CN114554223B - Concealed communication method based on video frame offset coding - Google Patents
Concealed communication method based on video frame offset coding Download PDFInfo
- Publication number
- CN114554223B CN114554223B CN202210101177.5A CN202210101177A CN114554223B CN 114554223 B CN114554223 B CN 114554223B CN 202210101177 A CN202210101177 A CN 202210101177A CN 114554223 B CN114554223 B CN 114554223B
- Authority
- CN
- China
- Prior art keywords
- magnetic field
- video
- field signal
- data
- bit
- 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.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/46—Embedding additional information in the video signal during the compression process
- H04N19/467—Embedding additional information in the video signal during the compression process characterised by the embedded information being invisible, e.g. watermarking
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention discloses a covert communication method based on video frame offset coding, which utilizes a magnetic field leaked when a CPU decodes a video to transmit data. The time of decoding the corresponding frame during video playing is changed by changing the timestamp information of some frames in the video, so that the time of the occurrence of the peak of the leakage magnetic field during the working of the CPU is changed, namely the energy distribution of the magnetic field is changed. The normal magnetic field signal segment generated by the non-offset frame is coded as 1, the abnormal magnetic field signal segment generated by the offset frame is coded as 0, and the steganographic information is leaked out by analyzing the energy distribution of the magnetic field. The method of the invention does not need to obtain video files for reading the video steganography information, only needs to use a magnetic sensor with high sampling rate to collect magnetic signals released when a CPU decodes the video when the video is played, analyzes the energy distribution characteristics of a magnetic field, recovers the content of the video steganography information from the magnetic signals, and can realize higher data transmission rate and hardly influence the size and quality of the video so as to ensure the transmission concealment.
Description
Technical Field
The invention belongs to the technical field of communication, relates to video steganography and magnetic field communication technology, and particularly relates to a covert communication method based on video frame offset coding.
Background
Reading of the current video steganography technology generally requires directly obtaining a video file, and analyzing the video file to obtain steganography information. But in some important places, the internal network is usually secured in a physical isolation manner to prevent the leakage of sensitive information. If the equipment is prohibited from being connected with the wireless network, the internal private network is physically isolated from the Internet of things, and the visitor is prohibited from taking pictures, recording and shooting pictures and operating sensitive equipment. In this case, a video file cannot be obtained, and thus reading the content encoded in the video through an electromagnetic or the like side channel is an important research direction. In the prior art, a related technology for deducing video coding content through a magnetic field signal leaked by a CPU (central processing unit) is available, but the technology is only suitable for occasions with low transmission rate and insensitivity to video size and quality, and how to design a new video steganography method to meet the requirement of high transmission rate and have no obvious influence on video size and quality to meet the requirement of covert transmission is an urgent problem to be solved.
Disclosure of Invention
The present invention aims to provide a covert communication method based on video frame offset coding, aiming at the defects of the prior art. The method utilizes a magnetic field leaked when a CPU works to transmit data, changes the playing time of a corresponding frame when a video is played by changing timestamp information of certain frames in the video, further changes the time when a peak of a magnetic field signal leaked when the CPU works, namely, changes the energy distribution of the magnetic field signal, collects the magnetic field signal released when the CPU decodes the video through a magnetic sensor with a high sampling rate, and analyzes the energy distribution of the magnetic field signal to leak steganographic information. The information reading of the existing video steganography method depends on the source file of the video, or the transmission rate is not high and the video quality is influenced. The method of the invention does not depend on the source file of the video, and utilizes the magnetic field signal generated when the CPU decodes the video to transmit the data, so that the transmission rate is greatly improved, the size of the video file is hardly influenced, the video quality can be ensured, and the transmission rate and the transmission concealment can be better ensured. The method greatly improves the rate and the concealment of the video steganography information reading, and can be used for the concealed communication in the scene with network physical isolation and higher transmission rate and transmission concealment requirements.
The invention discloses a covert communication method based on video frame offset coding, which comprises the following steps:
step 1: preprocessing data to be transmitted, firstly adopting a proper character coding set to convert the transmitted data into a binary sequence, and then adding a forward error correction code and a synchronous sequence, wherein the specific steps are as follows:
step 1.1: coding the data to be transmitted into a binary sequence M by adopting a proper character coding set;
step 1.2: adding Hamming forward error correction code to binary sequence M, where the check code length is r, and each check packet length is n =2 r -1, wherein the primary code length is k = n-r =2 r -1-r; if it is the most importantIf the original code of the next group is less than k characters, the characters are filled by 0, and the sequence after adding Hamming forward error correction code is recorded as M r ；
Step 1.3: will M r Cut into a series of fragments of length m
Adding length m at the beginning of each segment p Of (2) a synchronization sequence M p Get->
The synchronous sequence consists of 0 and 1 and is used for a synchronization and training decision device at a receiving end;
step 2: according to the fact that each bit of data in a binary system sequence to be transmitted corresponds to one video frame, the video frame corresponding to 1 is unchanged in appearance time, and the video frame corresponding to 0 is delayed by delta t The method comprises the following steps of generating timestamp information of each frame of an original video for video steganography, and then adjusting the occurrence time of each frame of the video by using a generated timestamp file, wherein the specific steps are as follows:
step 2.1: encode 0 and 1 using the principle: a video frame encodes a bit of data, if the binary number is 1, the corresponding video frame appears in the same time as the uncoded original video; if the binary number is 0, the corresponding video frame appears later by a certain time delta compared with the uncoded original video t . The amount of this lag time is determined by the video frame rate, typically 60 frames/s, which can be set to 3ms, with the remaining frame rates being scaled up or down. When the video is decoded, the moment of each frame corresponds to the peak value moment of the CPU leakage magnetic field, and the sampled magnetic field signal is divided into a plurality of magnetic field signals with the length of
The field signal segments of (1), where fps is the frame rate of the video, and each field signal segment represents a field generated by one frame of the video and also represents a field signal corresponding to a binary number of one bit in the data to be transmitted. Such asIf a proper starting point is selected for division, the peak of the magnetic field signal segment corresponding to 1 is ensured to appear at the middle position of the segment, the peak of the magnetic field of the segment corresponding to 0 appears at the two side positions of the segment, and 0 and 1 can be coded by utilizing the different magnetic field energy distribution of the two magnetic field segments. In particular, the beginning and the end of the video to be coded are left with T respectively before And T after Time is not coded and is left as a buffer. The middle part of the video is coded according to the principle that one frame corresponds to one binary bit, taking the nth bit of the data to be transmitted as an example: if the Nth bit is 1, the timestamp information written to the frame is ≥>
Same as when not encoded; if the Nth bit is 0, then the timestamp information written to the frame is ≥>
I.e. the frame is delayed by a time delta compared to the uncoded video t . And setting a time stamp for each frame of the original video in the same way to generate a time stamp file, wherein the time stamp file records the occurrence time information of each frame of the video.
Step 2.2: and (3) adjusting the occurrence time of each frame of the original video by using the timestamp file obtained in the step (2.1), wherein the occurrence time of each frame in the generated new video is the same as the specified time in the timestamp file.
And step 3: playing a video, placing a magnetic sensor with a high sampling rate near a CPU of playing equipment to collect a magnetic field signal, decoding the magnetic field signal, and recovering steganography information; the method comprises the following specific steps:
step 3.1: collecting;
when the video is played, a magnetic sensor with a high sampling rate is arranged near a CPU of the playing equipment to collect magnetic field signals, and the collected data is recorded to a local file through a signal collection card;
step 3.2: detecting a synchronous signal and finding out the position where the encoding starts;
after the file header information of the collected data is removed, the video is obtainedThe time-varying curve of the magnetic field signal leaked by the CPU during playing. The synchronization sequence M is utilized in the following manner based on sliding window matching p Finding the position where to start encoding data: firstly, by observing the obtained magnetic field waveform, the magnetic field signal segment corresponding to the code 1 can be found, the total energy of the signal is larger due to the existence of the peak of the magnetic field signal, and conversely, the total energy of the magnetic field signal segment of the code 0 is lower. Therefore, each magnetic field signal segment encoding one bit of data is mapped to a total energy value capable of representing the segment, so that the dense approximately continuous magnetic field signals obtained at the original high sampling rate are discretized into an energy value capable of representing the magnetic field signal segment corresponding to one encoded bit. Setting a synchronization sequence M p Is p, the sampling rate of the magnetic sensor is f s The frame rate of the played video is f r Then the total number of sampling points corresponding to the p-bit binary number is
This is the size of a sliding window over the magnetic field signal, after which the corresponding->
K values are uniformly sampled from the sampling points and summed as the total energy value of the magnetic field signal segment, which in turn maps a sliding window to a p-dimensional eigenvector E pi Feature vector E of the sliding window pi And a synchronization sequence M p Performing cross-correlation operation, and recording the cross-correlation result as C i . The sliding window slides on the whole magnetic field signal to sample in sequence, the calculation is repeated, and finally all the sliding windows and the synchronous sequence M are obtained p Array C of cross-correlation results. Note the overlap t between the sliding windows op Small enough to ensure accuracy of synchronisation, e.g. by taking t directly op =1. After the cross-correlation results of all sliding windows are obtained, the best cross-correlation result | C max The sliding window is considered as the position of the synchronization sequence, and the start time of the sliding window is the start time of the synchronization sequence in the signal, which is recorded as ^ greater than or equal to>
Step 3.3: slicing the magnetic field signal;
the step is to slice the sampled magnetic field signal from the position where the coded data starts, and each magnetic field signal segment obtained after slicing represents the magnetic field signal generated during the playing period of a video frame and also encodes the magnetic field signal corresponding to one data bit. Let m be the number of data bits to be transmitted and f be the sampling rate of the magnetic sensor s The frame rate of the played video is f r The sampling result is represented as data, and the time when the encoded data starts in the signal is
The sampled encoded magnetic field signal is divided into m slices, and each slice contains a number of sampling points ^ 4>
The ith data bit is divided into
(data[A，B]Representing an array of data between a and B), in such a way that the magnetic field signal segments corresponding to each of all m bits of data to be transmitted are obtained.
Step 3.4: a training decision device;
firstly, sampling each magnetic field signal segment obtained in step 3.3, and setting data of n points uniformly sampled on each magnetic field signal segment, wherein the total number of points of each magnetic field signal segment is known as
The sampling interval of this round of sampling is then
The sampled n points can represent the variation trend of the whole magnetic field signal segment. At this time we have obtained an n-dimensional vector for each coded bit。
An SVM (support vector machine) classification model is trained using a known p-bit sync sequence. The training data is the known p-bit synchronization sequence and the magnetic field signal sampling result corresponding to each bit, the input of the model is the obtained n-dimensional vector corresponding to each encoding bit, and the output is 0 or 1, namely, a mapping from the energy distribution of the magnetic field segments to the encoding bits of 0 or 1 is established.
Step 3.5: decoding judgment;
starting from the next coded bit after the synchronization sequence, i.e. the decision starting time is
t d For the duration of the synchronous sequence, the judgment process is that the sampling data of the magnetic field segment corresponding to each coding bit is input into a trained SVM model, the model outputs whether the coding bit corresponding to the magnetic field segment is 0 or 1, all m magnetic field segments are processed in sequence, and the obtained judged sequence is recorded as being/>
Step 3.6: correcting errors by using a Hamming code;
a plurality of
Splicing of sequences into one->
A sequence for Hamming code error correction, the corrected sequence being recorded as->
Step 3.7: carrying out binary conversion;
binary sequence to be received
And reverting to the information.
The invention has the beneficial effects that:
the invention utilizes the magnetic field leaked when the CPU decodes the video to transmit data. The method comprises the steps of adjusting the time when a video frame appears by changing timestamp information of some frames in a video to encode data to be transmitted, collecting a magnetic field signal leaked by a CPU (Central processing Unit) when the video is played by using a magnetic sensor with a high sampling rate, analyzing the energy distribution of the collected magnetic field signal to read video steganography content, and being used for covert communication. Especially, the invention changes the energy distribution of the magnetic field leaked when the CPU decodes the video by changing the timestamp information of some frames in the video, and encodes and decodes 0 and 1 by two different energy distributions of magnetic field signals; compared with the traditional video steganography technology, the video steganography method has the advantages that the steganography and reading of information can be realized only by collecting and analyzing the magnetic field signal released when the CPU decodes the video without acquiring the played video source file, meanwhile, the transmission rate of data is obviously improved, the size and the quality of the video are hardly influenced, and the transmission concealment is ensured.
Drawings
Fig. 1 is a flow chart of the video steganography and reading of the present invention.
Detailed Description
The method of the present invention is further described below with reference to the accompanying drawings.
The invention provides a covert communication method based on video decoding electromagnetic leakage and utilizing video frame time offset coding. A new video steganography and reading method is designed, 0 and 1 are coded by using a normal condition that a video frame is not shifted and an abnormal condition that the video frame is shifted, magnetic field signals leaked when a CPU decodes a video can be caused by the two frame types and present two different energy distributions, and 0 and 1 are decoded according to the different energy distributions of the two magnetic field signal segments. According to the method, the magnetic field signal released when the CPU decodes the video is collected only through the magnetic sensor with a higher sampling rate, the data carried by the video is analyzed according to the magnetic field signal, a video source file is not required to be obtained, and meanwhile, higher data transmission rate and better transmission concealment are achieved. The flow chart of video steganography and reading is shown in fig. 1, and mainly comprises the processes of adding an error correcting code, adding a synchronization sequence, generating a timestamp file according to a transmitted binary sequence, adjusting the moment of each frame of a video by using the generated timestamp file, playing the video and acquiring magnetic signals, synchronizing the magnetic signals to find the initial position of coding, slicing, training a classifier, decoding by using the classifier, correcting errors by using the error correcting code and the like.
The method comprises the following steps:
step 1: the method comprises the following steps of preprocessing data to be transmitted, firstly adopting a proper character coding set to convert the transmitted data into a binary sequence, and then adding a forward error correction code and a synchronous sequence, wherein the specific steps are as follows:
step 1.1: coding the data to be transmitted into a binary sequence M by adopting a proper character coding set;
step 1.2: adding Hamming forward error correction code to binary sequence M, where the check code length is r, and each check packet length is n =2 r -1, wherein the primary code length is k = n-r =2 r -1-r; if the original code of the last group is less than k characters, 0 is used for completion, and the sequence after adding the Hamming forward error correction code is recorded as M r ；
Step 1.3: will M r Cutting into a series of fragments of length m
Adding length m at the beginning of each segment p Of (2) a synchronization sequence M p Get->
The synchronous sequence consists of 0 and 1 and is used for a synchronization and training decision device at a receiving end; />
And 2, step: according to the fact that each bit of data in a binary sequence to be transmitted corresponds to one video frame, the video frame corresponding to 1 is unchanged in appearance time, and the video frame corresponding to 0 is delayed by delta t The method comprises the following specific steps of generating time stamp information of each frame of an original video for video steganography, and then adjusting the time of each frame of the video by using a generated time stamp file:
step 2.1: the principle is employed to encode 0 and 1: a video frame encodes a bit of data, if the binary number is 1, the corresponding video frame appears in the same time as the uncoded original video; if the binary number is 0, the corresponding video frame appears later by a certain time delta compared with the uncoded original video t . When the video is decoded, the moment of each frame corresponds to the peak moment of the CPU leakage magnetic field, and the sampled magnetic field signal is divided into a plurality of magnetic field signals with the length of
Each magnetic field signal segment represents a magnetic field generated by one frame of video and also represents a magnetic field signal corresponding to one-bit binary number in data to be transmitted. If a proper starting point is selected for division, the peak of the magnetic field signal segment corresponding to 1 is ensured to appear at the middle position of the segment, the peak of the magnetic field of the segment corresponding to 0 appears at the two side positions of the segment, and 0 and 1 can be coded by utilizing the different magnetic field energy distribution of the two magnetic field segments. In particular, the beginning and the end of the video to be coded are respectively left with T before And T after Time is not encoded and is left for buffering. The middle part of the video is coded according to the principle that one frame corresponds to one binary bit, taking the nth bit of the data to be transmitted as an example: if the Nth bit is 1, the timestamp information written to the frame is ≥>
Same as when not encoded; if the Nth bit is 0, then the timestamp information written to the frame is ≥>
I.e. the frame is delayed by a time delta compared to the uncoded video t . And setting a time stamp for each frame of the original video in the same way to generate a time stamp file, wherein the time stamp file records the occurrence time information of each frame of the video.
Step 2.2: and (3) adjusting the time of each frame of the original video by using the timestamp file obtained in the step (2.1), wherein the time of each frame in the generated new video is the same as the time specified in the timestamp file.
And step 3: playing a video, placing a magnetic sensor with a high sampling rate near a CPU of playing equipment to collect a magnetic field signal, decoding the magnetic field signal, and recovering steganography information; the method comprises the following specific steps:
step 3.1: collecting;
when a video is played, a magnetic sensor with a high sampling rate is placed near a CPU of the playing equipment to collect a magnetic field signal, and the collected data is recorded to a local file through a signal acquisition card;
step 3.2: detecting a synchronous signal and finding out the position where the coding starts;
after the file header information of the collected data is removed, a curve of the time-varying magnetic field signal leaked by the CPU in the video playing period is obtained. Utilizing the synchronization sequence M in the following manner based on sliding window matching p Finding the position where to start encoding data: firstly, by observing the obtained magnetic field waveform, the magnetic field signal segment corresponding to the code 1 can be found, the total energy of the signal is larger due to the existence of the magnetic field signal peak, and conversely, the total energy of the magnetic field signal segment of the code 0 is lower. Therefore, it is considered that each magnetic field signal segment encoding one bit of data is mapped to a total energy value capable of representing the segment, so that the dense approximately continuous magnetic field signal obtained at the original high sampling rate is discretized into an energy value capable of representing the magnetic field signal segment corresponding to one encoded bit. Setting a synchronization sequence M p Has a length p and a sampling rate f of the magnetic sensor s The frame rate of the played video is f r Then the total number of sampling points corresponding to the p-bit binary number is
This is the size of a sliding window over the magnetic field signal, after which the corresponding &'s in each bit of data>
Uniformly sampling k values in each sampling point, and summingFor this purpose, the total energy value of the magnetic field signal segments is determined, and a sliding window is mapped into a p-dimensional characteristic vector E pi Feature vector E of the sliding window pi And a synchronization sequence M p Performing cross-correlation operation, and recording the cross-correlation result as C i . The sliding window slides on the whole magnetic field signal to sample in sequence, the calculation is repeated, and finally all the sliding windows and the synchronous sequence M are obtained p Array C of cross-correlation results. Note the overlap t between the sliding windows op Small enough to ensure accuracy of synchronisation, e.g. by taking t directly op And =1. After the cross-correlation results of all sliding windows are obtained, the best cross-correlation result | C max The sliding window is considered as the position of the synchronization sequence, and the start time of the sliding window is the start time of the synchronization sequence in the signal, which is recorded as ^ greater than or equal to>
Step 3.3: slicing the magnetic field signal;
the step divides the magnetic field signal obtained by sampling from the position of the start of coding data, and each magnetic field signal segment obtained after division represents the magnetic field signal generated during the playing period of a video frame and also encodes the magnetic field signal corresponding to one data bit. Let m be the number of data bits to be transmitted and f be the sampling rate of the magnetic sensor s The frame rate of the played video is f r The sampling result is expressed as data, and the moment when the coded data starts in the signal is
The sampled encoded magnetic field signal is divided into m slices, each slice containing a number of sample points &>
The slice corresponding to the ith data bit is
In such a way, magnetic field signals corresponding to all m bits of data to be transmitted are obtainedAnd (4) fragment.
Step 3.4: a training decision device;
firstly, sampling each magnetic field signal segment obtained in step 3.3, and setting data of n points uniformly sampled on each magnetic field signal segment, wherein the total number of points of each magnetic field signal segment is known as
The sampling interval of this round of sampling is
The sampled n points can represent the variation trend of the whole magnetic field signal segment. At this point we have obtained an n-dimensional vector for each coded bit.
An SVM (support vector machine) classification model is trained using the known p-bit sync sequence. The training data is the known p-bit synchronization sequence and the magnetic field signal sampling result corresponding to each bit, the input of the model is the obtained n-dimensional vector corresponding to each encoding bit, and the output is 0 or 1, namely, a mapping from the energy distribution of the magnetic field segments to the encoding bits of 0 or 1 is established.
Step 3.5: decoding judgment;
starting from the next coded bit after the synchronization sequence, i.e. the decision starting time is
t d For the duration of the synchronous sequence, the judgment process is that the sampling data of the magnetic field segment corresponding to each coding bit is input into a trained SVM model, the model outputs whether the coding bit corresponding to the magnetic field segment is 0 or 1, all m magnetic field segments are processed in sequence, and the obtained judged sequence is recorded as being/>
Step 3.6: correcting errors by using Hamming codes;
a plurality of
The sequence is spliced again into a->
Sequence, correcting error by Hamming code, and recording the corrected sequence as ^ er>
Step 3.7: carrying out binary conversion;
binary sequence to be received
Reverting to information.
The method of the invention does not need to obtain video files for reading the video steganography information, only needs to use a magnetic sensor with high sampling rate to collect magnetic signals released when a CPU decodes the video when the video is played, analyzes the energy distribution characteristics of a magnetic field, recovers the content of the video steganography information from the magnetic signals, and can realize higher data transmission rate and hardly influence the size and quality of the video so as to ensure the transmission concealment.
Claims (5)
1. A covert communication method based on video frame offset coding is characterized by comprising the following steps:
step 1: preprocessing data to be transmitted, firstly adopting a character coding set, converting the transmitted data into a binary sequence, and adding a forward error correction code and a synchronous sequence;
step 2: corresponding each bit of data in a binary sequence to be transmitted to a video frame, and delaying the video frame occurrence time corresponding to 1 by delta t Generating time stamp information of each frame of the original video for video steganography, and then adjusting the time of each frame of the video by using the generated time stamp file;
and 3, step 3: playing a video, placing a magnetic sensor near a CPU of playing equipment to collect a magnetic field signal, obtaining an approximately continuous sampling signal, decoding the magnetic field signal, recovering steganography information, and realizing covert communication; the method comprises the following specific steps:
step 3.1: collecting;
when the video is played, the magnetic sensor is arranged near a CPU of the playing equipment to collect a magnetic field signal, an approximately continuous sampling signal is obtained, and the collected data is recorded to a local file through a signal acquisition card;
step 3.2: detecting a synchronous signal and finding out the position where the coding starts;
after removing file header information from the acquired data, obtaining a curve of the magnetic field signal leaked by the CPU along with the change of time during video playing; by means of sliding window matching, the synchronous sequence M is utilized p Finding the position where the encoded data starts, i.e. the start time of the synchronization sequence
Step 3.3: slicing the magnetic field signal;
slicing the sampled magnetic field signal from the position of starting to encode data, and encoding a magnetic field signal corresponding to a data bit to obtain each magnetic field signal segment representing the magnetic field signal generated during playing of a video frame, wherein the number of data bits to be transmitted is m, and the sampling rate of the magnetic sensor is f s The frame rate of the played video is f r The sampling result is expressed as data, and the moment when the coded data starts in the signal is
The sampled encoded magnetic field signal is divided into m slices, each slice containing a number of sample points &>
The slice corresponding to the ith data bit is:
step 3.4: a training decision device;
firstly, sampling each magnetic field signal segment obtained in step 3.3, and setting data of n points uniformly sampled on each magnetic field signal segment, wherein the total number of points of each magnetic field signal segment is known as
The sampling interval of this round of sampling is then
N points obtained by sampling can represent the variation trend of the whole magnetic field signal segment, so that an n-dimensional vector corresponding to each coded bit is obtained;
training an SVM (support vector machine) classification model by using a known p-bit synchronization sequence, wherein training data is the known p-bit synchronization sequence and a magnetic field signal sampling result corresponding to each bit, the input of the model is an obtained n-dimensional vector corresponding to each encoding bit, and the output is 0 or 1, namely establishing a mapping from the energy distribution of the magnetic field segment to the encoding bit of 0 or 1;
step 3.5: decoding judgment;
starting with the next coded bit after the synchronization sequence, i.e. with a start time of decision
t d For the duration of the synchronization sequence, the judgment process is that the sampling data of the magnetic field segment corresponding to each coding bit is input into a trained SVM model, the model outputs whether the coding bit corresponding to the magnetic field segment is 0 or 1, all m magnetic field segments are sequentially processed, and the obtained judged sequence is recorded as->
Step 3.6: correcting errors by using Hamming codes;
a plurality of
The sequence is spliced again into a->
A sequence for Hamming code error correction, the corrected sequence being recorded as->
Step 3.7: carrying out binary conversion;
binary sequence to be received
Reverting to information.
2. The concealment communication method based on video frame offset coding according to claim 1, characterized in that step 1 is specifically:
step 1.1: encoding data to be transmitted into a binary sequence M by adopting a character encoding set;
step 1.2: adding Hamming forward error correction code to binary sequence M, where the check code length is r, and each check packet length is n =2 r -1, wherein the primary code length is k = n-r =2 r -1-r; if the original code of the last group is less than k characters, 0 is used for completion, and the sequence after adding the Hamming forward error correction code is recorded as M r ；
Step 1.3: will M r Cut into a series of fragments of length m
Adding length m at the beginning of each segment p Of (2) a synchronization sequence M p Get >>
The synchronization sequence is composed of 0 and 1 and is used for the receiving endThe synchronization and training decider.
3. The concealment communication method based on video frame offset coding of claim 1, characterized in that, the step 2 specifically comprises the following steps:
step 2.1: coding 0 and 1: a video frame encodes a bit of data, if the binary number is 1, the corresponding video frame appears in the same time as the uncoded original video; if the binary number is 0, the corresponding video frame appears delayed by a time delta compared with the uncoded original video t (ii) a When the video is decoded, the moment of each frame corresponds to the peak moment of the CPU leakage magnetic field, and the sampled magnetic field signal is divided into a plurality of magnetic field signals with the length of
Of the magnetic field signal segment of (a), wherein f ps For the video frame rate, each magnetic field signal segment represents a magnetic field generated by one frame of video and also represents a magnetic field signal corresponding to one-bit binary number in data to be transmitted; selecting a proper starting point for dividing so as to ensure that the peak of the magnetic field signal segment corresponding to 1 appears in the middle position of the segment, the peak of the magnetic field of the segment corresponding to 0 appears in the positions at two sides of the segment, and encoding 0 and 1 by utilizing the different magnetic field energy distribution of the two magnetic field segments;
step 2.2: and (3) adjusting the time of each frame of the original video by using the timestamp file obtained in the step (2.1), wherein the time of each frame in the generated new video is the same as the time specified in the timestamp file.
4. The method of claim 3, wherein T is set aside at the beginning and end of the video to be encoded before And T after The time is not coded, the buffer is reserved, the middle part of the video is coded according to the principle that one frame corresponds to one binary bit, and the Nth bit of the data to be transmitted is coded as an example: if the Nth bit is 1, the time stamp information written into the frame is
Same as when not encoded; if the Nth bit is 0, then the timestamp information written to the frame is ≥>
I.e. the frame is delayed by a time delta compared to the uncoded video t And setting a time stamp for each frame of the original video in the same way to generate a time stamp file, wherein the time stamp file records the occurrence time information of each frame of the video.
5. The concealment communication method based on video frame offset coding of claim 1, characterized in that step 3.2 is specifically: firstly, mapping each magnetic field signal segment for encoding one bit of data into a total energy value capable of representing the segment, thereby discretizing dense approximately continuous magnetic field signals obtained by sampling into an energy value capable of representing the magnetic field signal segment corresponding to one encoding bit; setting a synchronization sequence M p Is p, the sampling rate of the magnetic sensor is f s The frame rate of the played video is f r Then the total number of sampling points corresponding to the p-bit binary number is
This is the size of a sliding window over the magnetic field signal, after which the corresponding->
Uniformly sampling k values in each sampling point, summing the k values to obtain a total energy value of the magnetic field signal segment, and mapping the sliding window to a p-dimensional characteristic vector E pi Feature vector E of the sliding window pi And a synchronization sequence M p Performing cross-correlation operation, and recording the cross-correlation result as C i (ii) a The sliding window slides on the whole magnetic field signal to sample in sequence, the calculation is repeated, and finally all the sliding windows and the synchronous sequence M are obtained p The cross-correlation result array C is obtained, and the cross-correlation of all the sliding windows is obtainedAfter the result, the cross-correlation result with the largest absolute value is selected as the best cross-correlation result | C- max The sliding window is regarded as the position of the synchronization sequence, and the starting time of the sliding window is the starting time of the synchronization sequence in the signal and is recorded as +>
/>
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210101177.5A CN114554223B (en) | 2022-01-27 | 2022-01-27 | Concealed communication method based on video frame offset coding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210101177.5A CN114554223B (en) | 2022-01-27 | 2022-01-27 | Concealed communication method based on video frame offset coding |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114554223A CN114554223A (en) | 2022-05-27 |
| CN114554223B true CN114554223B (en) | 2023-03-31 |
Family
ID=81673721
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210101177.5A Active CN114554223B (en) | 2022-01-27 | 2022-01-27 | Concealed communication method based on video frame offset coding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114554223B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2015121453A (en) * | 2015-06-04 | 2016-12-27 | Закрытое акционерное общество "ОРТКОМ" | METHOD FOR STEGANOGRAPHIC INFORMATION TRANSFER |
| CN109640096A (en) * | 2018-12-06 | 2019-04-16 | 浙江大学 | A kind of concealed communication method based on video decoding electromagnetic leakage |
| CN110190923A (en) * | 2019-05-31 | 2019-08-30 | 北京大学 | A kind of latent communication system and method based on non-cooperation blind source signal |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10110930B2 (en) * | 2015-07-02 | 2018-10-23 | Dialogic Corporation | Robust packet loss handling in recording real-time video |
-
2022
- 2022-01-27 CN CN202210101177.5A patent/CN114554223B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2015121453A (en) * | 2015-06-04 | 2016-12-27 | Закрытое акционерное общество "ОРТКОМ" | METHOD FOR STEGANOGRAPHIC INFORMATION TRANSFER |
| CN109640096A (en) * | 2018-12-06 | 2019-04-16 | 浙江大学 | A kind of concealed communication method based on video decoding electromagnetic leakage |
| CN110190923A (en) * | 2019-05-31 | 2019-08-30 | 北京大学 | A kind of latent communication system and method based on non-cooperation blind source signal |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114554223A (en) | 2022-05-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100499824C (en) | Methods and systems for preventing start code emulation at locations that include non-byte aligned and/or bit-shifted positions | |
| TWI480855B (en) | Extraction and matching of characteristic fingerprints from audio signals | |
| US4599723A (en) | Method of encoding data for serial transmission | |
| Li et al. | Detection of quantization index modulation steganography in G. 723.1 bit stream based on quantization index sequence analysis | |
| EP1964352A1 (en) | Apparatuses, methods and computer program for generating and interpreting a data stream with a series of segments having specified entry points | |
| CN1663281A (en) | Method for generating hashes from a compressed multimedia content | |
| WO2010021966A1 (en) | Feature optimization and reliability estimation for audio and video signature generation and detection | |
| BRPI0613720A2 (en) | method and device for watermarking time stamps, method and device for decoding time stamps | |
| CN107993669B (en) | Voice content authentication and tampering recovery method based on modification of least significant digit weight | |
| EP1468567A2 (en) | Methods and systems for start code emulation prevention and data stuffing | |
| CN109640096B (en) | Covert communication method based on video decoding electromagnetic leakage | |
| CN100550653C (en) | A kind of Code And Decode method of variable length structural information | |
| CN107481738B (en) | Real-time audio comparison method and device | |
| JP2012507047A (en) | Method and apparatus for extracting data encoded in media content | |
| Ali et al. | A review of digital forensics methods for JPEG file carving | |
| JP2004529526A (en) | Robust checksum | |
| US20100053335A1 (en) | System and method for measuring image quality of moving pictures | |
| US20100303279A1 (en) | Method and system for temporal synchronization | |
| CN114554223B (en) | Concealed communication method based on video frame offset coding | |
| CN116915363A (en) | Data transmission method, device and equipment | |
| WO2006070222A2 (en) | An apparatus and method for adaptive digital locking and soft evaluation of data symbols in a wireless digital communication system | |
| CN115695564B (en) | Efficient transmission method of Internet of things data | |
| CN101350198B (en) | Method for compressing watermark using voice based on bone conduction | |
| CN114051082B (en) | Steganography detection feature selection method and device based on distortion degree and information gain ratio | |
| CN114999502B (en) | Adaptive word framing based voice content watermark generation and embedding method and voice content integrity authentication and tampering positioning method |
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 |