CN105554509A - Video inter-frame tampering detection method based on macro-block type change characteristic - Google Patents

Abstract

The present invention provides a video inter-frame tampering detection method based on a macro-block type change characteristic. The method comprises a step of extracting the number of each type of macro blocks in the P frame and B frame of a video to be detected, a step of extracting the abnormal video frames in the video according to an extracted macro block quantity characteristic, a step of judging whether the video is tampered or not according to the quality of the obtained abnormal video frames, and judging a tampering type through the periodical detection of the abnormal frames for a video with a tampering suspect. According to the method, the inter-frame tampering type (comprising frame deletion and frame insertion) can be detected, the method has a good detection effect for the tampering video with MPEG-2, MPEG-4, H. 264 formats, at the same time, since related characteristics can be obtained in the decoding process, the detection efficiency of the algorithm is high.

Description

Based on macro block (mb) type variation characteristic frame of video between altering detecting method

Technical field

The present invention relates to video tamper detection method technical field, particularly, relate to altering detecting method between a kind of frame of video based on macro block (mb) type variation characteristic.

Background technology

Along with the fast development of the Internet and the continuous maturation of multimedia technology; digital multimedia particularly image and video file starts to enter daily life in a large number; the appearance of powerful authoring tool (such as Photoshop, video editing special edition); people are made to be revised original image, editing original video more easily; at these tool belts to people simultaneously easily; also result in very large hidden danger; in some particular surroundings such as individual privacy, judicial evidence collection, social stability, distort video and often can bring very large social effectiveness.Therefore video is become especially important with or without the judgement of distorting.Interframe tampering (frame deletion, frame insert) distorts as video the attention that the most basic in means, modal a kind of means of distorting receive Chinese scholars.

Existing inter-frame video tampering detection means are roughly divided into coding and non-coding two aspects.The detection means of non-coding aspect is generally detected video by the discontinuity of video frame content, video decode is needed to become picture, then extract the feature (such as optical flow field, velocity field etc.) of video content aspect, correlation technique can consume a large amount of computational resources in extraction feature.What adopt herein is the detection method of coding aspect, in the video tampering detection patent of the coding aspect published at present, following several sections are had to have certain similitude with context of methods: publication number is CN104469361A, the patent being entitled as " a kind of video with Motion Adaptive deletes frame evidence collecting method " quantizes the cymomotive force of P frame motion residuals data in video sequence, judge that deleting frame line is that the method is not good to the video Detection results of motion intense according to cymomotive force sequence; Publication number is CN103533377A, the patent being entitled as " a kind of based on H.264/AVC video delete frame altering detecting method " has different time domains and frequency domain characteristic according to the mean residual sequence of P frame before and after video is distorted, time domain specification and frequency domain characteristic are classified as double characteristic, the verification and measurement ratio of the method haves much room for improvement.Publication number is CN104093033A, be entitled as the patent of " method that number of frames is estimated is identified and deleted to a kind of frame of deleting of H.264/AVC video " by calculating the blurring effect sequence of video, be averaged in units of GOP and obtain MBAS sequence, and it can be used as characteristic use grader to carry out deleting the judgement of frame.Three kinds of above-mentioned methods all can only detect the frame deletion in interframe tampering, and what do not contain another kind of common frame insertion distorts pattern, has some limitations.

Summary of the invention

For defect of the prior art, the object of this invention is to provide altering detecting method between a kind of frame of video based on macro block (mb) type variation characteristic.

Based on an inter-frame video altering detecting method for macro block (mb) type variation characteristic, comprise the steps:

Step 1: decode operation is carried out to video, all kinds of number of macroblocks in statistics video P frame, B frame;

Step 2: extract abnormal P frame, the B frame in video sequence, obtain the abnormal frame sequence of former video;

Step 3: according to the quantity of abnormal frame, judges whether video is distorted through interframe;

Step 4: periodicity is carried out to the video that there is interframe and distort suspicion and detects, judge the type that video is distorted, described in the type of distorting comprise: frame deletion and frame insertion.

Preferably, step 1 comprises the steps:

Step 1.1: the coded format different according to video is chosen corresponding decoder and carried out partial decoding of h to video;

Step 1.2: decoding obtains the type of coding of each macro block in video P frame, B frame in units of the minimum macroblock size of current coding format;

Particularly, if the minimum macroblock size of current coding format is N × N, be divided into the size of N × 2N or 2N × 2N in some regions, then the macro block of N × 2N be denoted as the macro block of 2 N × N, the macro block of 2N × 2N is then denoted as the macro block of 4 N × N;

Step 1.3: the quantity of all kinds of macro block in statistic of classification P frame, B frame, the method for expressing of statistic is as follows:

Remember the n-th P frame: inter-coded macroblocks quantity i (n), jump coding number of macroblocks s (n);

Remember the n-th B frame: forward direction reference macroblock quantity b ₁(n), backward reference macroblock quantity b ₂(n).

Preferably, step 2 comprises the steps:

Step 2.1: for the statistic of associated macroblock in the B frame that step 1.3 obtains, calculates the ratio p (n) that forward direction reference macroblock accounts for unidirectional reference macroblock, as shown in formula (1):

$p\left(n\right)=\frac{b_1\left(n\right)}{b_1\left(n\right)+b_2\left(n\right)}---\left(1\right)$

Step 2.2: the abnormal frame being extracted P frame, B frame by criterion,

The criterion of P abnormal frame is:

i(n-1)<i(n)∩i(n)>i(n+1)∩s(n-1)>s(n)∩s(n)<s(n+1)

In formula: i (n-1) represents the inter-coded macroblocks quantity of (n-1)th P frame, s (n-1) represents the jump coding number of macroblocks of (n-1)th P frame, i (n+1) represents the inter-coded macroblocks quantity of (n+1)th P frame, and s (n+1) represents the jump coding number of macroblocks of (n+1)th P frame;

The criterion of B abnormal frame is:

p(n)<min{p(i)|n-j-1<i<n}

Step 2.3: by being judged as in step 2.2 that the frame of video of abnormal frame reverts to former video sequence, obtain abnormal frame sequence F (n), expression formula is as shown in the formula (2)

Wherein, the span of n is [1, N], and N is video totalframes.

Preferably, step 3 comprises the steps:

Step 3.1: the abnormal frame sequence of adding up according to step 2.3, the quantity M of P frame and B frame in statistics Video coding, statistics video abnormal frame quantity D, the computational methods of abnormal frame quantity D are as shown in the formula shown in (3):

$D={\mathrm{\&Sigma;}}_{i=1}^{N}F\left(i\right)---\left(3\right)$

Step 3.2: the D/M value calculating test video, D/M value is compared with threshold value T1, judge whether to distort operation through interframe according to comparative result, wherein, described D/M value accounts for the ratio of the total P frame of test video, B number of frames for the frame of video being judged as abnormal frame, wherein, the interval of threshold value T1 is (0,1);

When D/M value is more than or equal to threshold value T1, then this video is judged to distort operation through interframe;

When D/M value is less than threshold value T1, then this video is judged to be original video.

Preferably, step 4 comprises the steps:

Step 4.1: for the video being judged as in step 3 distorting through interframe, calculates frame deletion Suspected Degree G;

Step 4.2: frame deletion Suspected Degree G is compared with threshold value T2, judge to distort type according to the result compared, wherein, the interval of threshold value T2 is (0,1);

If G is more than or equal to threshold value T2, then this detection video is distorted type and is judged to be frame deletion;

If G is less than threshold value T2, then this detection video is distorted type and is judged to be that frame inserts.

6, altering detecting method between the frame of video based on macro block (mb) type variation characteristic according to claim 5, it is characterized in that, described step 4.1 comprises:

Step 4.1.1: this abnormal frame is calculated as deleting the Suspected Degree of point about cycle T, as shown in formula (4) to each abnormal frame:

$G(T,k)=(\underset{i=T+1}{\overset{k}{\&Sigma;}}F\left(i\right)*F\left(i-T\right)+\underset{i=k+T+1}{\overset{N}{\&Sigma;}}F\left(i\right)*F\left(i-T\right))/N---\left(4\right)$

In formula: G (T, k) expression is in the Suspected Degree in the T cycle about k position, parameter k represents the sequence number of this abnormal frame in former sequence of frames of video;

Step 4.1.2: calculate with each abnormal frame for deleting frame deletion Suspected Degree G (k) corresponding to point, as shown in formula (5):

$G\left(k\right)=\underset{T}{max}G(T,k)---\left(5\right)$

Step 4.1.3: extract the frame deletion Suspected Degree that maximum G (k) is this video, as shown in formula (6):

$G=\underset{k}{max}G\left(k\right)---\left(6\right)$

Compared with prior art, the present invention has following beneficial effect:

1, the method in the present invention can not only monitor the P frame of video, and the feature can also extracting B frame detects, and experiment shows that the method in the present invention has good Detection results at the video of MPEG-2, MPEG-4, H.264 form.

2, the inter-frame video altering detecting method based on macro block (mb) type variation characteristic provided by the invention directly can obtain correlated characteristic in decode procedure, and therefore detection speed is fast.

3, the present invention is while whether detection video is distorted, and can judge distorting between concrete frame of video means (frame deletion, frame insert) and distinguish.

Accompanying drawing explanation

By reading the detailed description done non-limiting example with reference to the following drawings, other features, objects and advantages of the present invention will become more obvious:

Fig. 1 is model framework figure provided by the invention;

Fig. 2 is the flow chart of abnormal frame sequential extraction procedures provided by the invention.

Embodiment

Below in conjunction with specific embodiment, the present invention is described in detail.Following examples will contribute to those skilled in the art and understand the present invention further, but not limit the present invention in any form.It should be pointed out that to those skilled in the art, without departing from the inventive concept of the premise, some distortion and improvement can also be made.These all belong to protection scope of the present invention.

As shown in Figure 1, according to the inter-frame video altering detecting method based on macro block (mb) type variation characteristic provided by the invention, comprise the steps:

Step 1: decode operation is carried out to video, all kinds of number of macroblocks in statistics video P frame, B frame;

Step 2: extract abnormal P frame, the B frame in video sequence, obtain the abnormal frame sequence of former video;

Step 3: according to the quantity of abnormal frame, judges whether video is distorted through interframe;

Step 4: for being judged as that interframe distorts the video of suspicion, by periodically detecting, judges that video distorts type (frame deletion and frame insert).

Preferably, described step 1 comprises the steps:

Step 1.1: according to the difference of video code model, chooses corresponding decoder and carries out partial decoding of h;

Step 1.2: in units of the minimum macroblock size of current coding format, decoding obtains the type of coding of each macro block in video P frame, B frame;

Particularly, if the minimum macroblock size of current coding format is N × N, be divided into the size of N × 2N or 2N × 2N in some regions, then the macro block of N × 2N be denoted as the macro block of 2 N × N, the macro block of 2N × 2N is then denoted as the macro block of 4 N × N, by that analogy;

Step 1.3: the quantity of all kinds of macro block in statistic of classification P frame, B frame, statistic is as follows:

N-th P frame: inter-coded macroblocks quantity i (n), jump coding number of macroblocks s (n)

N-th B frame: forward direction reference macroblock quantity b ₁(n), backward reference macroblock quantity b ₂(n).

Particularly, Fig. 2 is the flow chart of abnormal frame sequential extraction procedures, and described step 2 comprises the steps:

Step 2.1: for the statistic of associated macroblock in the B frame that step 1.3 obtains, calculates the ratio that forward direction reference macroblock accounts for unidirectional reference macroblock:

$p\left(n\right)=\frac{b_1\left(n\right)}{b_1\left(n\right)+b_2\left(n\right)}---\left(1\right)$

Step 2.2: the abnormal frame being extracted P frame, B frame by criterion:

The criterion of P abnormal frame is:

i(n-1)<i(n)∩i(n)>i(n+1)∩s(n-1)>s(n)∩s(n)<s(n+1)；

The criterion of B abnormal frame is:

p(n)<min{p(i)|n-j-1<i<n}

Step 2.3: for the frame of video being judged as abnormal frame in step 2.2, is reverted to former video sequence, and obtain abnormal frame sequence and express, expression formula is as shown in the formula (2)

Wherein, the span of n is [1, N], and N is video totalframes.

Preferably, described step 3 comprises the steps:

Step 3.1: express according to the abnormal frame sequence that step 2.3 is added up, the quantity M of P frame, B frame in statistics Video coding, statistics video abnormal frame quantity D, the computational methods of abnormal frame quantity D are as shown in the formula (3):

$D={\mathrm{\&Sigma;}}_{i=1}^{N}F\left(i\right)---\left(3\right)$

Step 3.2: the D/M value calculating test video, D/M value is compared with threshold value T1, judge whether to distort operation through interframe according to comparative result, wherein, described D/M value accounts for the ratio of the total P frame of test video, B number of frames for the frame of video being judged as abnormal frame, the interval of threshold value T1 is (0,1):

When D/M value is more than or equal to threshold value T1, then this video is judged to distort operation through interframe;

When D/M value is less than threshold value T1, then this video is judged to be original video.

Preferably, described step 4 comprises the steps:

Step 4.1: for the video being judged as the suspicion of distorting in step 3, calculate frame deletion Suspected Degree G, the computational process of frame deletion Suspected Degree G is as follows:

Step 4.1.1: this abnormal frame is calculated as deleting the Suspected Degree of point about cycle T, as shown in formula (4) to each abnormal frame:

$G(T,k)=(\underset{i=T+1}{\overset{k}{\&Sigma;}}F\left(i\right)*F\left(i-T\right)+\underset{i=k+T+1}{\overset{N}{\&Sigma;}}F\left(i\right)*F\left(i-T\right))/N---\left(4\right)$

In formula: G (T, k) expression is in the Suspected Degree in the T cycle about k position, parameter k represents the sequence number of this abnormal frame in former sequence of frames of video;

Step 4.1.2: calculate with each abnormal frame for deleting frame deletion Suspected Degree G (k) corresponding to point, as shown in formula (5):

$G\left(k\right)=\underset{T}{max}G(T,k)---\left(5\right)$

Step 4.1.3: extract the frame deletion Suspected Degree that maximum G (k) is this video, as shown in formula (6):

$G=\underset{k}{max}G\left(k\right)---\left(6\right)$

Step 4.2: frame deletion Suspected Degree G is compared with threshold value T2, judge to distort type according to the result compared, the interval of threshold value T2 is (0,1);

If G is more than or equal to threshold value T2, then this detection video is distorted type and is judged to be frame deletion;

If G is less than threshold value T2, then this detection video is distorted type and is judged to be that frame inserts.

Above specific embodiments of the invention are described.It is to be appreciated that the present invention is not limited to above-mentioned particular implementation, those skilled in the art can make various distortion or amendment within the scope of the claims, and this does not affect flesh and blood of the present invention.

Claims (6)

1. based on macro block (mb) type variation characteristic frame of video between an altering detecting method, it is characterized in that, comprise the steps:

Step 1: decode operation is carried out to video, all kinds of number of macroblocks in statistics video P frame, B frame;

Step 2: extract abnormal P frame, the B frame in video sequence, obtain the abnormal frame sequence of former video;

Step 3: according to the quantity of abnormal frame, judges whether video is distorted through interframe;

Step 4: periodicity is carried out to the video that there is interframe and distort suspicion and detects, judge the type that video is distorted, described in the type of distorting comprise: frame deletion and frame insertion.

2. altering detecting method between the frame of video based on macro block (mb) type variation characteristic according to claim 1, it is characterized in that, described step 1 comprises the steps:

Step 1.1: the coded format different according to video is chosen corresponding decoder and carried out partial decoding of h to video;

Step 1.2: decoding obtains the type of coding of each macro block in video P frame, B frame in units of the minimum macroblock size of current coding format;

Particularly, if the minimum macroblock size of current coding format is N × N, be divided into the size of N × 2N or 2N × 2N in some regions, then the macro block of N × 2N be denoted as the macro block of 2 N × N, the macro block of 2N × 2N is then denoted as the macro block of 4 N × N;

Step 1.3: the quantity of all kinds of macro block in statistic of classification P frame, B frame, the method for expressing of statistic is as follows:

Remember the n-th P frame: inter-coded macroblocks quantity i (n), jump coding number of macroblocks s (n);

Remember the n-th B frame: forward direction reference macroblock quantity b ₁(n), backward reference macroblock quantity b ₂(n).

3. altering detecting method between the frame of video based on macro block (mb) type variation characteristic according to claim 2, it is characterized in that, described step 2 comprises the steps:

Step 2.1: for the statistic of associated macroblock in the B frame that step 1.3 obtains, calculates the ratio p (n) that forward direction reference macroblock accounts for unidirectional reference macroblock, as shown in formula (1):

Step 2.2: the abnormal frame being extracted P frame, B frame by criterion,

The criterion of P abnormal frame is:

i(n-1)<i(n)∩i(n)>i(n+1)∩s(n-1)>s(n)∩s(n)<s(n+1)

In formula: i (n-1) represents the inter-coded macroblocks quantity of (n-1)th P frame, s (n-1) represents the jump coding number of macroblocks of (n-1)th P frame, i (n+1) represents the inter-coded macroblocks quantity of (n+1)th P frame, and s (n+1) represents the jump coding number of macroblocks of (n+1)th P frame;

The criterion of B abnormal frame is:

p(n)<min{p(i)|n-j-1<i<n}

Step 2.3: by being judged as in step 2.2 that the frame of video of abnormal frame reverts to former video sequence, obtain abnormal frame sequence F (n), expression formula is as shown in the formula (2)

Wherein, the span of n is [1, N], and N is video totalframes.

4. altering detecting method between the frame of video based on macro block (mb) type variation characteristic according to claim 3, it is characterized in that, described step 3 comprises the steps:

Step 3.1: the abnormal frame sequence of adding up according to step 2.3, the quantity M of P frame and B frame in statistics Video coding, statistics video abnormal frame quantity D, the computational methods of abnormal frame quantity D are as shown in the formula shown in (3):

Step 3.2: the D/M value calculating test video, D/M value is compared with threshold value T1, judge whether to distort operation through interframe according to comparative result, wherein, described D/M value accounts for the ratio of the total P frame of test video, B number of frames for the frame of video being judged as abnormal frame, wherein, the interval of threshold value T1 is (0,1);

When D/M value is more than or equal to threshold value T1, then this video is judged to distort operation through interframe;

When D/M value is less than threshold value T1, then this video is judged to be original video.

5. altering detecting method between the frame of video based on macro block (mb) type variation characteristic according to claim 1, it is characterized in that, described step 4 comprises the steps:

Step 4.1: for the video being judged as in step 3 distorting through interframe, calculates frame deletion Suspected Degree G;

Step 4.2: frame deletion Suspected Degree G is compared with threshold value T2, judge to distort type according to the result compared, wherein, the interval of threshold value T2 is (0,1);

If G is more than or equal to threshold value T2, then this detection video is distorted type and is judged to be frame deletion;

If G is less than threshold value T2, then this detection video is distorted type and is judged to be that frame inserts.

6. altering detecting method between the frame of video based on macro block (mb) type variation characteristic according to claim 5, it is characterized in that, described step 4.1 comprises:

Step 4.1.1: this abnormal frame is calculated as deleting the Suspected Degree of point about cycle T, as shown in formula (4) to each abnormal frame:

In formula: G (T, k) expression is in the Suspected Degree in the T cycle about k position, parameter k represents the sequence number of this abnormal frame in former sequence of frames of video;

Step 4.1.2: calculate with each abnormal frame for deleting frame deletion Suspected Degree G (k) corresponding to point, as shown in formula (5):

Step 4.1.3: extract the frame deletion Suspected Degree that maximum G (k) is this video, as shown in formula (6):

。