CN111510672A - Video tampering recovery processing method, system, storage medium and encoder - Google Patents
Video tampering recovery processing method, system, storage medium and encoder Download PDFInfo
- Publication number
- CN111510672A CN111510672A CN202010197707.1A CN202010197707A CN111510672A CN 111510672 A CN111510672 A CN 111510672A CN 202010197707 A CN202010197707 A CN 202010197707A CN 111510672 A CN111510672 A CN 111510672A
- Authority
- CN
- China
- Prior art keywords
- frame
- video
- key
- window
- frames
- 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.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/71—Indexing; Data structures therefor; Storage structures
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/80—Generation or processing of content or additional data by content creator independently of the distribution process; Content per se
- H04N21/83—Generation or processing of protective or descriptive data associated with content; Content structuring
- H04N21/835—Generation of protective data, e.g. certificates
- H04N21/8358—Generation of protective data, e.g. certificates involving watermark
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Signal Processing (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Security & Cryptography (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention belongs to the technical field of video information processing, and discloses a video tampering recovery processing method, a system, a storage medium and an encoder.A monitoring video is used as a deleting channel, and a video watermark consisting of highly compressed key frames and positioning data is transmitted through the deleting channel; embedding a watermark by fountain coding and extracting the watermark at a decoder end to realize intra-frame tampering detection recovery and inter-frame tampering detection; inter-frame tamper detection extracts frame indexes of all frames in a frame window from the last key frame verified as a real block, distinguishes inserted or deleted frames by using the extracted frame indexes, and determines the position of a new key frame; the verification and recovery process is repeated using the new key frame. Compared with the traditional video tampering recovery method, the method provided by the invention has the advantages that the quality of the recovered image is obviously better, the efficiency is better than that of the existing algorithm, and the quality range is about 33.2-41.6 dB.
Description
Technical Field
The invention belongs to the technical field of video information processing, and particularly relates to a video tampering recovery processing method, a video tampering recovery processing system, a storage medium and an encoder.
Background
Currently, surveillance videos are often used as case-finding and forensic evidence, and therefore it is important to ensure the authenticity and reliability of the surveillance videos. The method mainly comprises the steps of performing time domain tampering and airspace tampering on a monitored video, wherein the time domain tampering is to operate a frame sequence in an interframe editing process, and common attack types comprise frame addition, frame deletion and frame rearrangement; the airspace tampering refers to modifying the content of a single-frame image in the intraframe editing process, and common space tampering attacks can be divided into object addition, object deletion and object modification. In most video tampering detection methods, the existence of an original video is crucial, and compared with a common video tampering detection method, the video tampering recovery method has very obvious advantages, the original video is not needed in the video tampering recovery method, and after the video to be detected is tampered, the tampered area can be recovered through the video tampering recovery method, so that the video tampering recovery method capable of effectively recovering the tampered area after the video is detected to be tampered is needed.
In the related technology for researching image tampering detection and recovery, foreign experts propose a fragile watermarking method based on variable capacity for an image with rough texture. Aiming at the field of satellite and medical image processing, experts utilize data hiding and image uniformity, and a region-based tampering detection and recovery method is provided. The above two methods are only suitable for the case where the tampering ratio is small, and the recovery image is not quantitatively analyzed.
In the field of researching video tampering and recovery, aiming at the problems of low tampering positioning precision and poor quality of recovered images, an expert uses non-negative matrix decomposition (NMF) to provide a method for detecting and self-recovering by using extended NMF with different sparse constraints, a block-level watermark and a frame-level watermark are generated by using a hash function and the provided method, the block-level watermark and the frame-level watermark are embedded into a video to verify time domain tampering and space domain tampering, and finally, the intelligent characteristic of a basic matrix is used to partially recover the basic matrix and recover a tampered region by using the space-time continuous characteristic of the video. By using halftone technique, there is a method for protecting and restoring video image by using secure watermark, which is characterized in that a binary version of original video frame, i.e. reference frame, is generated at the encoder end by using halftone technique, the reference frame is embedded into other frames as watermark by using quantization index modulation technique, and after the decoder end identifies tampering, the original content is restored by inverse halftone technique. The above two methods can only partially recover the tampered region and do not perform quantitative analysis on the quality of the recovered video frame.
Early foreign and domestic scholars dedicated to the research of image tampering recovery methods and publicly published a plurality of research results, however, the research on video tampering recovery has just started in recent years, the two published results introduced above have the problem that only part of tampered regions can be recovered, the accuracy and efficiency of a video recovery algorithm are low, and the recovered video frames are not quantitatively analyzed.
Through the above analysis, the problems and defects of the prior art are as follows: at present, a method for recovering video tampering only partially recovers a tampered area, and the problems of low accuracy and efficiency of a video recovery algorithm exist, and a recovered video frame is not subjected to quantitative analysis.
The difficulty in solving the above problems and defects is: video data needs to be coded in the storage process, the video data volume is very large, and the performance requirement on the video detection and recovery algorithm is higher during processing
The significance of solving the problems and the defects is as follows: the authenticity and reliability of the monitoring video serving as evidence in the law enforcement process are very important, and if the problems can be well solved, the monitoring video has unprecedented significance for criminal investigation and national justice.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a video tampering recovery processing method, a system, a storage medium and an encoder.
The invention is realized in such a way, the video tampering recovery processing method takes a monitoring video as a deleting channel, and transmits a video watermark consisting of highly compressed key frames and positioning data through the deleting channel; embedding a watermark by fountain coding and extracting the watermark at a decoder end to realize intra-frame tampering detection recovery and inter-frame tampering detection; inter-frame tamper detection extracts frame indexes of all frames in a frame window from the last key frame verified as a real block, distinguishes inserted or deleted frames by using the extracted frame indexes, and determines the position of a new key frame; the verification and recovery process is repeated using the new key frame.
Further, the video tampering recovery processing method includes:
in a first step, a surveillance video is divided into non-overlapping portions, each portion having wlFrames, called frame windows, the length of which is as small as possible for real-time encoding, w being chosenlCovering a visual event lasting for one second for monitoring the frame number of the video within one second;
secondly, selecting a key frame from each frame window, wherein the frame number is k, forming a key section with the length of 2G +1 by taking the key frame as a central frame, G is a fixed value, the set corresponding to the key frame and the key section is k-G, k-G +1, al；
And thirdly, forming a reference frame by highly compressing the key frame, wherein the reference frame records the main content of the key frame. For other frames in the frame window which are not in the key section, forming a frame index according to the frame window number, the frame number and the distance between the other frames and the key frame in the same window;
and fourthly, embedding the reference frame into the key segment, and embedding the frame indexes of other frames which do not belong to the key segment into the same frame in the frame window.
Further, the selection of the key frame of the video tampering recovery processing method uses a histogram comparison method, that is, extracting the key frame based on the boundary position of the shot change in the frame window, and for each frame j, calculating a difference feature value:
in the formula, HistjAn intensity histogram representing frame j, b being the intensity value;
for any other frame j in the frame window, d (k) is larger than or equal to d (j), by selecting the key frame k, if the distance from the key frame to the initial frame in the frame window is smaller than G, and k is smaller than G, replacing G with k; similarly, if the key frame is less than G away from the last frame in the frame window, k > wlG, replacing k by wlG, each selected frame window is substantially similar in length, and the key frames in the frame window may represent the content of the frame window.
Further, the video tampering recovery processing method constructs a content reference frame R according to an existing compression method, firstly preprocesses a key frame image to obtain a gray key frame, decomposes the gray key frame into a plurality of blocks with the size of 8 × 8, divides the blocks into four types according to the number of edge points in each block, applies DCT transformation to each block, quantizes DCT coefficients through a standard JPEG quantization table, and according to the corresponding bit distribution of each type of quantization coefficients, a direct current coefficient is converted into unsigned binary bits, and an alternating current coefficient is converted into signed binary bits.
Further, the construction of the frame index of the video tampering recovery processing method refers to the frame index I of the frame j in the frame windowj＝(nwj,j,dkj) The following information is contained:
1) the frame window containing frame j is numbered nwjAnd a frame number j;
2) distance dk from frame j to key frame k in the same frame windowjAt G-wjAnd wj-G.
Further, according to the embedding of the reference frame and the frame index in the video tampering recovery processing method, each reference frame is regarded as a bit sequence, each reference frame is bit information, wherein the length of the encoding symbol is represented, and the number of the encoding symbols is represented, each frame in the key segment is decomposed into blocks with the size of:
wherein N represents the number of blocks of a frame decomposition, b2Representing the size of the block, W representing the surveillance video width, and H representing the surveillance video height;
for each frame j, the N blocks that are decomposed are denoted Bj,1,...,Bj,NEmbedding reference frames in these blocks using fountain coding B j,p1 ≦ p ≦ N and k-G ≦ j ≦ k + G, each of the (2G +1) N blocks carrying the slave reference frame R1,...,RnWatermark payload symbols obtained by fountain coding, Bj,pBy Wj,pIs represented and calculated as R1,...,RnPseudo-random linear combination of (2), Wj,pIs equal to S L, verifying that the watermark is represented by a hash function Bj,pAnd Wj,pConsists of the following components:
Hj,p＝Hash(Bj,p,Wj,p,j,p,key)。
further, the video tamper recovery processing method simultaneously detecting intra-frame and inter-frame tampering by a decoder includes:
firstly, the speed of detecting the intra-frame tampering is very high, and in the intra-frame tampering detection module, a surveillance video is divided into a plurality of monitoring videos with the length wlIn the initial frame window, determining a key frame k and extracting therefrom a verification watermark (W'k,p,H′k,p) N, for p ═ 1.. N, extracted H'k,pEqual to:
corresponding to p, if in key frame k
Considered to be a real block, e is setk,p0; otherwise, set ek,pThe tamper rate of key frame k is calculated as 1:
the error map for key frame k is defined as the set:
{ek,p|p＝1,...,N}；
if all blocks in the key frame k are considered to be real, moving to the next window and repeating the steps; if the tampering rate is lower than the threshold rho, TR < rho, recovering the tampered block, recovering the reference frame of the key frame by extracting the recovery watermark embedded in the key section, and replacing the block which is considered to be tampered in the key frame with the corresponding block in the reference frame;
under the condition that rho < TR < 1, the tampered block cannot be recovered, the decoder can only locate the tampering, and the error map represents the position of the tampered block; under the condition that TR is 1, extracting a frame index in the same frame window, if the extracted frame index indicates that the frame belongs to the selected window and is correctly sequenced, searching the key frame again and determining the number of the key frame, and repeating the steps; otherwise, the decoder judges that the frame is tampered, and executes the frame tampering detection module.
It is another object of the present invention to provide a program storage medium for receiving user input, the stored computer program causing an electronic device to perform the steps comprising: the monitoring video is used as a deleting channel, and the video watermark consisting of the highly compressed key frame and the positioning data is transmitted through the deleting channel; embedding a watermark by fountain coding and extracting the watermark at a decoder end to realize intra-frame tampering detection recovery and inter-frame tampering detection; inter-frame tamper detection extracts frame indexes of all frames in a frame window from the last key frame verified as a real block, distinguishes inserted or deleted frames by using the extracted frame indexes, and determines the position of a new key frame; the verification and recovery process is repeated using the new key frame.
Another object of the present invention is to provide a video tamper recovery processing system implementing the video tamper recovery processing method, the video tamper recovery processing system including:
a deletion channel establishing module for taking the monitoring video as a deletion channel;
the video watermark composition module is used for transmitting the video watermark consisting of the highly compressed key frames and the positioning data through a deletion channel;
the watermark embedding and extracting module is used for embedding the watermark by adopting fountain coding and extracting the watermark at a decoder end;
and the recovery and detection module is used for realizing intra-frame tampering detection recovery and inter-frame tampering detection.
Another object of the present invention is to provide an encoder having the video tamper recovery processing system installed therein.
By combining all the technical schemes, the invention has the advantages and positive effects that: the invention provides a fountain code-based video tampering recovery method for a monitoring video scene with a static background. The monitoring video is used as a deleting channel, a video watermark consisting of highly compressed key frames and positioning data is transmitted through the channel, the watermark is embedded by fountain coding and extracted at a decoder end, and intra-frame tampering detection recovery and inter-frame tampering detection are realized. Compared with the traditional video tampering recovery method, the method provided by the invention has the advantages that the quality of the recovered image is obviously better, the efficiency is better than that of the existing algorithm, and the quality range is about 33.2-41.6 dB. The PSNR value of an image is a description mode for evaluating the image quality, and is an existing evaluation standard. The unit of PSNR is dB, the larger the value is, the smaller the distortion is, the PSNR is higher than 40dB, which indicates that the image quality is excellent (namely, the image is very close to the original image); generally indicating image quality is good at 30-40 dB (i.e. distortion is perceptible but acceptable); a common reference at 30 dB; poor image quality is illustrated at 20-30 dB; finally, PSNR is unacceptable below 20dB images. 33.2-41.6 are the algorithmic performances described in psnr for the algorithms proposed in this algorithm. Compared with the existing algorithm, the embodiment of the invention has better recovery capability and calculation performance.
Drawings
Fig. 1 is a flowchart of a video tampering recovery processing method according to an embodiment of the present invention.
Fig. 2 is a schematic structural diagram of a video tamper recovery processing system according to an embodiment of the present invention;
in the figure: 1. a delete channel establishment module; 2. a video watermark composition module; 3. a watermark embedding and extracting module; 4. a recovery and detection module.
Fig. 3 is a flowchart of an implementation of a video tamper recovery processing method according to an embodiment of the present invention.
Fig. 4 is a schematic diagram of a key segment in a frame window according to an embodiment of the present invention.
FIG. 5 is a representation of bit allocation provided by an embodiment of the present invention;
in the figure: (a) block type 1; (b) block type 2; (c) block type 3; (d) block type 4.
Fig. 6 is a flow chart of a decoder system according to an embodiment of the present invention.
FIG. 7 is a schematic diagram of an original key frame, a watermarked key frame, and a tampered key frame provided by an embodiment of the present invention.
Fig. 8 is a schematic diagram of a test video tamper rate and a recovered image PSNR according to an embodiment of the present invention.
Fig. 9 is a schematic diagram of tamper detection of a test video according to an embodiment of the present invention.
Fig. 10 is a schematic diagram of tamper detection of a test video according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In view of the problems in the prior art, the present invention provides a method, a system, a storage medium, and an encoder for video tamper recovery processing, and the present invention is described in detail below with reference to the accompanying drawings.
As shown in fig. 1, the video tampering recovery processing method provided by the present invention includes the following steps:
s101: the monitoring video is used as a deleting channel, and the video watermark consisting of the highly compressed key frame and the positioning data is transmitted through the deleting channel;
s102: and fountain codes are adopted to embed watermarks, and extraction is carried out at a decoder end, so that intra-frame tampering detection recovery and inter-frame tampering detection are realized.
As shown in fig. 2, the video tamper recovery processing system provided by the present invention includes:
and a deletion channel establishing module 1, configured to use the monitoring video as a deletion channel.
And the video watermark composing module 2 is used for transmitting the video watermark composed of the highly compressed key frames and the positioning data through the deleting channel.
And the watermark embedding and extracting module 3 is used for embedding the watermark by adopting fountain coding and extracting the watermark at a decoder end.
And the recovery and detection module 4 is used for realizing intra-frame tamper detection recovery and inter-frame tamper detection.
The technical solution of the present invention is further described below with reference to the accompanying drawings.
As shown in fig. 3, the video tampering recovery processing method provided by the present invention includes the following steps:
in a first step, a surveillance video is divided into non-overlapping portions, each portion having wlFrames, called frame windows, the length of which should be as small as possible for real-time encoding and large enough for the encoding process to be effective, therefore, reasonable selection of the length of the frame window, e.g. selection of w, is requiredlTo monitor the number of frames in a second of video, a visual event lasting for one second is overlaid.
Second, from each frame windowWherein, a key frame (frame number is k) is selected, and a key segment with length 2G + 1 is formed by taking the key frame as a central frame, G is a set fixed value, the set corresponding to the key frame and the key segment is k-G, k-G +1, al。
And thirdly, forming a reference frame by highly compressing the key frame, wherein the reference frame records the main content of the key frame. And for other frames in the frame window which are not in the key section, forming a frame index by the frame window number, the frame number and the distance between the other frames and the key frame in the same window.
And fourthly, embedding the reference frame into the key segment, and embedding the frame indexes of other frames which do not belong to the key segment into the same frame in the frame window.
The selection of key frames of the present invention is crucial in the encoding scheme, and the content of the video is represented by the constructed sequence of key frames. In the present example, key frames are extracted using a histogram comparison based method, i.e., based on the boundary position of shot changes in the frame window. For each frame j, compute the difference feature value:
in the formula (1), HistjAn intensity histogram representing frame j, and b is the intensity value.
For any other frame j within the frame window, d (k) ≧ d (j) is satisfied, by which the key frame k is selected. If the distance from the key frame to the initial frame in the frame window is less than G (namely k < G), replacing G with k; similarly, if the key frame is less than G away from the last frame in the frame window (i.e., k > w)l-G) replacing k by wl-G. Each selected frame window is substantially similar in length, and the key frames in the frame window can represent the content of the frame window. The selection of key frames by histogram comparison method guarantees local differences (such as shot boundaries).
The reference frame of the present invention is constructed such that the embedded reference frame does not change the visual information effect of the surveillance video, the size of the constructed reference frame should be as small as possible, and thus a highly compressed version of the key frame is used as the reference frame.A content reference frame R is constructed according to the existing compression method, a key frame image is first preprocessed to obtain a gray key frame, the gray key frame is decomposed into a number of blocks having a size of 8 × 8, which are divided into four types according to the number of edge points in each block, as shown in FIG. 5. DCT transform (discrete cosine transform) is applied to each block and DCT coefficients are quantized through a standard JPEG quantization table.A corresponding bit allocation of each type of quantization coefficients is determined according to the table in FIG. 5, DC coefficients are converted into unsigned binary bits, AC coefficients are converted into signed binary bits.5 (a) -5 (d) blocks retain 80, 60, 40 and 22 bit information, respectively, and two additional bit recording types of blocks are used so the average block method uses 52.5.52.
The construction of the frame index of the invention refers to the frame index I of the frame j in the frame windowj＝(nwj,j,dkj) The following information is contained:
1) the frame window containing frame j is numbered nwjAnd a frame number j.
2) Distance dk from frame j to key frame k in the same frame windowjAt G-wjAnd wj-G.
According to the embedding of the reference frame and the frame index, each reference frame is regarded as a bit sequence, each bit is information of a bit, wherein the length of a coding symbol is represented, and the number of the coding symbols is represented. Each frame in the critical section is broken down into blocks of size, the size of the blocks indicating the precision of the tamper bits, as represented by equation (2):
wherein N represents the number of blocks of a frame decomposition, b2Denotes the block size, W denotes the surveillance video width, and H denotes the surveillance video height.
For each frame j, the N blocks that are decomposed are denoted Bj,1,...,Bj,NEmbedding reference frames in these blocks using fountain coding B j,p1 ≦ p ≦ N and k-G ≦ j ≦ k + G, each of the above (2G +1) N blocks carrying the slave reference frame R1,...,RnWatermark payload symbols obtained by fountain coding. B isj,pBy Wj,pIs represented and calculated as R1,...,RnPseudo-random linear combination of, thus Wj,pIs equal to S L the pseudo-random linear combination method is used because it is used in a random linear fountain (R L F) and a verification watermark is embedded in each block, the decoder treats the received symbol as correctly transmitted, representing the real block, or as deleted, the verification watermark being represented by a hash function Bj,pAnd Wj,pThe composition is as follows:
Hj,p＝Hash(Bj,p,Wj,p,j,p,key) (3)
embedding B when video frame j belongs to a key segmentj,pComprises (W)j,p,Hj,p) (ii) a When video frame j does not belong to the key segment, embed Bj,pIs that the information ofj＝(nwj,j,dkj). The embedding method of the other frames in the non-key section is different from the embedding method of the frames in the key section.
Regarding the embedding algorithm, consider the following:
1) imperceptibility. Embedding of the watermark should not degrade video quality;
2) on the other hand, since embedding reference frames together with the hash function requires a large capacity, the common method mainly embeds 3L SB in the luminance component Y (3 b/pixel) or red, green, and blue (RGB) bit-planes (9 b/pixel).
3) Robustness: a robust or fragile watermark is selected for video processing, such as compression processing.
Three features of the watermarking method are inherently interrelated, for example, increasing embedding capacity reduces imperceptibility and robustness. The frame index only contains the frame window number, the frame number and the distance from the frame to the key frame, and the capacity occupied by the frame index is very small compared with the reference frame, so that the only relevant problem of embedding the frame index is imperceptibility and robustness.
The frame index of each frame in the key segment is embedded into the corresponding frame, and the method for embedding the frame index and embedding the reference frame should keep a distance, for example, the frame index can be embedded in the first L SB (least significant bit algorithm L eastsignificantit), and the reference frame is embedded in the second L SB.
The flow chart of the decoder of the present invention is shown in fig. 6, by which both intra-frame and inter-frame tampering can be detected.
Firstly, the speed of detecting the intra-frame tampering is very high, and in the intra-frame tampering detection module, a surveillance video is divided into a plurality of monitoring videos with the length wlIn an initial frame window, determining a key frame k and extracting therefrom a verification watermark (W'k,p,H′k,p) N, for p ═ 1.. N, extracted H'k,pEqual to:
corresponding to p, if in key frame k
Considered to be a real block, e is setk,p0; otherwise, set ek,pThe tamper rate of key frame k is calculated as 1:
the error map for key frame k is defined as the set:
{ek,p|p＝1,...,N}(6)
if all blocks in key frame k are considered to be authentic, move to the next window and repeat the above steps. If the tampering rate is lower than the threshold ρ, i.e. TR < ρ, the tampered block can be recovered, the reference frame of the key frame is recovered by extracting the recovery watermark embedded in the key section, and then the block in the key frame considered tampered is replaced with the corresponding block from the reference frame.
In case ρ < TR < 1, the tampered block cannot be recovered and the decoder can only locate the tampering, the error map indicating the location of the tampered block. In the case where TR is 1, there is a possibility that a key frame is erroneously selected due to tampering, a frame index within the same frame window is extracted, and if the extracted frame index indicates that the frames belong to the selected window and have been correctly ordered, the key frame is searched again and the number thereof is determined, and the above steps are repeated. Otherwise, the decoder judges that the frame is tampered, and executes the frame tampering detection module.
The inter-frame tampering detection module first extracts the frame indexes of all frames in the frame window starting from the last key frame verified as a real block, and can distinguish the inserted or deleted frames by using the extracted frame indexes and determine the position of a new key frame. The verification and recovery process is then repeated using these new key frames. When the surveillance video is not tampered or only intra-frame tampered, the proposed method uses only an intra-frame tampering detection module, which can detect with very fast execution speed.
The invention relates to a fountain code-based video tampering recovery method, which takes a monitoring video as a deleting channel, transmits a video watermark consisting of a highly compressed key frame and positioning data through the channel, adopts fountain codes to embed the watermark and extracts the watermark at a decoder end, thereby realizing intra-frame tampering detection recovery and inter-frame tampering detection. Meanwhile, compared with the traditional video tampering recovery method, the method proves that the quality of the image recovered by the algorithm is more advantageous, and the quality range is about 33.2-41.6 dB.
The technical effects of the present invention will be described in detail with reference to experiments.
1. The ten test videos used in the experimental evaluation were from the REWIND video copy-mobile forgery database, the resolution of each video sequence was 240 × 320 pixels and a frame rate of 30 frames per second, the length of the frame window, WlSet to 20 frames, therefore, capable of covering visual events lasting around 0.67 seconds, video frames are divided into N-1200 size 8 × 8The frame index of the key frame and the reference frame are embedded in the second L SB of the Y luminance component, therefore, the capacity of embedding is equal to 64bpb (number of bits per block), or equivalent to 1bpp (number of bits per pixel).
A critical segment of length 2G + 1 is selected to be equal to 5 or, equivalently, set to G-2, with 32 bits in each 8 × 8 block in the critical segment for embedding a hash function generated by the MD5 algorithm, with the symbol length S L also set to 32, i.e., 32 bits for embedding fountain-coded reference frames into each block within the critical segmentjIs set to 24 bits, of which 14 bits are dedicated to the frame window number nw j5 bits are used for the number j of the video frame in the frame window, 5 bits are used for the distance dk from the frame j to the key frame in the same frame windowj. The decoder selects from each block Bj,pExtract frame index I'j,2p-1And l'j,2p. Selecting a known frame index I'jSo as to satisfy the following conditions:
1)σ(I′j)＝#{p∈{1,2,...,2N}|I′j,p＝I′jis the maximum value;
2)σ(I′j) T (N), where T (N) denotes a threshold value dependent on N;
in the experiment, t (N) ═ N/10 (120) is set, that is, the detected frame index reaches at least 120 to be confirmed.
2. Intra-frame tamper detection and recovery experiment
The performance of the proposed method in recovering the content of the key frame is shown in the figure (which is fig. 7), when the video is subjected to intra-frame tampering, the content of the key frame changes, and the original key frame, the watermarked key frame and the tampered key frame (with a tampering rate of 16.5%) are respectively shown in the figures (which is fig. 7) (a) -7 (c), which are fig. 7) (d) showing the recovered key frame, wherein the tampered part is replaced by the corresponding part in the recovered reference frame. The content of the key frame and the content of the key segment are associated, so the tamper rate refers to the tamper rate in the key segment. The modification of the tampered key frames by the recovered video frames can be clearly seen by comparing the recovered video frames with the tampered key frames. The highest reversible tamper rate is found by experimentation, as shown in the figure (which is fig. 8) each line represents a particular key frame of one of the test videos of the REWIND video data set. The corresponding frame window is randomly tampered 20 times, TR equals 10%, 20%, …, 60% and 67%. The average PSNR value of the recovered key frames decreases as the tampering rate increases. The standard deviation of the calculated PSNR from the mean PSNR decreased from 0.54 at TR 10% to 0.13 at TR 67%. It has been found experimentally that the standard deviation is equal to 0.42 for TR 20%, 0.28 for TR 30%, 0.22 for TR 40%, 0.19 for TR 50% and 0.14 for TR 60%. When the tampering rate exceeds 67%, the key frame will not be recovered, but the method can still detect tampering. The average PSNR of the recovered key frames is between 33.dB and 41.6dB, and is acceptable in the scenes of judicial appraisal, law enforcement application and the like.
3. Interframe tampering detection experiment
The inter-frame tampering detection is carried out on the proposed method, the first test video consists of 130 frames, the 31 st to 45 th frames are replaced by the 5 th to 19 th frames, the 71 th to 100 th frames are deleted from the video, as shown in the figure (which is figure 9), the tampered video is input, the proposed method carries out inter-frame tampering detection on the tampered video, the 31 st to 45 th frame numbers are successfully extracted, and the 71 th to 100 th frames are judged to be deleted. The 41 th frame to the 70 th frame of the second test video are replaced by a continuous 30-frame video sequence of other videos, the extraction frame index module cannot generate a valid index set when the method is detected, and the 30 frames are not identified as key sections. Therefore, the decoder declares the 30 frames as an inserted frame, and recovers the original frame number of the frame in the key segment from the information available in the frame index of the remaining frames, as shown in the figure (which is fig. 10).
It should be noted that the embodiments of the present invention can be realized by hardware, software, or a combination of software and hardware. The hardware portion may be implemented using dedicated logic; the software portions may be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or specially designed hardware. Those skilled in the art will appreciate that the apparatus and methods described above may be implemented using computer executable instructions and/or embodied in processor control code, such code being provided on a carrier medium such as a disk, CD-or DVD-ROM, programmable memory such as read only memory (firmware), or a data carrier such as an optical or electronic signal carrier, for example. The apparatus and its modules of the present invention may be implemented by hardware circuits such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of hardware circuits and software, e.g., firmware.
The above description is only for the purpose of illustrating the present invention and the appended claims are not to be construed as limiting the scope of the invention, which is intended to cover all modifications, equivalents and improvements that are within the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. A video tampering recovery processing method is characterized in that a monitoring video is used as a deleting channel, and a video watermark consisting of highly compressed key frames and positioning data is transmitted through the deleting channel; embedding a watermark by fountain coding and extracting the watermark at a decoder end to realize intra-frame tampering detection recovery and inter-frame tampering detection; inter-frame tamper detection extracts frame indexes of all frames in a frame window from the last key frame verified as a real block, distinguishes inserted or deleted frames by using the extracted frame indexes, and determines the position of a new key frame; the verification and recovery process is repeated using the new key frame.
2. The video tamper recovery processing method according to claim 1, wherein the video tamper recovery processing method includes:
in a first step, a surveillance video is divided into non-overlapping portions, each portion having wlFrames, called frame windows, the length of which is as small as possible for real-time encoding, w being chosenlFor monitoring video within one secondFrame number, covering visual events lasting for one second;
secondly, selecting a key frame from each frame window, wherein the frame number is k, forming a key section with the length of 2G +1 by taking the key frame as a central frame, G is a fixed value, the set corresponding to the key frame and the key section is k-G, k-G +1, al；
Thirdly, highly compressing the key frame to form a reference frame, recording the main content of the key frame by the reference frame, and forming a frame index for other frames which are not in the key section in the frame window according to the frame window number, the frame number and the distance between the other frames and the key frame in the same window;
and fourthly, embedding the reference frame into the key segment, and embedding the frame indexes of other frames which do not belong to the key segment into the same frame in the frame window.
3. The video tamper recovery processing method according to claim 2, wherein the selection of the key frames of the video tamper recovery processing method uses a histogram comparison based method, that is, extracting key frames based on the boundary position of shot changes in a frame window, and for each frame j, calculates a difference feature value:
in the formula, HistjAn intensity histogram representing frame j, b being the intensity value;
for any other frame j in the frame window, d (k) is larger than or equal to d (j), by selecting the key frame k, if the distance from the key frame to the initial frame in the frame window is smaller than G, and k is smaller than G, replacing G with k; similarly, if the key frame is less than G away from the last frame in the frame window, k > wlG, replacing k by wlG, each selected frame window is substantially similar in length, and the key frames in the frame window may represent the content of the frame window.
4. The video tampering recovery processing method according to claim 2, wherein the video tampering recovery processing method comprises constructing a content reference frame R according to an existing compression method, preprocessing a key frame image to obtain a gray key frame, decomposing the gray key frame into a plurality of blocks with a size of 8 × 8, dividing the blocks into four types according to the number of edge points in each block, applying DCT transform to each block, quantizing DCT coefficients by a standard JPEG quantization table, converting DC coefficients into unsigned binary bits and AC coefficients into signed binary bits according to the corresponding bit allocation of each type of quantized coefficients.
5. The video tamper recovery processing method according to claim 2, wherein the frame index of the video tamper recovery processing method is constructed by referring to a frame index I of a frame j within a frame windowj＝(nwj,j,dkj) The following information is contained:
1) the frame window containing frame j is numbered nwjAnd a frame number j;
2) distance dk from frame j to key frame k in the same frame windowjAt G-wjAnd wj-G.
6. The video tamper recovery processing method according to claim 2, wherein the embedding of the reference frames and the frame indexes of the video tamper recovery processing method, each reference frame is regarded as a bit sequence, each reference frame is information of a bit, wherein the information represents a length of a coding symbol and is a number of the coding symbols, each frame in the key segment is decomposed into blocks of a size, and the size of each block represents a precision representation of the tamper bits:
wherein N represents the number of blocks of a frame decomposition, b2Representing the size of the block, W representing the surveillance video width, and H representing the surveillance video height;
for each frame j, the N blocks that are decomposed are denoted Bj,1,...,Bj,NEmbedding reference frames in these blocks using fountain coding Bj,p1 ≦ p ≦ N and k-G ≦ j ≦ k + G, each of the (2G +1) N blocks carrying the slave reference frame R1,...,RnWatermark payload symbols obtained by fountain coding, Bj,pBy Wj,pIs represented and calculated as R1,...,RnPseudo-random linear combination of (2), Wj,pIs equal to S L, verifying that the watermark is represented by a hash function Bj,pAnd Wj,pConsists of the following components:
Hj,p＝Hash(Bj,p,Wj,p,j,p,key)。
7. the video tamper recovery processing method of claim 2, wherein the video tamper recovery processing method detecting intra-frame and inter-frame tampering simultaneously by a decoder comprises:
firstly, the speed of detecting the intra-frame tampering is very high, and in the intra-frame tampering detection module, a surveillance video is divided into a plurality of monitoring videos with the length wlIn the initial frame window, determining a key frame k and extracting therefrom a verification watermark (W'k,p,H′k,p) N, for p ═ 1.. N, extracted H'k,pEqual to:
corresponding to p, if in key frame k
Considered to be a real block, e is setk,p0; otherwise, set ek,pThe tamper rate of key frame k is calculated as 1:
the error map for key frame k is defined as the set:
{ek,p|p＝1,...,N}；
if all blocks in the key frame k are considered to be real, moving to the next window and repeating the steps; if the tampering rate is lower than the threshold rho, TR < rho, recovering the tampered block, recovering the reference frame of the key frame by extracting the recovery watermark embedded in the key section, and replacing the block which is considered to be tampered in the key frame with the corresponding block in the reference frame;
under the condition that rho < TR < 1, the tampered block cannot be recovered, the decoder can only locate the tampering, and the error map represents the position of the tampered block; under the condition that TR is 1, extracting a frame index in the same frame window, if the extracted frame index indicates that the frame belongs to the selected window and is correctly sequenced, searching the key frame again and determining the number of the key frame, and repeating the steps; otherwise, the decoder judges that the frame is tampered, and executes the frame tampering detection module.
8. A program storage medium for receiving user input, the stored computer program causing an electronic device to perform the steps comprising: the monitoring video is used as a deleting channel, and the video watermark consisting of the highly compressed key frame and the positioning data is transmitted through the deleting channel; embedding a watermark by fountain coding and extracting the watermark at a decoder end to realize intra-frame tampering detection recovery and inter-frame tampering detection; inter-frame tamper detection extracts frame indexes of all frames in a frame window from the last key frame verified as a real block, distinguishes inserted or deleted frames by using the extracted frame indexes, and determines the position of a new key frame; the verification and recovery process is repeated using the new key frame.
9. A video tamper recovery processing system for implementing the video tamper recovery processing method according to any one of claims 1 to 7, the video tamper recovery processing system comprising:
a deletion channel establishing module for taking the monitoring video as a deletion channel;
the video watermark composition module is used for transmitting the video watermark consisting of the highly compressed key frames and the positioning data through a deletion channel;
the watermark embedding and extracting module is used for embedding the watermark by adopting fountain coding and extracting the watermark at a decoder end;
and the recovery and detection module is used for realizing intra-frame tampering detection recovery and inter-frame tampering detection.
10. An encoder, characterized in that the encoder is equipped with the video tamper recovery processing system of claim 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010197707.1A CN111510672A (en) | 2020-03-19 | 2020-03-19 | Video tampering recovery processing method, system, storage medium and encoder |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010197707.1A CN111510672A (en) | 2020-03-19 | 2020-03-19 | Video tampering recovery processing method, system, storage medium and encoder |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111510672A true CN111510672A (en) | 2020-08-07 |
Family
ID=71871702
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010197707.1A Pending CN111510672A (en) | 2020-03-19 | 2020-03-19 | Video tampering recovery processing method, system, storage medium and encoder |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111510672A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114125469A (en) * | 2021-11-08 | 2022-03-01 | 北京天宇威视科技股份有限公司 | MP4 file repair method and device, electronic equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105488434A (en) * | 2015-12-03 | 2016-04-13 | 江苏师范大学 | Mark based vector map completeness authentication method |
| CN108650491A (en) * | 2018-05-15 | 2018-10-12 | 西安电子科技大学 | A kind of video watermark detection method towards monitoring system |
| CN109412754A (en) * | 2018-10-22 | 2019-03-01 | 北京理工大学 | A kind of data storage, distribution and access method encoding cloud |
-
2020
- 2020-03-19 CN CN202010197707.1A patent/CN111510672A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105488434A (en) * | 2015-12-03 | 2016-04-13 | 江苏师范大学 | Mark based vector map completeness authentication method |
| CN108650491A (en) * | 2018-05-15 | 2018-10-12 | 西安电子科技大学 | A kind of video watermark detection method towards monitoring system |
| CN109412754A (en) * | 2018-10-22 | 2019-03-01 | 北京理工大学 | A kind of data storage, distribution and access method encoding cloud |
Non-Patent Citations (1)
| Title |
|---|
| VAHIDEH AMANIPOUR: "Video-Tampering Detection and Content Reconstruction via Self-Embedding", 《IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114125469A (en) * | 2021-11-08 | 2022-03-01 | 北京天宇威视科技股份有限公司 | MP4 file repair method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6285775B1 (en) | Watermarking scheme for image authentication | |
| Hsu et al. | DCT-based watermarking for video | |
| US8032758B2 (en) | Content authentication and recovery using digital watermarks | |
| Sadek et al. | Robust video steganography algorithm using adaptive skin-tone detection | |
| Chamlawi et al. | Digital image authentication and recovery: employing integer transform based information embedding and extraction | |
| Tabassum et al. | A digital video watermarking technique based on identical frame extraction in 3-Level DWT | |
| Ravi et al. | Compression noise based video forgery detection | |
| Su et al. | A practical design of digital video watermarking in H. 264/AVC for content authentication | |
| Amanipour et al. | Video-tampering detection and content reconstruction via self-embedding | |
| Kumar et al. | An improved data-hiding approach using skin-tone detection for video steganography | |
| Ponni alias Sathya et al. | Non‐redundant frame identification and keyframe selection in DWT‐PCA domain for authentication of video | |
| US9087377B2 (en) | Video watermarking method resistant to temporal desynchronization attacks | |
| CN111510672A (en) | Video tampering recovery processing method, system, storage medium and encoder | |
| Chen et al. | H. 264 video authentication based on semi-fragile watermarking | |
| CN116095341A (en) | Watermark embedding method, device, equipment and storage medium | |
| CN114598887A (en) | Anti-recompression video watermarking method for controlling bit rate increase | |
| Kumar et al. | Video quality evaluation using DWT-SPIHT based watermarking technique | |
| CN113596575A (en) | Video fingerprint data generation method, comparison method and equipment | |
| Sakib et al. | A robust DWT-based compressed domain video watermarking technique | |
| Tsai et al. | Highly imperceptible video watermarking with the Watson's DCT-based visual model | |
| Agilandeeswari et al. | A novel block based video in video watermarking algorithm using discrete wavelet transform and singular value decomposition | |
| EP2544143B1 (en) | Method for watermark detection using reference blocks comparison | |
| KR100500144B1 (en) | Method of embeding and extracting digital watermark for authenticating digitial image | |
| Ogawa et al. | Digital watermarking technique for motion pictures based on quantization | |
| CN111182365B (en) | Compressed domain video watermark embedding and extracting method based on visual sensitive block |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200807 |
|
| RJ01 | Rejection of invention patent application after publication |