A video data tamper-proofing method and system

Abstract

The application relates to the field of video tamper-proofing technology, in particular to a video data tamper-proofing method and system, which comprises the following steps: performing one-level encryption on each frame of video data to be analyzed after encoding processing; determining the code stream data characteristics of each frame of video image based on the frame type, the quantization parameter and the code rate of each frame of encoded video image; performing frequency domain analysis on each frame of encoded video image to determine the amplitude spectrum and the phase spectrum of each frame of video image, so as to determine the frequency domain characteristics of each frame of video image; obtaining the importance degree of each frame of video image by using the frequency domain characteristics of each frame of video image; classifying all the frames of encoded video image; and performing two-level encryption on the one-level encrypted video frame by using an elliptic curve encryption algorithm and by adopting different encryption security strengths according to the importance degree of the video image in each category. Thus, the tamper-proofing capability of the video data is enhanced.

Description

Technical Field

[0001] This application relates to the field of video anti-tampering technology, specifically to a method and system for preventing video data from being tampered with. Background Technology

[0002] Preventing video data tampering helps ensure the authenticity and credibility of video content, especially in areas where video is relied upon as evidence, such as the judiciary, news reporting, and public safety. Video, as an important form of evidence, plays a crucial role in many scenarios. If video data is tampered with, it may lead to erroneous conclusions or mislead the public, causing irreparable consequences.

[0003] Currently, video encryption algorithms are often used to prevent video data tampering. Existing video encryption algorithms encrypt segments of the entire video frame using the same key for both encryption and decryption, resulting in poor security. Furthermore, since each segment contains continuous time-series video data, unauthorized access can lead to the leakage of important data. In addition, with continuous acquisition of high-resolution video data, the scale and complexity of the acquired video data increase significantly. Existing video encryption methods incur huge computational resource overhead and data transmission bandwidth requirements when processing video data, making them prone to data loss, especially in complex network environments and under bandwidth constraints, thus failing to guarantee the security of video data. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this application is to provide a method and system for preventing video data from being tampered with. The specific technical solution adopted is as follows: In a first aspect, embodiments of this application provide a method for preventing video data from being tampered with, the method comprising the following steps: Each frame of the video data to be analyzed is encoded and then encrypted at level one. Based on the frame type, quantization parameters, and bit rate of each encoded video frame, the characteristics of the bitstream data of each video frame are determined. Frequency domain analysis is performed on each frame of encoded video image to determine the amplitude spectrum and phase spectrum of each frame of video image. The amplitude difference coefficient and phase difference coefficient of each frame of video image are obtained by the difference in amplitude spectrum distribution and phase spectrum distribution between each frame of encoded video image and its adjacent frame of video image, so as to determine the frequency domain characteristics of each frame of video image. The importance of each video frame is obtained by utilizing the frequency domain features of each frame; all encoded video frames are classified according to the differences in bitstream data features and frequency domain features between different frames; and different encryption security strengths are adopted based on the importance of the video images in each category, and the video frames after the first-level encryption are encrypted using the elliptic curve encryption algorithm.

[0005] In one embodiment, determining the bitstream data characteristics of each frame of video image includes: Calculate the mean and dispersion of the quantization parameters of all macroblocks in each frame of the encoded video image, and determine the quantization parameter characteristics of each frame of the encoded video image based on the mean and the dispersion. The frame types of video images are represented digitally; The bitstream data features of each frame of video image are a vector composed of the digital representation of the frame type, the quantization parameter features, and the bit rate.

[0006] In one embodiment, the quantization parameter features are positively correlated with the mean and negatively correlated with the degree of dispersion.

[0007] In one embodiment, determining the amplitude difference coefficient includes: Calculate the mean of all amplitudes in the amplitude spectrum of each video frame, denoted as the first mean, and calculate the fusion result of the difference between the first mean of each video frame and its adjacent video frames; The amplitude difference coefficient is the ratio of the fusion result of each frame of video image to the first mean.

[0008] In one embodiment, determining the phase difference coefficient includes: Calculate the mean of the absolute values ​​of all phases in the phase spectrum of each video frame, denoted as the second mean, and calculate the fusion value of the difference between the second mean of each video frame and its adjacent video frames; The phase difference coefficient is the ratio of the fusion value of each frame of video image to the second mean.

[0009] In one embodiment, the frequency domain features of each video frame are a vector composed of the amplitude difference coefficient and the phase difference coefficient.

[0010] In one embodiment, the importance of each video frame is the sum of the amplitude difference coefficient and the phase difference coefficient.

[0011] In one embodiment, classifying all encoded frame video images includes: A clustering algorithm is used to classify all encoded video frames. The distance metric in the clustering algorithm is the reciprocal of the normalized sum of the similarity of bitstream data features and the similarity of frequency domain features between different video frames.

[0012] In one embodiment, the method of employing different encryption security strengths based on the importance of video images in each category includes: Each cluster is classified into levels according to the importance of the cluster core image frames within each cluster. The importance of the cluster core image frames is positively correlated with the level of the corresponding cluster, and the number of levels is equal to the number of encryption security strengths. Image frames within each cluster level are subjected to secondary encryption with a corresponding encryption security strength, where the encryption security strength is positively correlated with the cluster level.

[0013] Secondly, embodiments of this application also provide a video data anti-tampering system, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0014] This application has at least the following beneficial effects: This application encodes and encrypts each frame of video data, thus strictly protecting each frame and preventing tampering from altering the overall video data by modifying single frames or parts of the video content. The encryption process for each frame increases the difficulty of cracking, significantly improving the integrity and fidelity of the video content. By analyzing the bitstream data characteristics of each frame of encoded video image, the differences between different video frames can be accurately distinguished, giving each video frame a unique identifier. This feature extraction helps to quickly identify key frames and changing frames when processing large amounts of video data, thereby improving the efficiency of video tampering detection. Furthermore, by analyzing the frequency domain characteristics of each frame of video image, the differences in its amplitude spectrum and phase spectrum are obtained, thereby enhancing the identification capability of video data in the frequency domain. Based on the differences in bitstream data characteristics and frequency domain characteristics between video frames, this application can intelligently classify different video frames and automatically adjust the encryption strength according to the importance of each type of video frame. This adaptive encryption strategy can ensure video security while avoiding unnecessary resource waste, and at the same time make video encryption more flexible and efficient, enhancing the anti-tampering capability of video data. This application performs secondary encryption on the video data after primary encryption, so that the encrypted image frames are in a scrambled state during transmission and are dynamically encrypted according to the importance of different image frame groups, preventing attacks and tampering during video transmission. Under the premise of ensuring data transmission security, it can meet the real-time requirements of secure data transmission under resource constraints, thereby improving the security of video data. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating the steps of a video data anti-tampering method provided in one embodiment of this application; Figure 2 A flowchart for determining the distance metric between different frames of video images. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a video data anti-tampering method and system proposed according to this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0019] The following description, in conjunction with the accompanying drawings, details a specific scheme for a video data anti-tampering method and system provided in this application.

[0020] Please see Figure 1 The diagram illustrates a flowchart of a video data anti-tampering method according to an embodiment of this application, which includes the following steps: S1 performs level 1 encryption on each frame of the video data to be analyzed after video encoding processing.

[0021] This embodiment uses a drone equipped with a high-definition camera to acquire video data from a monitored scene, denoted as the video data to be analyzed. The acquired video data undergoes preprocessing to improve video quality. The preprocessing steps include geometric correction to avoid geometric distortion; and filtering and denoising, employing a filtering algorithm to remove noise signals from the video image. This embodiment uses Gaussian filtering, but implementers can choose other feasible existing filtering algorithms; this embodiment does not impose any restrictions on this.

[0022] For each frame of the video data to be analyzed, this embodiment uses H.265 encoding based on image content complexity to encode each frame of the video data, resulting in encoded video stream data. H.265 encoding is a well-known existing technology, and its specific process will not be described in detail here.

[0023] First, considering the security of the image, each frame of video image undergoes primary encryption to ensure basic confidentiality. For primary encryption, this embodiment employs a lightweight AES-128 encryption method to encrypt each encoded image frame, obtaining the encrypted image frame ciphertext and the corresponding key for each frame. Using lightweight AES encryption for each video frame ensures basic security without consuming significant computational resources. The AES-128 encryption method is existing technology, and its specific process will not be detailed here. Implementers can use other feasible existing encryption algorithms to perform primary encryption on each encoded video frame.

[0024] S2, based on the frame type, quantization parameters and bit rate of each encoded video frame, determines the bitstream data characteristics of each video frame.

[0025] Traditional video slicing encryption algorithms typically use the same key to encrypt and decrypt images across all image frame segments in the entire video. Furthermore, each segment contains continuous time-series video data, resulting in a lack of data security against malicious attacks. On the other hand, while applying dynamic encryption to each image frame segment can ensure high data security and achieve near-untampering, this encryption method incurs enormous computational resource overhead, and the real-time performance of secure data transmission is challenged.

[0026] Furthermore, traditional video data slicing encryption methods divide the entire video into several segments before encrypting them, making each segment a continuous frame of video content. If the encrypted data is compromised, it can lead to the leakage of a continuous video segment, significantly increasing the probability of leaking critical data. Dynamic encryption of each video segment is difficult to guarantee in real-time secure video transmission in drone swarm collaborative operation scenarios due to computational resource limitations, and may even result in data loss.

[0027] To address the aforementioned issues, this embodiment first considers the bitstream characteristics between different encoded image frames. It analyzes and processes the bitstream data of each image frame obtained during the video encoding process. Specifically, it obtains the frame type, macroblock coding quantization parameter QP, and bitrate of each encoded video image frame. The image frame type is an IPB frame, namely a key frame, a prediction frame, or a bidirectional interpolation frame.

[0028] The bitstream data generated during the encoding of video images contains structured information that can characterize the features of image frames. Some key components can reflect the complexity and information capacity of the image frames.

[0029] Frame type is the most basic feature of an image frame. An I-frame is a keyframe containing complete image information of a scene, reflecting scene transitions and key visual content. P-frames and B-frames depend on I-frames and reflect the temporal redundancy of the image. In this embodiment, the frame types of different image frames are digitally represented. Image frames belonging to the I-frame, P-frame, and B-frame types are digitally represented as 2, 1, and 0, respectively. The implementer may set other digital representation methods at his own discretion, and this embodiment does not impose any restrictions on this.

[0030] The quantization parameter QP determines the compression effect and distortion level of different macroblocks in an image frame. A lower QP corresponds to high-quality images with high detail retention, indicating that the image frame is of higher importance; while a higher QP means that the image frame loses more information. QP can be used as one of the indicators to judge whether the content of an image frame is important. For any encoded image frame, the mean value of the quantization parameter QP of all macroblocks in the frame is calculated. With degree of dispersion The quantization parameter features of any given image frame are constructed, wherein the quantization parameter features are positively correlated with the mean and negatively correlated with the degree of dispersion. The degree of dispersion can be calculated using variance, standard deviation, coefficient of variation, etc.; this embodiment uses variance as the method for calculating the degree of dispersion.

[0031] The quantization parameter features of each frame of video image after encoding in this embodiment The expression is: In the formula, To ensure that the value is greater than 0 and to avoid a denominator of 0, this embodiment... The implementer can set it according to the actual situation, and this embodiment does not impose any restrictions on it.

[0032] Bitrate is a macroscopic indicator that represents the amount of information carried by an image frame. Its unit is bits. Image frames with a higher bitrate usually contain more texture and motion information, while image frames with a low bitrate usually correspond to simple scene images.

[0033] Based on the above analysis, for the encoded t-th frame video image, construct the bitstream data features of the t-th frame video image. ,in, Let be the digital representation of the frame type of the t-th frame of the video image. This represents the normalized result of the quantization parameter features of the t-th frame of the video image. This represents the normalized bitrate of the t-th frame of the video image. This embodiment uses the maximum-minimum normalization method for both the normalization of the quantization parameter features and the normalization of the bitrate. Implementers can choose other feasible normalization methods; this embodiment does not impose any restrictions on this.

[0034] S3. Perform frequency domain analysis on each frame of encoded video image to determine the amplitude spectrum and phase spectrum of each frame of video image. Obtain the amplitude difference coefficient and phase difference coefficient of each frame of video image by the difference in amplitude spectrum distribution and phase spectrum distribution between each frame of encoded video image and its adjacent frame of video image, so as to determine the frequency domain characteristics of each frame of video image.

[0035] During drone-based filming, the content of the footage changes significantly. Some static scenes are judged to have high content complexity due to the drone's movement, resulting in highly sensitive image frames in the bitstream data characteristics. However, for more important scenes, the drone hovers to capture significant changes in the footage. In these cases, relying solely on bitstream data characteristics is insufficient to distinguish the image frames that truly contain highly sensitive and important information.

[0036] Based on the above analysis, this embodiment considers the differences between each image frame and its neighboring frames to construct the frequency domain features of each video image frame. Specifically: Perform a Fourier transform on each encoded image frame to obtain the amplitude spectrum and phase spectrum of each image frame. The amplitude spectrum and phase spectrum of each image frame are denoted as follows: , ,in These are the discrete frequency points in the frequency domain transformed from an image frame. This is the domain defined by transforming the image frame into the frequency domain.

[0037] According to the Fourier translation theorem, when a drone moves, the scene captured by the camera undergoes a global translation. The change in the phase spectrum between adjacent frames is linear along the direction of movement, and the amplitude spectrum remains unchanged after the translation. For consecutive image frames, the changes in the global amplitude and phase spectra are relatively small. However, when the drone hovers to capture important footage, the scene includes, in addition to the background, dynamic ROIs (Regions of Interest) that are constantly moving and changing. The continuous movement of these targets introduces new edge texture features into the video image, causing changes in the local energy distribution in the amplitude spectrum. These changes in edge texture features also lead to changes in the phase spectrum, generally manifesting as local phase abrupt changes, concentrated along the direction of movement. Therefore, for image frames containing important information, the average changes in the global amplitude and phase spectra are relatively large.

[0038] In summary, this embodiment calculates the amplitude difference coefficient and phase difference coefficient of each frame of video image. First, the mean of all amplitudes in the amplitude spectrum of each frame of video image is calculated and denoted as the first mean. The mean of the absolute values ​​of all phases in the phase spectrum of each frame of video image is calculated and denoted as the second mean. The fusion result of the difference between each frame of video image and the first mean of its adjacent frame of video image is calculated. The fusion value of the difference between each frame of video image and the second mean of its adjacent frame of video image is calculated. Based on the first mean and the fusion result, the amplitude difference coefficient of each video frame is determined; based on the second mean and the fusion value, the phase difference coefficient of each video frame is determined.

[0039] It should be noted that difference indicates the degree of difference between two variables, which can be calculated using methods such as the absolute value of the difference, the square of the difference, or the ratio; fusion indicates the combination of multiple variables, which can be calculated using methods such as addition, multiplication, addition-multiplication fusion, or taking the mean.

[0040] In this embodiment, the expressions for the amplitude difference coefficient and phase difference coefficient of each frame of video image are as follows: In the formula, , , The first , , The first mean of the image frames , , The first , , The second mean of the image frames, Indicates the first Amplitude difference coefficient of frame video images, Indicates the first Phase difference coefficients of frame t video images. Constructing the frequency domain features of the t-th frame video image. .

[0041] The larger the values ​​of the amplitude difference coefficient and the phase difference coefficient, the more real motion information the image frame contains and the higher its importance. Conversely, the smaller the values, the more likely it is to be a scene image or a static scene acquired by the drone.

[0042] S4. Obtain the importance of each video frame by utilizing the frequency domain features of each video frame; classify all encoded video frames according to the differences in bitstream data features and frequency domain features between different video frames; and use elliptic curve cryptography to perform secondary encryption on the video frames after primary encryption based on the importance of the video images in each category, employing different encryption security strengths.

[0043] Furthermore, the similarity of bitstream data features and frequency domain features between different frames of video images were analyzed; In the formula, The cosine similarity function is used. Let be the similarity of bitstream data features between the t-th frame and the s-th frame of the video image. Let be the frequency domain feature similarity between the t-th frame and the s-th frame of the video image. Let be the bitstream data features of the s-th frame of the video image. Let be the frequency domain features of the s-th frame of the video image.

[0044] By comparing the similarities of different image frames in multidimensional features such as frame type, quantization parameter QP, and bit rate, the differences in bitstream features between different image frames can be characterized. The closer the value is to 1, the more similar the two image frames are in terms of structural complexity. The smaller the value, the more significant the differences are between the two video frames in terms of scene content and encoding method. The closer the value is to 1, the greater the probability that the two video frames have a large number of motion features in the frequency domain, and that they belong to the same type of highly sensitive and important image frames. Conversely, the smaller the value is, the less probability that the two video frames belong to the same type of image frames.

[0045] Furthermore, considering both the differences in bitstream data characteristics and frequency domain characteristics, all image frames are classified. This embodiment uses the K-means clustering algorithm to cluster all image frames, with K set to 15. Implementers can set this value according to their actual needs; this embodiment does not impose any restrictions. The distance metric between different video frames during the clustering process is defined by the following formula. In the formula, Let be the distance metric between the t-th and s-th video frames, and Norm() be the normalization function. The K-means clustering algorithm ultimately outputs 15 different types of image frame clusters. The K-means clustering algorithm is a well-known existing technology; implementers can choose other feasible clustering algorithms, and this embodiment does not impose any restrictions. The flowchart for determining the distance metric between different video frames is as follows: Figure 2As shown.

[0046] Then, the image frames in the entire video data are scrambled and encrypted according to their importance level before being transmitted.

[0047] Specifically, the sum of the amplitude difference coefficient and the phase difference coefficient of each video frame is used as the importance of each video frame. Each cluster is classified into levels according to the importance of the cluster center image frame within each cluster. The importance of the cluster center image frame is positively correlated with the level of the corresponding cluster, and the number of levels is equal to the number of encryption security strengths when the image frame after the first level of encryption is encrypted using the elliptic curve encryption algorithm.

[0048] In this embodiment, all clusters are arranged in descending order of importance of the cluster center image frame. Every three clusters are divided into one level, and a total of 15 clusters are divided into 5 levels. The first three clusters have the highest level, the next three clusters have the next highest level, and the level of subsequent clusters decreases sequentially.

[0049] Elliptic curve cryptography algorithms with different elliptic curves are used to perform secondary encryption on image frames encrypted at the first level, based on their varying importance. The elliptic curves selected follow the NIST SP800-186 standard, with encryption strengths of 256, 224, 192, 128, and 112 respectively, ranked from highest to lowest importance. The specific curve selection is not strictly limited and can be chosen by the implementer based on the implementation scenario. Elliptic curve cryptography algorithms are well-known existing technologies, and their specific processes will not be elaborated upon.

[0050] Ultimately, this method achieves two-level encryption for secure transmission of video data. For a group of image frames containing important information, the encryption strength is high, making it difficult to crack. For a group of data with lower encryption strength, the information contained is not of high importance, and even if cracked, continuous and valid video data cannot be obtained. Compared to traditional video segment encryption methods, this embodiment achieves higher-strength dynamic encryption with a low number of video segment groups, resulting in lower computational resource overhead compared to traditional dynamic segment encryption. It can simultaneously ensure both real-time data transmission and security when multiple drones are operating collaboratively.

[0051] Based on the same inventive concept as the above method, this application embodiment also provides a video data anti-tampering system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described video data anti-tampering methods.

[0052] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0053] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0054] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.

Claims

1. A method for preventing video data from being tampered with, characterized in that, The method includes the following steps: Each frame of the video data to be analyzed is encoded and then encrypted at level one. Based on the frame type, quantization parameters, and bit rate of each encoded video frame, the characteristics of the bitstream data of each video frame are determined. Frequency domain analysis is performed on each frame of encoded video image to determine the amplitude spectrum and phase spectrum of each frame of video image. The amplitude difference coefficient and phase difference coefficient of each frame of video image are obtained by the difference in amplitude spectrum distribution and phase spectrum distribution between each frame of encoded video image and its adjacent frame of video image, so as to determine the frequency domain characteristics of each frame of video image. The importance of each video frame is obtained by utilizing the frequency domain features of each frame; all encoded video frames are classified according to the differences in bitstream data features and frequency domain features between different frames; and different encryption security strengths are adopted based on the importance of the video images in each category, and the video frames after the first-level encryption are encrypted using the elliptic curve encryption algorithm.

2. The video data anti-tampering method as described in claim 1, characterized in that, The determination of the bitstream data characteristics of each frame of video image includes: Calculate the mean and dispersion of the quantization parameters of all macroblocks in each frame of the encoded video image, and determine the quantization parameter characteristics of each frame of the encoded video image based on the mean and the dispersion. The frame types of video images are represented digitally; The bitstream data features of each frame of video image are a vector composed of the digital representation of the frame type, the quantization parameter features, and the bit rate.

3. The video data anti-tampering method as described in claim 2, characterized in that, The quantification parameter is positively correlated with the mean and negatively correlated with the degree of dispersion.

4. The video data anti-tampering method as described in claim 1, characterized in that, The determination of the amplitude difference coefficient includes: Calculate the mean of all amplitudes in the amplitude spectrum of each video frame, denoted as the first mean, and calculate the fusion result of the difference between the first mean of each video frame and its adjacent video frames; The amplitude difference coefficient is the ratio of the fusion result of each frame of video image to the first mean.

5. A video data anti-tampering method as described in claim 1, characterized in that, The determination of the phase difference coefficient includes: Calculate the mean of the absolute values ​​of all phases in the phase spectrum of each video frame, denoted as the second mean, and calculate the fusion value of the difference between the second mean of each video frame and its adjacent video frames; The phase difference coefficient is the ratio of the fusion value of each frame of video image to the second mean.

6. The video data anti-tampering method as described in claim 1, characterized in that, The frequency domain features of each video frame are a vector composed of the amplitude difference coefficient and the phase difference coefficient.

7. The video data anti-tampering method as described in claim 1, characterized in that, The importance of each video frame is the sum of the amplitude difference coefficient and the phase difference coefficient.

8. A video data anti-tampering method as described in claim 1, characterized in that, The classification of all encoded video frames includes: A clustering algorithm is used to classify all encoded video frames. The distance metric in the clustering algorithm is the reciprocal of the normalized sum of the similarity of bitstream data features and the similarity of frequency domain features between different video frames.

9. A video data anti-tampering method as described in claim 8, characterized in that, The encryption security strength is determined by the importance of video images in each category, and includes: Each cluster is classified into levels according to the importance of the cluster core image frames within each cluster. The importance of the cluster core image frames is positively correlated with the level of the corresponding cluster, and the number of levels is equal to the number of encryption security strengths. Image frames within each cluster level are subjected to secondary encryption with a corresponding encryption security strength, where the encryption security strength is positively correlated with the cluster level.

10. A video data anti-tampering system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-9.

Images