD+β×(C-3×N×H×W×Q)2
C＝C1+C2+C3
wherein lDRepresenting the loss of the steganographic arbiter, NxHxWxQ is the expected load set before training, N is the video frame number, H is the video height, W is the video width, Q is the channel embedding rate, which is the average embedded secret message ratio per channelThe number of the bits.
Different from the image steganography network, the network uses a three-dimensional convolution network, so the set embedding capacity is distributed in the whole video sequence, the embedding capacity of the steganography generator in each frame of image in the video sequence is automatically adjusted according to the spatial domain characteristic information of the steganography generator through the learning of the network, and the number of secret message bits required to be embedded in the whole video is averagely distributed for 32 frames of video images.
In the two-stage training, the first stage sets β to 0 and α to a constant of 1; in the next stage, the parameter beta is set to 10 in order to ensure the optimization of the objective function by reasonably setting the actual training requirement-7。CkTo embed the secret message payload in the three channels of samples, it is defined as:
in the above formula pi,j,kRefers to the steganographic generator generating the corresponding pixel xi,jThe modified probability value for the k-th channel,
the probability of the embedding value being +1, -1, 0, respectively, +1, -1 corresponds to the probability of the embedding method being one plus or one minus, and 0 represents no embedding.
After the video steganography confrontation modifies the probability matrix, a corresponding embedded modification position graph is needed to be obtained to embed the secret message, and the embedded modification position graph is generated by an optimal embedded simulator as follows.
Wherein, i, j, k, c respectively represent the ith row and jth column pixel, p in the c channel in the image of the kth framec i,j,kEmbedding probabilities into a modified probability matrix generated for a steganography generator; n isc i,j,kRandom numbers generated for even distribution between 0 and 1;
mc i,j,kfor the modification strategy, when the value is 0, the lowest bit of the pixel point is not modified, the pixel is skipped when the secret message is embedded, when the value is not 0, the lowest bit of the pixel is compared with the message bit, and if the value is the same, the pixel is not modified; if different, in accordance with mc i,j,kThe lowest order bit of the pixel is modified, i.e. mc i,j,kWhen the value is 1, the pixel is increased by one; m isc i,j,kAt-1, the pixel is decremented by one.
But the function can not generate the gradient back propagation transfer gradient of the network in the actual training, so that the training time is too long. In order to solve the problem of discontinuous embedding function, an optimal embedding activation function based on a tanh function is introduced as follows:
tan h is the hyperbolic tangent function:
in the formula, lambda is a scaling factor, the gradient of the function changing in the step-shaped state is controlled, and different scaling factors correspond to the change of the function.
The modified position graph is formed by a modified strategy, when the modified strategy is 0, the pixel of the dense video is not embedded with the message, and the bit is skipped during extraction; when the modification strategy is not 0, the lowest bit of the pixel of the secret video is the secret message bit.
The training for generating the countermeasure network in the model training is divided into two stages, firstly, a first video is trained to generate the countermeasure network so that a pseudo video which accords with natural semantics can be generated, after 3000 rounds of iterative training, a second generation network is trained to be responsible for generating the embedded modification matrix, and the training is carried out for 800 times. For the training in the first network, training of the arbiter is performed after the generator is trained once. In the training, an Adam optimizer with a learning rate of 0.0002 is used for training the model, and the optimizer is responsible for adjusting parameters in the neural network so as to optimize the output of the neural network and minimize a loss function. Each batch of video samples was trained 3800 times for a total of 500 epochs.
In conclusion, the invention can effectively solve the problems that the existing network can not directly process the video carrier with large data volume and the optimization modification strategy in the traditional video information hiding algorithm is difficult to construct, and effectively realizes the information hiding by taking the video as the carrier.
As shown in fig. 4, a schematic structural diagram of an embodiment of an information steganography apparatus based on a video carrier according to the present invention may include:
pre-training the generated video to generate a countermeasure network 401 for generating a carrier video;
generating a countermeasure network 402 by pre-training generated steganography, and generating a modified probability matrix based on a carrier video;
generating a countermeasure network 402 by pre-training generated steganography, and adaptively generating an optimal embedded modification position diagram by the modification probability matrix through an optimal binary embedded function;
the generated steganography is trained in advance to generate a countermeasure network 402, and the generated steganography is further used for embedding secret information in a position corresponding to the optimal embedding modification position diagram of the carrier video to generate a dense video.
The working principle of the information steganography apparatus based on a video carrier disclosed in this embodiment is the same as that of the above-mentioned information steganography method based on a video carrier, and is not described herein again.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. The device disclosed by the embodiment corresponds to the method disclosed by the embodiment, so that the description is simple, and the relevant points can be referred to the method part for description.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A method of steganography of information based on a video carrier, comprising:
generating a modified probability matrix for a countermeasure network according to steganography generated by pre-training based on a carrier video, wherein the carrier video is a pseudo video which is generated by a network and accords with natural semantics;
the modification probability matrix is subjected to self-adaptive generation of an optimal embedding modification position diagram through an optimal binary embedding function;
and embedding secret information at the position corresponding to the optimal embedding modification position diagram of the carrier video to generate a secret video.
2. The method of claim 1, further comprising, before generating the modified probability matrix for the reactive network based on the carrier video according to the pre-trained steganography generation, the method further comprising:
and generating a carrier video according to the video generated by the pre-training.
3. The method of claim 2, wherein the generating the countermeasure network generation carrier video from the pre-trained generated video comprises:
and taking noise as the input of the video generation countermeasure network generated by the pre-training to generate the carrier video consisting of the foreground, the background and the mask.
4. The method of claim 3, wherein the generating the carrier video with foreground, background and mask by using noise as an input of the pre-training generated video generation countermeasure network comprises:
and taking noise as an input of the video generation countermeasure network generated by the pre-training, generating the foreground and the mask through a foreground generator in the video generation countermeasure network generated by the pre-training, and generating the background through a background generator in the video generation countermeasure network generated by the pre-training, wherein the carrier video consisting of the foreground, the background and the mask.
5. The method of claim 4, wherein the generating the modified probability matrix for the reactive network according to the pre-trained steganography based on the carrier video comprises:
and taking the foreground in the carrier video as the input of the pre-generated steganography generation countermeasure network, and generating a modified probability matrix through a steganography generator in the pre-generated steganography generation countermeasure network.
6. An information steganography apparatus based on a video carrier, comprising:
the method comprises the steps that a pair of rejection networks are generated through pre-training, and the pair of rejection networks are used for generating a modified probability matrix based on a carrier video, wherein the carrier video is a pseudo video which is generated by the network and accords with natural semantics;
the hidden writing generated by pre-training generates a countermeasure network, and is also used for adaptively generating an optimal embedded modification position diagram through an optimal binary embedding function by the modification probability matrix;
and the steganography generated by the pre-training generates a countermeasure network, and is also used for embedding secret information in the position corresponding to the optimal embedding modification position diagram of the carrier video to generate a dense video.
7. The apparatus of claim 6, further comprising:
and training the generated video in advance to generate a countermeasure network for generating the carrier video.
8. The apparatus of claim 7, wherein the pre-training generated video generation countermeasure network is specifically configured to:
and generating the carrier video consisting of the foreground, the background and the mask by taking the noise as an input.
9. The apparatus of claim 8, wherein the pre-training generated video generation countermeasure network comprises: a foreground generator and a background generator; wherein:
the foreground generator is used for generating the foreground and the mask by taking noise as input;
and the background generator is used for generating the background by taking noise as input, wherein the carrier video is composed of the foreground, the background and a mask.
10. The apparatus of claim 9, wherein the pre-training generated steganographic generation countermeasure network comprises: a steganography generator; wherein:
the steganography generator is used for generating a modified probability matrix by taking the foreground in the carrier video as input.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111626840.5A CN114339258B (en) | 2021-12-28 | 2021-12-28 | Information steganography method and device based on video carrier |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111626840.5A CN114339258B (en) | 2021-12-28 | 2021-12-28 | Information steganography method and device based on video carrier |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114339258A true CN114339258A (en) | 2022-04-12 |
| CN114339258B CN114339258B (en) | 2024-05-10 |
Family
ID=81015791
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111626840.5A Active CN114339258B (en) | 2021-12-28 | 2021-12-28 | Information steganography method and device based on video carrier |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114339258B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114900586A (en) * | 2022-04-28 | 2022-08-12 | 中国人民武装警察部队工程大学 | Information steganography method and device based on DCGAN |
| CN115086503A (en) * | 2022-05-25 | 2022-09-20 | 清华大学深圳国际研究生院 | Information hiding method, device, equipment and storage medium |
| CN115348360A (en) * | 2022-08-11 | 2022-11-15 | 国家电网有限公司大数据中心 | Self-adaptive embedded digital label information hiding method based on GAN |
| CN117714712A (en) * | 2024-02-01 | 2024-03-15 | 浙江华创视讯科技有限公司 | Data steganography method, equipment and storage medium for video conference |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104010193A (en) * | 2014-05-29 | 2014-08-27 | 中国科学院信息工程研究所 | Video steganographic method based on macro block partition mode disturbance |
| CN104519361A (en) * | 2014-12-12 | 2015-04-15 | 天津大学 | Video steganography analysis method based on space-time domain local binary pattern |
| CN104837011A (en) * | 2015-05-04 | 2015-08-12 | 中国科学院信息工程研究所 | Content self-adaptive video steganalysis method |
| CN104853215A (en) * | 2015-04-17 | 2015-08-19 | 中国科学院信息工程研究所 | Video steganography method based on motion vector local optimality preservation |
| CN107563155A (en) * | 2017-08-08 | 2018-01-09 | 中国科学院信息工程研究所 | A kind of safe steganography method and device based on generation confrontation network |
| CN110188706A (en) * | 2019-06-03 | 2019-08-30 | 南京邮电大学 | Neural network training method and detection method based on facial expression in the video for generating confrontation network |
| CN111768326A (en) * | 2020-04-03 | 2020-10-13 | 南京信息工程大学 | High-capacity data protection method based on GAN amplification image foreground object |
| CN111860162A (en) * | 2020-06-17 | 2020-10-30 | 上海交通大学 | Video crowd counting system and method |
| US20210150769A1 (en) * | 2019-11-14 | 2021-05-20 | Qualcomm Incorporated | High efficiency image and video compression and decompression |
| CN113242469A (en) * | 2021-04-21 | 2021-08-10 | 南京大学 | Self-adaptive video transmission configuration method and system |
-
2021
- 2021-12-28 CN CN202111626840.5A patent/CN114339258B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104010193A (en) * | 2014-05-29 | 2014-08-27 | 中国科学院信息工程研究所 | Video steganographic method based on macro block partition mode disturbance |
| CN104519361A (en) * | 2014-12-12 | 2015-04-15 | 天津大学 | Video steganography analysis method based on space-time domain local binary pattern |
| CN104853215A (en) * | 2015-04-17 | 2015-08-19 | 中国科学院信息工程研究所 | Video steganography method based on motion vector local optimality preservation |
| CN104837011A (en) * | 2015-05-04 | 2015-08-12 | 中国科学院信息工程研究所 | Content self-adaptive video steganalysis method |
| CN107563155A (en) * | 2017-08-08 | 2018-01-09 | 中国科学院信息工程研究所 | A kind of safe steganography method and device based on generation confrontation network |
| CN110188706A (en) * | 2019-06-03 | 2019-08-30 | 南京邮电大学 | Neural network training method and detection method based on facial expression in the video for generating confrontation network |
| US20210150769A1 (en) * | 2019-11-14 | 2021-05-20 | Qualcomm Incorporated | High efficiency image and video compression and decompression |
| CN111768326A (en) * | 2020-04-03 | 2020-10-13 | 南京信息工程大学 | High-capacity data protection method based on GAN amplification image foreground object |
| CN111860162A (en) * | 2020-06-17 | 2020-10-30 | 上海交通大学 | Video crowd counting system and method |
| CN113242469A (en) * | 2021-04-21 | 2021-08-10 | 南京大学 | Self-adaptive video transmission configuration method and system |
Non-Patent Citations (2)
| Title |
|---|
| S. THENMOZHI, ET AL: "Information Security Using Image Steganography Based on Generative Adversarial Networks", 《2021 THIRD INTERNATIONAL CONFERENCE ON INVENTIVE RESEARCH IN COMPUTING APPLICATIONS (ICIRCA)》 * |
| 王耀杰等: "基于生成对抗网络的图像隐藏方案", 《信息网络安全》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114900586A (en) * | 2022-04-28 | 2022-08-12 | 中国人民武装警察部队工程大学 | Information steganography method and device based on DCGAN |
| CN114900586B (en) * | 2022-04-28 | 2024-04-16 | 中国人民武装警察部队工程大学 | Information steganography method and device based on DCGAN |
| CN115086503A (en) * | 2022-05-25 | 2022-09-20 | 清华大学深圳国际研究生院 | Information hiding method, device, equipment and storage medium |
| CN115086503B (en) * | 2022-05-25 | 2023-09-22 | 清华大学深圳国际研究生院 | Information hiding method, device, equipment and storage medium |
| CN115348360A (en) * | 2022-08-11 | 2022-11-15 | 国家电网有限公司大数据中心 | Self-adaptive embedded digital label information hiding method based on GAN |
| CN115348360B (en) * | 2022-08-11 | 2023-11-07 | 国家电网有限公司大数据中心 | GAN-based self-adaptive embedded digital tag information hiding method |
| CN117714712A (en) * | 2024-02-01 | 2024-03-15 | 浙江华创视讯科技有限公司 | Data steganography method, equipment and storage medium for video conference |
| CN117714712B (en) * | 2024-02-01 | 2024-05-07 | 浙江华创视讯科技有限公司 | Data steganography method, equipment and storage medium for video conference |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114339258B (en) | 2024-05-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112288658B (en) | Underwater image enhancement method based on multi-residual joint learning | |
| CN114339258A (en) | Information steganography method and device based on video carrier | |
| Barni et al. | CNN detection of GAN-generated face images based on cross-band co-occurrences analysis | |
| CN108764270B (en) | Information hiding detection method integrated by using convolutional neural network | |
| Ubhi et al. | Neural style transfer for image within images and conditional GANs for destylization | |
| CN112634117B (en) | End-to-end JPEG domain image steganography method based on generation of countermeasure network | |
| CN109816612A (en) | Image enchancing method and device, computer readable storage medium | |
| CN111681155B (en) | GIF dynamic image watermarking method based on deep learning | |
| CN115378574B (en) | Lightweight dynamic image data encryption method and system | |
| CN112651917A (en) | Space satellite low-illumination image enhancement method based on generation countermeasure network | |
| Hsu et al. | A high-capacity QRD-based blind color image watermarking algorithm incorporated with AI technologies | |
| CN115131188A (en) | Robust image watermarking method based on generation countermeasure network | |
| Wang et al. | A survey on digital image steganography | |
| CN111726472B (en) | Image anti-interference method based on encryption algorithm | |
| CN111768326A (en) | High-capacity data protection method based on GAN amplification image foreground object | |
| CN113379606B (en) | Face super-resolution method based on pre-training generation model | |
| CN114900586B (en) | Information steganography method and device based on DCGAN | |
| CN109729070B (en) | Detection method of network heterogeneous concurrent steganography channel based on CNN and RNN fusion model | |
| Wang et al. | Data hiding during image processing using capsule networks | |
| CN117391920A (en) | High-capacity steganography method and system based on RGB channel differential plane | |
| Revi et al. | Gan-generated fake face image detection using opponent color local binary pattern and deep learning technique | |
| CN117131520A (en) | Two-stage image privacy protection method and system based on dynamic mask and generation recovery | |
| Babaheidarian et al. | Decode and transfer: A new steganalysis technique via conditional generative adversarial networks | |
| Kumari et al. | Optimized dwt based digital image watermarking and extraction using rnn-lstm | |
| CN113436094A (en) | Gray level image automatic coloring method based on multi-view attention mechanism |
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 |