CN102065286B - Motion vector-based video watermark embedding and extracting method - Google Patents

Abstract

The invention discloses a motion vector-based video watermark embedding and extracting method. The method comprises the following watermark embedding steps of: a, during coding, acquiring ith motion vector (Xi, Yi) of a video frame of a coding end; b, acquiring the characteristic vector (V1i, V2i) of the motion vector (Xi, Yi) of the coding end according to the combination relationship of decimal parts of the transverse coordinate Xi and the longitudinal coordinate Yi of the motion vector (Xi, Yi) of the coding end; c, adjusting the transverse coordinate Xi or the longitudinal coordinate Yi of the motion vector (Xi, Yi) of the coding end according to the modification rule so that the transverse coordinate V1i and the longitudinal coordinate V2i of the characteristic vector (V1i, V2i) of the motion vector of the coding end are as same as Wk and Wk+1 of two watermarks to be embedded; the method comprises the following watermark extracting steps of: d, during decoding, acquiring ith motion vector (Xi, Yi) of a video frame of a decoding end; e, acquiring the characteristic vector (V1i, V2i) of the motion vector (Xi, Yi) of the decoding end according to the combination relationship of decimal parts of the transverse coordinate Xi and the longitudinal coordinate Yi of the motion vector (Xi, Yi) of the decoding end; and f, extracting watermark information of the motion vector (Xi, Yi), wherein the Wk and the Wk+1 of the two watermarks are the transverse coordinate V1i and the longitudinal coordinate V2i of the characteristic vector (V1i, V2i) of the motion vector (Xi, Yi) of the decoding end respectively.

Description

Video watermark based on motion vector embeds and method for distilling

Technical field

The present invention relates to multimedia digital watermark embedding and method for distilling, particularly a kind of video watermark based on motion vector embeds and method for distilling.

Background technology

Along with the fast development of Computer Multimedia Technology and network technology, Digital Medias such as digital picture, video, audio frequency duplicate and propagation becomes more and more convenient.But being easy to copy of Digital Media also makes it be faced with serious safety problem.Digital watermark technology is a kind of copyright protection means that proposed in recent years, and it is through embedding visible or sightless information in Digital Media, the right of attribution that can the reference numerals media piece, illegal propagation that can also the control figure medium.

The embedding and the extraction scheme of video digital watermark mainly contain 3 kinds at present: the video digital watermark of (1) uncompressed domain; (2) video digital watermark of encoding domain; (3) video digital watermark of compression domain.The video digital watermark of uncompressed domain mainly is divided into spatial domain watermark and transform domain watermark.This scheme can be utilized the technical scheme of digital figure watermark, and watermark embedding method is many, but increases the bit rate of video code flow easily and be that watermark information possibly destroyed encoding.The video digital watermark of encoding domain generally embeds through revising DCT coefficient or motion vector.This scheme anti-attack ability is more intense, but embeds and need partly revise encoder when extracting watermark.The video digital watermark of compression domain is directly embedded into watermark in the video code flow after the encoding compression.This scheme need not decoded and encoded, but the embedded quantity of watermark is fewer.

Existing through revising the method for motion vector embed watermark, the abscissa that has plenty of through revising motion vector embeds a watermark with in the ordinate one or two, and the abscissa and the ordinate that have plenty of through the modification motion vector embed two watermarks.Though a kind of method in back can embed two watermarks through revising a motion vector, needs to revise the abscissa and the ordinate of motion vector, and is bigger to the damage of original video.

Summary of the invention

The objective of the invention is to, solve in the existing video watermark embedding and method for distilling, the bigger technical problem of original video damage based on motion vector.

For achieving the above object, the present invention provides a kind of video watermark based on motion vector to embed and method for distilling, comprises watermark embedding and watermark extracting two big steps,

Described watermark embed step is:

Step a when coding, obtains i motion vector (X of coding side frame of video _i, Y _i), i ∈ 1,2 ..., m}, X _iThe expression abscissa, Y _iThe expression ordinate, m is the sum of motion vector in the frame of video;

Step b is according to said coding side motion vector (X _i, Y _i) abscissa X _iWith ordinate Y _iThe syntagmatic of fractional part, obtain said coding side motion vector (X _i, Y _i) characteristic vector (V1 _i, V2 _i);

Step c is according to revising the said coding side motion vector (X of rule adjustment _i, Y _i) abscissa X _iOr ordinate Y _iOne of them makes the characteristic vector (V1 of said coding side motion vector _i, V2 _i) abscissa V1 _i, ordinate V2 _iWith two watermark W that will embed _k, W _K+1Identical;

Described watermark extracting step is:

Steps d when decoding, is obtained i motion vector (X of decoding end frame of video _i, Y _i);

Step e is according to said decoding end motion vector (X _i, Y _i) abscissa X _iWith ordinate Y _iThe syntagmatic of fractional part, obtain said decoding end motion vector (X _i, Y _i) characteristic vector (V1 _i, V2 _i);

Step f extracts motion vector (X _i, Y _i) in watermark information, two watermark W _k, W _K+1Be respectively decoding end motion vector (X _i, Y _i) characteristic vector (V1 _i, V2 _i) abscissa V1 _i, ordinate V2 _i

Beneficial effect of the present invention is, through comparing motion vector abscissa and ordinate and treat the relation of embed watermark, revises the abscissa or the ordinate of motion vector, is implemented in and embeds two watermarks in the motion vector.The present invention is embed watermark in the bigger motion vector of amplitude, and the amplitude of motion vector is bigger, explains that object motion is violent in the frame of video, and the picture that amended motion vector recovers has trickle distortion, and is not obvious in the scene of strenuous exercise.Watermark embedding method of the present invention is simple, only need simply judge with plus and minus calculation motion vector to get final product embed watermark, and computation complexity is little.And, good compatibility is arranged with different video compression standards.Simultaneously, guaranteeing that watermark embeds under the situation of quantity, less to the modification of motion vector, therefore less to the damage of original video.

Description of drawings

Fig. 1 the present invention is based on the video watermark embedding of motion vector and the video system sketch map of method for distilling for application;

Fig. 2 is the flow chart of the method for watermark embedding of the present invention;

Fig. 3 is the flow chart of the method for watermark extracting of the present invention.

Embodiment

To combine the accompanying drawing in the embodiment of the invention below, the technical scheme in the embodiment of the invention is carried out clear, intactly description, obviously, described embodiment only is a part of embodiment of the present invention, rather than whole embodiment.Based on embodiments of the invention, the every other embodiment that those of ordinary skills are obtained under the prerequisite of not making creative work belongs to the scope that the present invention protects.

As shown in Figure 1, for application the present invention is based on the video watermark embedding of motion vector and the video system sketch map of method for distilling.Said video system comprises coding side and decoding end.At coding side, the binary system watermark bit that watermarking images is converted into one dimension flows, and sends into video encoder, and video encoder is encoded to original video, and watermark bit stream is embedded motion vector, forms video flowing.In decoding end, Video Decoder extracts the watermark sequence that embeds in the motion vector, recovers watermarking images and frame of video.

Before the embedding of watermark and leaching process are elaborated, thought of the present invention is illustrated.The present invention is comparing motion vector abscissa and ordinate and treat the relation of embed watermark at first; Revise then motion vector abscissa or ordinate one of them; Be implemented in and embed two watermarks in the motion vector, thus less to the modification of motion vector, minimum to the damage of original video.Method of the present invention is applied in the encoding and decoding of B frame and P frame.

Carrying out before watermark embeds, the motion vector of macro block in B frame and the P frame is being sorted according to particular order, obtaining motion vector sequence.In the present embodiment, according to from left to right, rank order from top to bottom obtains the motion vector sequence { (X of macro block with the motion vector of macro block ₁, Y ₁), (X ₂, Y ₂) ..., (X _m, Y _m), m is the sum of macroblock motion vector in the current video frame.The binary system watermark bit stream that needs to embed is { W ₁, W ₂... W _n, n is the watermark sum.

For example H.264, AVS, MPEG-4 or the like the present invention can be applicable in the video compression technology that precision is 1/4 pixel.According to the syntagmatic of the fractional part of motion vector abscissa and ordinate, confirm the characteristic vector of motion vector.

The syntagmatic of the fractional part of described motion vector abscissa and ordinate does, for i motion vector (X _i, Y _i), abscissa X _iFractional part be x _i, ordinate Y _iFractional part be y _iWith abscissa X _iFractional part x _iNumerical value be enlarged into 4 times, and take absolute value, promptly calculate | x _i* 4|; With ordinate Y _iFractional part y _iNumerical value be enlarged into 4 times, and take absolute value, promptly calculate | y _i* 4|.Calculate | x _i* 4| and | y _i* the distance P between the 4| _i=mod (4+|x _i* 4|-|y _i* 4|, 4).

Because motion vector is 1/4 pixel precision, thus the abscissa of motion vector and ordinate fractional part multiply by 4 and take absolute value after be 0,1,2 or 3.Then | x _i* 4| and | y _i* the numerical value of 4| is 0,1,2 or 3.0,1,2,3 regard a circulation as, i.e. 0-3=1,0-1=3.

According to | x _i* 4|, | y _i* 4| parity and P _iValue divide and always to have 8 kinds of situation, as shown in table 1.Table 1 is abscissa, ordinate fractional part and the characteristic vector abscissa of motion vector, the corresponding relation of ordinate.

Through | x _i* the parity of 4| and P _iValue come calculated characteristics vector (V1 _i, V2 _i), the abscissa V1 of characteristic vector _i=mod (| x _i* 4|, 2), the ordinate V2 of characteristic vector _iBe P _i/ 2 integer part, as shown in table 1.

Table 1

|x i*4|	|y i*4|	P i	V1 i	V2 i
					Odd number	Odd number	0	1	0
Odd number	Even number	1	1	0
					Odd number	Odd number	2	1	1
Odd number	Even number	3	1	1
					Even number	Even number	0	0	0
Even number	Odd number		1	0						0
					Even number	Even number	2	0	1
Even number	Odd number	3	0	1

The present invention need preestablish threshold value threshold, and motion vector magnitude and described threshold value are done comparison, the carrier that amplitude embeds as watermark greater than the motion vector of said threshold value.The embedded capacity and the Y-PSNR PSNR (Peak Signal Noise Ratio) that influence watermark understood in choosing of threshold value, adjusts within the specific limits according to different requirement in the application.

Embedding and leaching process in the face of watermark is elaborated down.

As shown in Figure 2, the embedding grammar of video watermark of the present invention is:

Step 101: beginning.

Step 102: when coding, obtain i motion vector (X of coding side frame of video _i, Y _i), i ∈ 1,2 ..., m}, X _iThe expression abscissa, Y _iThe expression ordinate.

Step 103: the motion vector (X of calculation code end frame of video _i, Y _i) amplitude $A_i=\sqrt{{\mathrm{<figure-callout\; id="2"\; label="X\; i"\; filenames="H2009102381789E00021.png"\; state="\{\{state\}\}">X</figure-callout>}}_i^{}+{\mathrm{<figure-callout\; id="2"\; label="Y\; i"\; filenames="H2009102381789E00021.png"\; state="\{\{state\}\}">Y</figure-callout>}}_i^{}},$ Compare amplitude A _iWith the size of threshold value threshold, whether judge amplitude greater than threshold value, if then execution in step 104, if not then execution in step 110.

Step 104: according to coding side motion vector (X _i, Y _i) abscissa X _iWith ordinate Y _iThe syntagmatic of fractional part, obtain motion vector (X _i, Y _i) characteristic vector (V1 _i, V2 _i);

Obtain abscissa X _iFractional part x _i, and the fractional part y of ordinate Yi _i

Calculate | x _i* 4| and | y _i* the distance P between the 4| _i=mod (4+|x _i* 4|-|y _i* 4|, 4);

Through | x _i* the parity of 4| and P _iValue come calculated characteristics vector (V1 _i, V2 _i), the abscissa V1 of characteristic vector _i=mod (| x _i* 4|, 2), the ordinate V2 of characteristic vector _iBe P _i/ 2 integer part.

Step 105: through adjustment motion vector (X _i, Y _i) abscissa Xi or ordinate Y _iOne of them makes the characteristic vector (V1 of this motion vector _i, V2 _i) abscissa V1 _i, ordinate V2 _iWith two watermark W that will embed _k, W _K+1Identical, expression has embedded two watermark W in this motion vector _k, W _K+1Particularly, through to motion vector (X _i, Y _i) abscissa X _iWith ordinate Y _iIn one add reducing, vow (X a motion _i, Y _i) embed two watermark W in the amount _k, W _K+1

According to abscissa V1 _i, ordinate V2 _iWith two watermark W that will embed _k, W _K+1Relation revise the abscissa X of motion vector _iOr ordinate Y _i, revise rule as follows:

Work as V1 _i=W _k, V2 _i=W _K+1And mod (P _i, 2)=0 o'clock, V1 perhaps worked as _i=W _k, V2 _i=W _K+1And mod (P _i, 2)=1 o'clock, abscissa X _iWith ordinate Y _iRemain unchanged;

Work as V1 _i=W _k, V2 _i≠ W _K+1And mod (P _i, 2)=0 o'clock, ordinate Y revised _iFor

Work as V1 _i=W _k, V2 _i≠ W _K+1And mod (P _i, 2)=1 o'clock, ordinate Y revised _iFor $(Y_i-\mathrm{Sgn}\left(Y_i\right)*\frac{1}{4})*(1-\mathrm{\&delta;}\left(Y_i\right))+\frac{3}{4}*\mathrm{\&delta;}\left(Y_i\right);$

Work as V1 _i≠ W _k, V2 _i=W _K+1And mod (P _i, 2)=0 o'clock, V1 perhaps worked as _i≠ W _k, V2 _i≠ W _K+1And mod (P _i, 2)=1 o'clock, the modification abscissa does

Work as V1 _i≠ W _k, V2 _i=W _K+1And mod (P _i, 2)=1 o'clock, V1 perhaps worked as _i≠ W _k, V2 _i≠ W _K+1And mod (P _i, 2)=0 o'clock, the modification abscissa does $(X_i-\mathrm{Sgn}\left(X_i\right)*\frac{1}{4})*(1-\mathrm{\&delta;}\left(X_i\right))+\frac{3}{4}*\mathrm{\&delta;}\left(X_i\right);$

Wherein, $\mathrm{Sgn}\left(X_i\right)=\left\{\begin{array}{cc}1& X_i\&GreaterEqual;0\\ 0& X_i<0\end{array}\right.,$ $\mathrm{Sgn}\left(Y_i\right)=\left\{\begin{array}{cc}1& Y_i\&GreaterEqual;0\\ 0& Y_i<0\end{array}\right.,$ $\mathrm{\&delta;}\left(X_i\right)=\left\{\begin{array}{cc}1& X_i=0\\ 0& X_i\&NotEqual;0\end{array}\right.,$

$\mathrm{\&delta;}\left(Y_i\right)=\left\{\begin{array}{cc}1& Y_i=0\\ 0& Y_i\&NotEqual;0\end{array}\right..$

Step 106 is calculated the amplitude of amended motion vector $A_i=\sqrt{{\mathrm{<figure-callout\; id="2"\; label="X\; i"\; filenames="H2009102381789E00021.png"\; state="\{\{state\}\}">X</figure-callout>}}_i^{}+{\mathrm{<figure-callout\; id="2"\; label="Y\; i"\; filenames="H2009102381789E00021.png"\; state="\{\{state\}\}">Y</figure-callout>}}_i^{}},$ Compare amplitude A _iWith the size of threshold value threshold, whether judge amplitude greater than threshold value, if then execution in step 107, if not then execution in step 108.

Step 107, k=k+2, promptly expression is with two current watermark W _k, W _K+1Embed motion vector (X _i, Y _i), will embed two new watermarks.

Step 108, k remains unchanged, and expression is with two current watermark W _k, W _K+1Embed motion vector (X _i, Y _i), and with two watermark W _k, W _K+1Embed next motion vector once more.The motion vector that prevents to satisfy condition originally causes when video watermark extracts, just not extract the watermark that is embedded in this motion vector through not satisfying the condition greater than threshold value after revising.

Step 109 judges that whether k is greater than watermark sum n, if then finish; If not, execution in step 110 then.

Step 110, i=i+1, expression is to next motion vector embed watermark.

Step 111 judges that whether i is greater than motion vector sum m, if then finish; If not, execution in step 102 then.

As shown in Figure 3, the method for distilling of video watermark of the present invention is:

Step 201: beginning.

Step 202: when decoding, obtain the motion vector (X of decoding end frame of video _i, Y _i), i ∈ 1,2 ..., m}, X _iThe expression abscissa, Y _iThe expression ordinate.

Step 203: the motion vector (X of computes decoded end frame of video _i, Y _i) amplitude $A_i=\sqrt{{\mathrm{<figure-callout\; id="2"\; label="X\; i"\; filenames="H2009102381789E00021.png"\; state="\{\{state\}\}">X</figure-callout>}}_i^{}+{\mathrm{<figure-callout\; id="2"\; label="Y\; i"\; filenames="H2009102381789E00021.png"\; state="\{\{state\}\}">Y</figure-callout>}}_i^{}},$ Compare amplitude A _iWith the size of threshold value threshold, whether judge amplitude greater than threshold value, if then execution in step 204, if not then execution in step 207.

Step 204: according to decoding end motion vector (X _i, Y _i) abscissa X _iWith ordinate Y _iThe syntagmatic of fractional part, extract decoding end motion vector (X _i, Y _i) in watermark information;

Obtain abscissa X _iFractional part x _i, and ordinate Y _iFractional part y _i

Calculate | x _i* 4| and | y _i* the distance P between the 4| _i=mod (4+|x _i* 4|-|y _i* 4|, 4);

Calculated characteristics vector (V1 _i, V2 _i), the abscissa V1 of characteristic vector _i=mod (| x _i* 4|, 2), the ordinate V2 of characteristic vector _iBe P _i/ 2 integer part;

Motion vector (X then _i, Y _i) in watermark information be W _k=V1 _i, W _K+1=V2 _i

Step 205:k=k+2, i.e. expression will be extracted two new watermarks.

Step 206: judge k whether greater than watermark sum n, if termination routine then; If not, execution in step 207 then.

Step 207:i=i+1, expression is extracted watermark to next motion vector.

Step 208: judge that whether i is greater than motion vector sum m, if then finish; If not, execution in step 202 then.

The present invention is embed watermark in the bigger motion vector of amplitude, and the amplitude of motion vector is bigger, explains that object motion is violent in the frame of video, and the picture that amended motion vector recovers has trickle distortion, and is not obvious in the scene of strenuous exercise.Watermark embedding method of the present invention is simple, only need simply judge with plus and minus calculation motion vector to get final product embed watermark, and computation complexity is little.And, good compatibility is arranged with different video compression standards.Simultaneously, guaranteeing that watermark embeds under the situation of quantity, less to the modification of motion vector, therefore less to the damage of original video.

The above description of this invention is illustrative, and nonrestrictive, and those skilled in the art is understood, and within spirit that claim limits and scope, can carry out many modifications, variation or equivalence to it, but they will fall in protection scope of the present invention all.

Claims (7)

1. the video watermark based on motion vector embeds and method for distilling, comprises watermark embedding and watermark extracting two big steps, it is characterized in that described watermark embed step is:

Step a when coding, obtains i motion vector (X of coding side frame of video _i, Y _i), i ∈ 1,2 ..., m}, X _iThe expression abscissa, Y _iThe expression ordinate, m is the sum of motion vector in the frame of video;

Step b is according to said coding side motion vector (X _i, Y _i) abscissa X _iWith ordinate Y _iThe syntagmatic of fractional part, obtain said coding side motion vector (X _i, Y _i) characteristic vector (V1 _i, V2 _i);

Step c is according to revising the said coding side motion vector (X of rule adjustment _i, Y _i) abscissa X _iOr ordinate Y _iOne of them makes the characteristic vector (V1 of said coding side motion vector _i, V2 _i) abscissa V1 _i, ordinate V2 _iWith two watermark W that will embed _k, W _K+1Identical;

Described watermark extracting step is:

Steps d when decoding, is obtained i motion vector (X of decoding end frame of video _i, Y _i);

Step e is according to said decoding end motion vector (X _i, Y _i) abscissa X _iWith ordinate Y _iThe syntagmatic of fractional part, obtain said decoding end motion vector (X _i, Y _i) characteristic vector (V1 _i, V2 _i);

Step f extracts motion vector (X _i, Y _i) in watermark information, two watermark W _k, W _K+1Be respectively decoding end motion vector (X _i, Y _i) characteristic vector (V1 _i, V2 _i) abscissa V1 _i, ordinate V2 _i

Motion vector (the X that also comprises calculation code end frame of video between described step a and the step b _i, Y _i) amplitude

Judge that described amplitude is whether greater than the step of a predetermined threshold value, if then execution in step b if not, then obtains next motion vector;

In the step that described watermark embeds, after described step c, also comprise

Calculate the amplitude of amended motion vector Judge amended amplitude A _iWhether greater than the step of said threshold value, if, then with two current watermark W _k, W _K+1Embed motion vector (X _i, Y _i), if not, then with two current watermark W _k, W _K+1Embed motion vector (X _i, Y _i), and with two watermark W _k, W _K+1Embed next motion vector once more;

In the step of described watermark extracting, after described steps d, before the step e, also comprise the motion vector (X of computes decoded end frame of video _i, Y _i) amplitude Judge that described amplitude is whether greater than the step of a predetermined threshold value, if then execution in step e if not, then obtains next motion vector.

2. the video watermark based on motion vector as claimed in claim 1 embeds and method for distilling, it is characterized in that, described step b does,

Obtain said coding side motion vector (X _i, Y _i) abscissa X _iFractional part x _i, and ordinate Y _iFractional part y _i

Calculate | x _i* 4| and | y _i* the distance P between the 4| _i=mod (4+|x _i* 4|-|y _i* 4|, 4);

Said coding side motion vector (X _i, Y _i) characteristic vector (V1 _i, V2 _i) abscissa V1 _i=mod (| x _i* 4,2), ordinate V2 _iBe P _i/ 2 integer part.

3. the described video watermark based on motion vector of claim 1 embeds and method for distilling, it is characterized in that, the modification rule among the described step c does,

Work as V1 _i=W _k, V2 _i=W _K+1And mod (P _i, 2)=0 o'clock, V1 perhaps worked as _i=W _k, V2 _i=W _K+1And mod (P _i, 2)=1 o'clock, abscissa X _iWith ordinate Y _iRemain unchanged;

Work as V1 _i=W _k, V2 _i≠ W _K+1And mod (P _i, 2)=0 o'clock, ordinate Y revised _iFor

Work as V1 _i=W _k, V2 _i≠ W _K+1And mod (P _i, 2)=1 o'clock, ordinate Y revised _iFor $(Y_i-\mathrm{Sgn}\left(Y_i\right)*\frac{1}{4})*(1-\mathrm{\&delta;}\left(Y_i\right))+\frac{3}{4}*\mathrm{\&delta;}\left(Y_i\right);$

Work as V1 _i≠ W _k, V2 _i=W _K+1And mod (P _i, 2)=0 o'clock, V1 perhaps worked as _i≠ W _k, V2 _i≠ W _K+1And mod (P _i, 2)=1 o'clock, the modification abscissa does

Work as V1 _i≠ W _k, V2 _i=W _K+1And mod (P _i, 2)=1 o'clock, V1 perhaps worked as _i≠ W _k, V2 _i≠ W _K+1And mod (P _i, 2)=0 o'clock, the modification abscissa does $(X_i-\mathrm{Sgn}\left(X_i\right)*\frac{1}{4})*(1-\mathrm{\&delta;}\left(X_i\right))+\frac{3}{4}*\mathrm{\&delta;}\left(X_i\right);$

Wherein, $\mathrm{Sgn}\left(X_i\right)=\left\{\begin{array}{cc}1& X_i\&GreaterEqual;0\\ 0& X_i<0\end{array}\right.,$ $\mathrm{Sgn}\left(Y_i\right)=\left\{\begin{array}{cc}1& Y_i\&GreaterEqual;0\\ 0& Y_i<0\end{array}\right.,$ $\mathrm{\&delta;}\left(X_i\right)=\left\{\begin{array}{cc}1& X_i=0\\ 0& X_i\&NotEqual;0\end{array}\right.,$ $\mathrm{\&delta;}\left(Y_i\right)=\left\{\begin{array}{cc}1& Y_i=0\\ 0& Y_i\&NotEqual;0\end{array}\right..$

4. the video watermark based on motion vector as claimed in claim 1 embeds and method for distilling, it is characterized in that described step e is to obtain said decoding end motion vector (X _i, Y _i) abscissa X _iFractional part x _i, and ordinate Y _iFractional part y _iCalculate | x _i* 4| and | y _i* the distance P between the 4| _i=mod (4+|x _i* 4|-|y _i* 4|, 4); Said decoding end motion vector (X _i, Y _i) characteristic vector (V1 _i, V2 _i) abscissa V1 _i=mod (| x _i* 4|, 2), ordinate V2 _iBe P _i/ 2 integer part.

5. the video watermark based on motion vector as claimed in claim 1 embeds and method for distilling, it is characterized in that described method is applied to the video compression technology that precision is 1/4 pixel.

6. the video watermark based on motion vector as claimed in claim 5 embeds and method for distilling, it is characterized in that, H.264 described video compression technology is based on, AVS or MPEG-4 standard.

7. the video watermark based on motion vector as claimed in claim 1 embeds and method for distilling, it is characterized in that described method is applied to the B frame or the P frame of frame of video.

Images