CN104125430A - Method and device for detecting video moving objects as well as video monitoring system - Google Patents

Abstract

The invention provides a method and a device for detecting video moving objects as well as a video monitoring system in an embodiment, wherein the method comprises the steps of obtaining the motion vector (MV) and the integer cosine transform (ICT) coefficient of moving image blocks in a video to be detected, screening the moving image blocks according to the ICT coefficient to obtain candidate moving blocks, screening the candidate moving blocks according to the characteristic information of the MV of the candidate moving blocks to obtain target moving blocks of the video, and then performing video moving object detection on the target moving blocks. The method and the device for detecting the video moving objects are capable of taking both calculated quantity and detection accuracy into account.

Description

Video moving object detection method, device and video monitoring system

Technical field

The embodiment of the present invention relates to image processing techniques, relates in particular to a kind of video moving object detection method, device and video monitoring system.

Background technology

It is a key technology during intelligent video is processed and analyzed that video frequency motion target detects, and it is follow-up basis of carrying out target following and behavioural analysis, in video monitoring, video frequency searching and the fields such as man-machine interaction based on video, plays an important role.

Along with popularizing of high-definition camera, amount of video in large-scale supervisory control system is huge, in video moving object detection method, need data volume to be processed to increase, the method that prior art video frequency motion target detects is more, mainly can be divided into following two classes: optical flow method and motion tracking method.Wherein, optical flow method need to estimate in conjunction with light stream, and the space clustering features such as color, brightness, edge of combining object video carry out Video segmentation, and in order to reach enough precision, the amount of calculation needing is large, and the speed of iteration convergence is uncertain; The basis of motion tracking method is the characteristic matching of the picture frame of video, and the algorithm by characteristic matching has reduced needs data volume to be processed, but its precision is lower, and undetected and flase drop easily occurs.That is to say, the video moving object detection method of prior art is difficult to reach balance between amount of calculation and precision.

Summary of the invention

The embodiment of the present invention provides a kind of video moving object detection method, device and video monitoring system, to take into account amount of calculation and accuracy of detection.

First aspect, the embodiment of the present invention provides a kind of video moving object detection method, comprising:

Obtain motion vector MV and the integer cosine transformation ICT coefficient of moving image piece in video to be detected;

According to described ICT coefficient, described moving image piece is screened, obtain candidate's moving mass;

According to the characteristic information of the MV of described candidate's moving mass, described candidate's moving mass is screened, obtain the target travel piece of described video, to carry out video frequency motion target detection according to described target travel piece.

In the possible implementation of the first of first aspect, describedly according to described ICT coefficient, described moving image piece is screened, obtain candidate's moving mass, comprising:

Obtain the ICT coefficient matrix of each moving image piece;

To the DC component in the ICT coefficient matrix of each moving image piece and the summation of low-frequency ac component, obtain the average energy corresponding with each moving image piece;

Determine that average energy in each moving image piece is greater than the first moving image piece of preset value, described the first moving image piece is described candidate's moving mass.

The implementation possible according to the first of first aspect or first aspect, in the possible implementation of the second, describedly according to the characteristic information of the MV of described candidate's moving mass, described candidate's moving mass is screened, obtains the target travel piece of described video, comprising:

The MV of each candidate's moving mass is carried out to medium filtering in spatial domain, obtain the MV after medium filtering;

Adopt formula (1) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain mould homogeneity measure Con _m:

${\mathrm{Con}}_m=\left\{\begin{array}{cc}\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}\left|\right|{\mathrm{MV}}_i^{T_N}|-|{\mathrm{MV}}_i^{T_j}\left|\right|}{N}& \mathrm{if}\left(\underset{1\&le;j\&le;N}{\mathrm{\&Sigma;}}\right|{\mathrm{MV}}_i^{T_j}|\&NotEqual;0)\\ \&infin;& \mathrm{else}\end{array}\right.---\left(1\right)$

Adopt formula (2) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain direction homogeneity measure Con _d:

${\mathrm{Con}}_d=\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}|{\mathrm{ang}}_i^{T_N}-{\mathrm{ang}}_i^{T_j}|}{N}---\left(2\right)$

Wherein, T _jrepresent j sequential in time domain, N represents got sequential number, represent respectively B _iat sequential T _jon MV _iamplitude and deflection;

According to the MV mould homogeneity measure Con of each candidate's moving mass _mwith MV direction homogeneity measure Con _dand the mould value of MV, determine whether each candidate's moving mass is target travel piece.

The implementation possible according to the second of first aspect, in the third possible implementation, described according to the MV mould homogeneity measure Con of each candidate's moving mass _mwith MV direction homogeneity measure Con _dand the mould value of MV, determine that whether each candidate's moving mass is target travel piece, comprising:

Judge i candidate's moving mass B _iwhether meet formula (3) or formula (4):

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cap;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ \left|{\mathrm{MV}}_i\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(3\right)$

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cup;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ {\mathrm{\&lambda;}}_{({\mathrm{Con}}_m,{\mathrm{Con}}_d)}\&CenterDot;\left|{\mathrm{MV}}_t\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(4\right)$

Wherein, α is default Con _mthreshold value, β is default Con _dthreshold value, for default weight threshold, described weight threshold for MV mould homogeneity measure Con _mwith MV direction homogeneity measure Con _dfunction;

If i candidate's moving mass B _imeet formula (3) or formula (4), i candidate's moving mass B _ifor target travel piece.

According to the first of first aspect, first aspect to any one in the third possible implementation, in the 4th kind of possible implementation, described in obtain the motion vector MV of moving image piece in video to be detected, comprising:

Obtain the original motion vector RMV of described moving image piece;

The RMV of described moving image piece is carried out to preliminary treatment, obtain motion vector MV;

Wherein, described preliminary treatment comprises at least one in following processing procedure:

RMV is spatially carried out to the normalization of moving image piece size;

The RMV of the moving image piece of I type and the moving image piece of P_SKIP type is set to without motion vector;

Adopt formula (5) to carry out the normalization in sequential to RMV:

$\mathrm{NMV}\left(B_c^i\right)=\frac{\mathrm{RMV}\left(B_c^i\right)}{c-r}---\left(5\right)$

Wherein, for i moving image piece in present frame, the call number that c is present frame, the call number that r is reference frame.

According to any one in four kinds of possible implementations of the first to the of first aspect, first aspect, in the 5th kind of possible implementation, before the described motion vector MV and integer cosine transformation ICT coefficient that obtains moving image piece in video to be detected, also comprise:

The compressed bit stream of video to be detected is carried out to half decoding, obtain described moving image piece.

Second aspect, the embodiment of the present invention provides a kind of video frequency motion target checkout gear, comprising:

Acquisition module, for obtaining motion vector MV and the integer cosine transformation ICT coefficient of video moving image piece to be detected;

Coarse sizing module, for according to described ICT coefficient, described moving image piece being screened, obtains candidate's moving mass;

Fine screening module, for described candidate's moving mass being screened according to the characteristic information of the MV of described candidate's moving mass, obtains the target travel piece of described video, to carry out video frequency motion target detection according to described target travel piece.

In the possible implementation of the first of second aspect, described coarse sizing module, comprising:

ICT coefficient matrix extraction unit, for obtaining the ICT coefficient matrix of each moving image piece;

Average energy computing unit, for DC component and the summation of low-frequency ac component of the ICT coefficient matrix to each moving image piece, obtains the average energy corresponding with each moving image piece;

Candidate's moving mass determining unit, for determining that each moving image piece average energy is greater than the first moving image piece of preset value, described the first moving image piece is described candidate's moving mass.

The third aspect, the embodiment of the present invention provides a kind of video monitoring system, comprising: front end video capture device, processing server, memory device and client device,

Wherein, described front end video capture device is for gathering video and described video compression coding is become to code stream to be transferred to described processing server at the scene;

Described processing server is connected with described front end video capture device, for: according to the video data of described front end video capture device transmission, obtain motion vector MV and the integer cosine transformation ICT coefficient of moving image piece in video to be detected; According to described ICT coefficient, described moving image piece is screened, obtain candidate's moving mass; According to the characteristic information of the MV of described candidate's moving mass, described candidate's moving mass is screened, obtain the target travel piece of described video, to carry out video frequency motion target detection according to described target travel piece; And described video data is left on described memory device;

Described client device is connected with described processing server, for to described processing server request video data, and shows described video data.

In the possible implementation of the first of the third aspect, described processing server specifically for:

Obtain the ICT coefficient matrix of each moving image piece;

To the DC component in the ICT coefficient matrix of each moving image piece and the summation of low-frequency ac component, obtain the average energy corresponding with each moving image piece;

Determine that average energy in each moving image piece is greater than the first moving image piece of preset value, described the first moving image piece is described candidate's moving mass.

The implementation possible according to the first of the third aspect or the third aspect, in the possible implementation of the second, described processing server specifically for:

The MV of each candidate's moving mass is carried out to medium filtering in spatial domain, obtain the MV after medium filtering;

Adopt formula (1) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain mould homogeneity measure Con _m:

${\mathrm{Con}}_m=\left\{\begin{array}{cc}\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}\left|\right|{\mathrm{MV}}_i^{T_N}|-|{\mathrm{MV}}_i^{T_j}\left|\right|}{N}& \mathrm{if}\left(\underset{1\&le;j\&le;N}{\mathrm{\&Sigma;}}\right|{\mathrm{MV}}_i^{T_j}|\&NotEqual;0)\\ \&infin;& \mathrm{else}\end{array}\right.---\left(1\right)$

Adopt formula (2) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain direction homogeneity measure Con _d:

${\mathrm{Con}}_d=\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}|{\mathrm{ang}}_i^{T_N}-{\mathrm{ang}}_i^{T_j}|}{N}---\left(2\right)$

Wherein, T _jrepresent j sequential in time domain, N represents got sequential number, represent respectively B _iat sequential T _jon MV _iamplitude and deflection;

According to the MV mould homogeneity measure Con of each candidate's moving mass _mwith MV direction homogeneity measure Con _dand the mould value of MV, determine whether each candidate's moving mass is target travel piece.

The implementation possible according to the second of the third aspect, in the third possible implementation, described processing server specifically for:

Judge i candidate's moving mass B _iwhether meet formula (3) or formula (4):

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cap;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ \left|{\mathrm{MV}}_i\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(3\right)$

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cup;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ {\mathrm{\&lambda;}}_{({\mathrm{Con}}_m,{\mathrm{Con}}_d)}\&CenterDot;\left|{\mathrm{MV}}_t\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(4\right)$

Wherein, α is default Con _mthreshold value, β is default Con _dthreshold value, for default weight threshold, described weight threshold for MV mould homogeneity measure Con _mwith MV direction homogeneity measure Con _dfunction;

If i candidate's moving mass B _imeet formula (3) or formula (4), i candidate's moving mass B _ifor target travel piece.

Embodiment of the present invention video moving object detection method, device and video monitoring system, by obtaining motion vector MV and the integer cosine transformation ICT coefficient of moving image piece in video to be detected, and according to described ICT coefficient and described MV, moving image piece in video to be detected is carried out to the screening of two levels respectively, by the thicker screening of ground floor granularity, remove the background noise in video to be detected, while making second layer screening, only need in candidate's moving mass, carry out, realize the reduction of amount of calculation; By the thinner screening of second layer granularity, more accurately determine target travel piece again, guaranteed precision; Particularly, ground floor screens as moving image piece is screened according to ICT coefficient, obtains candidate's moving mass; Second layer screening, for described candidate's moving mass is screened according to the characteristic information of the MV of candidate's moving mass, is obtained the target travel piece of described video, realizes video frequency motion target and detects, and taken into account precision and amount of calculation that video frequency motion target detects.

Accompanying drawing explanation

In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, to the accompanying drawing of required use in embodiment or description of the Prior Art be briefly described below, apparently, accompanying drawing in the following describes is some embodiments of the present invention, for those of ordinary skills, do not paying under the prerequisite of creative work, can also obtain according to these accompanying drawings other accompanying drawing.

Fig. 1 is the flow chart of video moving object detection method embodiment mono-of the present invention;

Fig. 2 is the flow chart of video moving object detection method embodiment bis-of the present invention;

Fig. 3 is the schematic diagram of the ICT coefficient matrix of moving image piece;

Fig. 4 is the flow chart of video moving object detection method embodiment tri-of the present invention;

Fig. 5 is the structural representation of video frequency motion target checkout gear embodiment mono-of the present invention;

Fig. 6 is the structural representation of video frequency motion target checkout gear embodiment bis-of the present invention;

Fig. 7 is the structural representation of video frequency motion target checkout gear embodiment tri-of the present invention;

Fig. 8 is the structural representation of video frequency motion target checkout gear embodiment tetra-of the present invention;

Fig. 9 is the structural representation of video monitoring system embodiment mono-of the present invention.

Embodiment

For making object, technical scheme and the advantage of the embodiment of the present invention clearer, below in conjunction with the accompanying drawing in the embodiment of the present invention, technical scheme in the embodiment of the present invention is clearly and completely described, obviously, described embodiment is the present invention's part embodiment, rather than whole embodiment.Embodiment based in the present invention, those of ordinary skills, not making the every other embodiment obtaining under creative work prerequisite, belong to the scope of protection of the invention.

It is video monitoring system that intelligent video is processed with an important application of analytical technology, and video monitoring system comprises front end video capture device, processing server, memory device and client device.Wherein, front end video capture device can be CCTV camera, as web camera, analog video camera, digital video recorder (Digital Video Recorder, be called for short: the equipment such as DVR), be responsible for gathering at the scene video image and this video image compression coding is become to code stream so that Internet Transmission, thereby be convenient to remote monitoring; Processing server can be management server, media server, for receiving in the past, hold the code stream of video capture device and process and analyze, and bit stream data or video data are recorded and be kept on disk array, and play for program request to client device forwards video code flow; Memory device can be disk array, deposits the video data of processing after server process; Disk array is responsible for the storage of video data, can the attached storage of Adoption Network (Network Attached Storage, is called for short: NAS), (Storage Area Network, is called for short: SAN) or server self storage storage area network; Processing server can also comprise management function, comprises the functions such as login, authentication, business scheduling of being responsible for user, and this part function also can be realized by management server independently; Client device is responsible for running client software, is connected to after processing server, can ask video data, and decodes and show, for user, checks on-the-spot video image.Processing server can receive the access of a plurality of client devices, connects, as IP network between each system by network.

Along with popularizing of high-definition camera, processing server needs the treating capacity of video data to be processed to increase, therefore, the embodiment of the present invention provides a kind of new video moving object detection method and video moving object detection method device, and this video frequency motion target checkout gear can be arranged on processing server, form a kind of new video monitoring system, under the prerequisite that guarantees accuracy of detection, reduce amount of calculation, improve detection efficiency.

Fig. 1 is the flow chart of video moving object detection method embodiment mono-of the present invention, and as shown in Figure 1, the method for the present embodiment can comprise:

Step 101, the motion vector MV that obtains moving image piece in video to be detected and integer cosine transformation ICT coefficient.

The granularity of cutting apart of moving image piece can be 4x4,8x8,8x16,16x16 etc., and described ICT coefficient is comprised of DC component and alternating current component.Conventional Video coding mode has a H.264(video high compression technology at present) and MPEG(Moving Picture Experts Group, Chinese is: Motion Picture Experts Group), adopt the H.264 video of form to there is how new characteristic, for example macro block (mb) type and cut size are more diversified, for example can macro block unification be divided into minimum cut size to 4 * 4 macro block, reference frame number is more, the ability of its compression is stronger, but for conventional moving target detecting method, its detection difficulty is larger.

For the video of form H.264, can carry out to video to be detected the code stream analysis of half decoding, from the H.264 code stream through more than half decodings, can extract again the original motion vector of moving image piece (Raw Motion Vector, be called for short: RMV) and integer cosine transformation (Integer Cosine Transform, abbreviation: ICT) coefficient.For RMV, carry out preliminary treatment again, can obtain motion vector (Motion Vector, the abbreviation: MV) that can supply the moving image piece of subsequent treatment.

Step 102, according to described ICT coefficient, described moving image piece is screened, obtain candidate's moving mass.

Prior art is screened moving image piece according to the size of MV estimated value conventionally, and to obtain candidate's moving mass, the MV of macro block is larger conventionally, and this macro block is that the possibility of target travel piece is larger.But, if video to be detected obtains for taking, for example, at night, under environment, take the video obtaining in the insufficient situation of light, its image information quantity not sufficient, causes MV estimated value unreliable, easily causes undetected or flase drop.In addition, because the MV without texture region in video easily occurs extremely, therefore, according to MV estimated value, screening the candidate's moving mass obtaining may be the pseudo-motion piece without in texture region, and this is that another causes undetected factor.

The present embodiment adopts a kind of new thinking to obtain candidate's moving mass.Particularly, due to for the video of form H.264, its ICT coefficient can reflect energy value, and conventionally, the energy value of certain macro block is higher, and this macro block is that the possibility of target travel piece is larger.Therefore, the step 102 of the present embodiment can be screened described moving image piece according to ICT coefficient, and the macro block that energy value is greater than to preset value is defined as candidate's moving mass.

The method that step 102 is screened moving image piece according to ICT coefficient, can filter out without texture region by comparatively simple calculating, and conventionally in a video, the data volume shared without texture region is very large, even be greater than the shared data volume of candidate's moving mass, therefore, the method for this step can reduce amount of calculation effectively, does not affect again accuracy of detection.

Step 103, according to the characteristic information of the MV of described candidate's moving mass, described candidate's moving mass is screened, obtain the target travel piece of described video, to carry out video frequency motion target detection according to described target travel piece.

Particularly, the characteristic information of MV can comprise: the size of MV, i.e. and the direction of the mould of MV, and MV, and at the MV of time domain mould homogeneity measure Con _mwith the MV direction homogeneity measure Con in time domain _d, wherein, the MV mould homogeneity measure Con of time domain _mfor describing the consistency of macro block MV size between different frame in time domain, the MV mould homogeneity measure Con of time domain _dfor describing the consistency of macro block MV direction between different frame in time domain.

Moving target is generally rigid objects, can think that the core of moving target has motion consistency, and the consistency of the size and Orientation of MV is all higher; May there is different motion morphologies in the marginal portion of moving target, thereby cause the size of MV of marginal portion of moving target and the size and Orientation of the MV of direction and core may have larger gap, therefore we can adopt the mode of medium filtering to suppress MV noise in spatial domain, both can retain MV details, again can filtering noise.

And in time domain, because the time interval between consecutive frame is extremely short, moving target can be approximately even motion, the size and Orientation of the MV of same macro block has higher consistency, therefore, can adopt a macro block at the MV of time domain mould homogeneity measure Con _mwith the MV direction homogeneity measure Con in time domain _d, with the MV mould value of this macro block, form a reliability constraint condition, according to this reliability constraint condition, candidate's moving mass is screened, the macro block that meets this reliability constraint condition is defined as to target travel piece.

The present embodiment, by obtaining motion vector MV and the integer cosine transformation ICT coefficient of moving image piece in video to be detected, and according to described ICT coefficient and described MV, moving image piece in video to be detected is carried out to the screening of two levels respectively, by the thicker screening of ground floor granularity, remove the background noise in video to be detected, while making second layer screening, only need in candidate's moving mass, carry out, realize the reduction of amount of calculation; By the thinner screening of second layer granularity, more accurately determine target travel piece again, guaranteed precision; Particularly, ground floor screens as moving image piece is screened according to ICT coefficient, obtains candidate's moving mass; Second layer screening, for described candidate's moving mass is screened according to the characteristic information of the MV of candidate's moving mass, is obtained the target travel piece of described video, realizes video frequency motion target and detects, and taken into account precision and amount of calculation that video frequency motion target detects.

Adopt several specific embodiments below, the technical scheme of embodiment of the method shown in Fig. 1 is elaborated.

Fig. 2 is the flow chart of video moving object detection method embodiment bis-of the present invention, and the present embodiment, on embodiment mono-basis, has further been introduced moving image piece is screened to the concrete grammar that obtains candidate's moving mass.As shown in Figure 2, the method for the present embodiment can comprise:

Step 201, the motion vector MV that obtains moving image piece in video to be detected and integer cosine transformation ICT coefficient.

Step 202, obtain the ICT coefficient matrix of each moving image piece.

H.264 in the video of form, the minimum particle size of macro block is 4 * 4, the moving image piece that the matrix that can be therefore 4 * 4 by the Data Segmentation of video to be detected is unit, Fig. 3 is the schematic diagram of the ICT coefficient matrix of moving image piece, as shown in Figure 3, the ICT coefficient of a moving image piece is comprised of with 15 AC coefficients that exchange that are positioned at other positions of matrix a direct current DC coefficient that is positioned at the matrix upper left corner.Generally speaking, the average energy of DC coefficient major embodiment moving image piece, the grain distribution of AC coefficient major embodiment moving image piece, and in matrix, the size distribution of the numerical value of each coefficient is, larger the closer to the coefficient in the upper left corner.

If video to be detected is extended formatting, the data of video to be detected are cut apart in the unit that can be macro block by the minimum particle size of this video to be detected.

Step 203, according to the ICT coefficient matrix of each moving image piece, calculate respectively the average energy of each moving image piece.

ICT coefficient matrix is comprised of DC component and alternating current component, and conventionally, the average energy of moving image piece can obtain by the summation to important in ICT coefficient matrix.Consider in ICT coefficient matrix, DC component and low-frequency ac component are larger on the impact of average energy, therefore, can be directly to the DC component in the ICT coefficient matrix to each moving image piece and the summation of low-frequency ac component, the numerical value that summation is obtained is as the average energy of each moving image piece.

Conventionally, DC component is arranged in the upper left corner of ICT coefficient matrix, and low-frequency ac component is three positions adjacent with described DC component, and DC component and low-frequency ac component are four elements in the upper left corner in ICT coefficient matrix.

Therefore, step 203 specifically can comprise:

The first step: obtain four coefficient values in the ICT coefficient matrix of each moving image piece, described four coefficients comprise DC coefficient and three ac coefficients adjacent with described DC coefficient.Three ac coefficients adjacent with described DC coefficient in this step refer in matrix three ac coefficients in the close upper left corner, i.e. AC ₁, AC ₄, AC ₅.

Second step: four of each moving image piece coefficient values are sued for peace respectively, obtain the average energy corresponding with each moving image piece.

That is, adopt formula (6) to calculate and obtain the average energy ICT that each moving image piece is corresponding:

ICT=DC+AC ₁+AC ₄+AC ₅ (6)

It should be noted that, if video to be detected is extended formatting, the size of moving image piece can not be 4 * 4, be for example 8 * 8, calculate average energy herein and also can adopt more coefficient value, for example, for 8 * 8 moving image piece, also can adopt that in matrix, 9 coefficient value sums in the close upper left corner are as the average energy of a moving image piece, the present invention is not construed as limiting this.

Step 204, determine that average energy in each moving image piece is greater than the first moving image piece of preset value, described the first moving image piece is described candidate's moving mass.

Preset value can be set according to the empirical value of the average energy of background noise, when in moving image piece, average energy is less than preset value, can think that this moving image piece belongs to background noise, this moving image piece can no longer carry out any processing to it in follow-up step.

Step 205, according to the characteristic information of the MV of described candidate's moving mass, described candidate's moving mass is screened, obtain the target travel piece of described video, to carry out video frequency motion target detection according to described target travel piece.

The present embodiment, by obtaining the ICT coefficient matrix of each moving image piece, according to the ICT coefficient matrix of each moving image piece, calculate respectively the average energy of each moving image piece, again by average energy being greater than to the first moving image piece of preset value, be defined as candidate's moving mass, the filtration of realization to background noise, has reduced to obtain the amount of calculation of target travel piece.

Fig. 4 is the flow chart of video moving object detection method embodiment tri-of the present invention, and the present embodiment, on the basis of embodiment mono-and embodiment bis-, has further been introduced candidate's moving mass is screened to the concrete grammar that obtains target travel piece.As shown in Figure 4, the method for the present embodiment can comprise:

Step 401, the motion vector MV that obtains moving image piece in video to be detected and integer cosine transformation ICT coefficient.

Step 402, according to described ICT coefficient, described moving image piece is screened, obtain candidate's moving mass.

Step 403, the MV of each candidate's moving mass is carried out to medium filtering in spatial domain, obtain the MV after medium filtering.

Particularly, establishing i candidate in candidate's moving mass, to transport piece be B _i, to B _imV in spatial domain, carry out median-filtered result MV _ican pass through formula (7) obtains:

${\mathrm{MV}}_i=\mathrm{median}\left({\mathrm{mv}}_{i^{\&prime;}}^s\right)---\left(7\right)$

Wherein, represent that i candidate transports piece B _ithe spatially MV of 4 neighborhoods set, i candidate of this set-inclusion transports piece B _iadjacent four macro blocks and the B of left side, right side, upside, downside _ithe MV of itself, median represents this set to ask medium filtering, namely with the mean value of the MV of five macro blocks in this set, as i candidate, transports piece B _imedian-filtered result MV _i.

Step 404, the MV after adopting formula (1) to the medium filtering of i candidate's moving mass Bi _icalculate, obtain B _iat the MV of time domain mould homogeneity measure Con _m:

${\mathrm{Con}}_m=\left\{\begin{array}{cc}\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}\left|\right|{\mathrm{MV}}_i^{T_N}|-|{\mathrm{MV}}_i^{T_j}\left|\right|}{N}& \mathrm{if}\left(\underset{1\&le;j\&le;N}{\mathrm{\&Sigma;}}\right|{\mathrm{MV}}_i^{T_j}|\&NotEqual;0)\\ \&infin;& \mathrm{else}\end{array}\right.---\left(1\right)$

Adopt formula (2) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain direction homogeneity measure Con _d:

${\mathrm{Con}}_d=\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}|{\mathrm{ang}}_i^{T_N}-{\mathrm{ang}}_i^{T_j}|}{N}---\left(2\right)$

Wherein, T _jrepresent j sequential in time domain, represent that respectively i candidate transports piece B _iat current sequential T _jon amplitude and the deflection of MV, they are that i candidate transports piece B _iat sequential T _jon rear orientation projection, N represents got sequential number.

By formula (1), (2), can be found out MV mould and direction homogeneity measure Con _m, Con _dcan detect i candidate and transport piece at T _jsequential and T _j-1, T _j-2..., and and T _j-Nsequential is compared the consistent degree of the motion state of this macro block.

Step 405, according to the MV mould homogeneity measure Con of each candidate's moving mass _mwith MV direction homogeneity measure Con _dand the mould value of MV, determine whether each candidate's moving mass is target travel piece.

Further particularly, step 405 specifically can comprise:

The first step: judge i candidate's moving mass B _iwhether meet formula (3) or formula (4):

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cap;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ \left|{\mathrm{MV}}_i\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(3\right)$

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cup;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ {\mathrm{\&lambda;}}_{({\mathrm{Con}}_m,{\mathrm{Con}}_d)}\&CenterDot;\left|{\mathrm{MV}}_t\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(4\right)$

Wherein, α is default Con _mthreshold value, β is default Con _dthreshold value, for default weight threshold, described weight threshold for MV mould homogeneity measure Con _mwith MV direction homogeneity measure Con _dfunction.

Formula (3) represents, i candidate's moving mass B _imV mould homogeneity measure Con _mwith MV direction homogeneity measure Con _dall meet default requirement, and the current MV size of this candidate's moving mass meets default threshold value, this candidate's moving mass is current is motion state, by this candidate's moving mass B _ibe defined as target travel piece.

Formula (4) represents, i candidate's moving mass B _imV mould homogeneity measure Con _mwith MV direction homogeneity measure Con _din only have one of them to meet default requirement, and the current MV size of this candidate's moving mass is multiplied by a weight threshold meet afterwards default threshold value, by this candidate's moving mass B _ibe defined as target travel piece.Mould consistency and direction consistency are to weigh whether candidate's moving mass is two dimensions of target travel piece.Can be expressed as the value of parameter a, b can be according to adjusting in practical engineering application.For example, if direction consistency has the effect of better difference moving mass and non-moving mass in engineering, strengthen its weight, strengthen the value of b.

Second step: if i candidate's moving mass B _imeet formula (3) or formula (4), i candidate's moving mass B _ifor target travel piece.

The present embodiment, by the MV of each candidate's moving mass is carried out to medium filtering in spatial domain, obtains the MV after medium filtering, and according to the MV mould homogeneity measure Con of each candidate's moving mass _mwith MV direction homogeneity measure Con _dand the mould value of MV, judge whether the above-mentioned MV characteristic information of each candidate's moving mass meets reliability constraint condition, determine whether each candidate's moving mass is target travel piece, realize the meticulous screening that candidate's moving mass is carried out.

Further, in above-mentioned each embodiment, described in obtain the motion vector MV of moving image piece in video to be detected, can comprise:

Step 1: the original motion vector RMV that obtains described moving image piece.

Further, described moving image piece can be by carrying out half decoding acquisition to the compressed bit stream of video to be detected.

Conventionally, moving image piece in this step can be by acquisition that the compressed bit stream of video to be detected is decoded, in embodiments of the present invention, can carry out half decoding to the compressed bit stream of video to be detected, just can obtain the moving image piece of the ICT coefficient needing in the method that can extract original motion vector RMV and the various embodiments described above, therefore can reduce the whole amount of calculation of testing process.

Step 2: the RMV of described moving image piece is carried out to preliminary treatment, obtain motion vector MV.

Wherein, described preliminary treatment comprises at least one in following processing procedure:

RMV is spatially carried out to the normalization of moving image piece size;

The RMV of the moving image piece of I type and the moving image piece of P_SKIP type is set to without motion vector;

Adopt formula (5) to carry out the normalization in sequential to RMV:

$\mathrm{NMV}\left(B_c^i\right)=\frac{\mathrm{RMV}\left(B_c^i\right)}{c-r}---\left(5\right)$

Wherein, for i moving image piece in present frame, the call number that c is present frame, the call number that r is reference frame.

In order to make the better effects if of video moving object detection method of the present invention, described preprocessing process can comprise above-mentioned all processing procedures, and its concrete grammar is as follows:

First, can spatially carry out the normalization of macroblock size to RMV, object is to obtain in the same size and uniform motion vector field, needn't repeat to consider the size of each piece when subsequent treatment.Because the cut size of RMB minimum is 4 * 4,4 * 4 that therefore the macro block that is greater than 4 * 4 can be covered by it are directly copied RMV.

Secondly, the RMV that can type be the macro block of I macro block or P_SKIP is set to, without motion vector, be 0.Reason be the macro block of P_SKIP type without pixel residual error, without motion vector residual error, occurring without texture region in the background piece of being everlasting, often there is the match point of a plurality of optimums in the pixel in this class region, causes the RMV value that obtains inaccurate.

Again, can carry out the normalization in sequential to RMB, object is that the reference frame unification of all macro blocks of present frame is equivalent to former frame, thus the validity during judgement that can improve the reliability constraint condition forming at the characteristic information by MV.

Fig. 5 is the structural representation of video frequency motion target checkout gear embodiment mono-of the present invention, as shown in Figure 5, the device 500 of the present embodiment can comprise: acquisition module 1, coarse sizing module 2 and fine screening module 3, wherein, acquisition module 1 can be for obtaining motion vector MV and the integer cosine transformation ICT coefficient of moving image piece in video to be detected; Coarse sizing module 2 can, for according to described ICT coefficient, described moving image piece being screened, be obtained candidate's moving mass; Fine screening module 3 can be obtained the target travel piece of described video for described candidate's moving mass being screened according to the characteristic information of the MV of described candidate's moving mass, to carry out video frequency motion target detection according to described target travel piece.

The device of the present embodiment, can possess corresponding functional module for the technical scheme of embodiment of the method shown in execution graph 1, and it is similar that it realizes principle, repeats no more herein.

The technique effect of the present embodiment is, by obtaining motion vector MV and the integer cosine transformation ICT coefficient of moving image piece in video to be detected, and according to described ICT coefficient and described MV, moving image piece in video to be detected is carried out to the screening of two levels respectively, by the thicker screening of ground floor granularity, remove the background noise in video to be detected, while making second layer screening, only need in candidate's moving mass, carry out, realize the reduction of amount of calculation; By the thinner screening of second layer granularity, more accurately determine target travel piece again, guaranteed precision; Particularly, ground floor screens as moving image piece is screened according to ICT coefficient, obtains candidate's moving mass; Second layer screening, for described candidate's moving mass is screened according to the characteristic information of the MV of candidate's moving mass, is obtained the target travel piece of described video, realizes video frequency motion target and detects, and taken into account precision and amount of calculation that video frequency motion target detects.

Fig. 6 is the structural representation of video frequency motion target checkout gear embodiment bis-of the present invention, as shown in Figure 6, the device 600 of the present embodiment is on the basis of Fig. 5 shown device structure, further, described coarse sizing module can also comprise: ICT coefficient matrix extraction unit 21, average energy computing unit 22 and candidate's moving mass determining unit 23, wherein, ICT coefficient matrix extraction unit 21, can be for obtaining the ICT coefficient matrix of each moving image piece; Average energy computing unit 22, can, for according to the ICT coefficient matrix of each moving image piece, calculate respectively the average energy of each moving image piece; Candidate's moving mass determining unit 23, can be greater than the first moving image piece of preset value for average energy in definite each moving image piece, described the first moving image piece is described candidate's moving mass.

Further, described average energy computing unit 22 specifically can be for: obtain four coefficient values in the ICT coefficient matrix of each moving image piece, described four coefficients comprise DC coefficient and three ac coefficients adjacent with described DC coefficient; Four of each moving image piece coefficient values are sued for peace respectively, obtain the average energy corresponding with each moving image piece.

The device of the present embodiment, can be for the technical scheme of embodiment of the method shown in execution graph 2, and it realizes principle and technique effect is similar, repeats no more herein.

Fig. 7 is the structural representation of video frequency motion target checkout gear embodiment tri-of the present invention, as shown in Figure 7, the device 700 of the present embodiment is on the basis of Fig. 6 shown device structure, further, fine screening module 3 can comprise: median filter unit 31, consistency computing unit 32 and target travel piece determining unit 33, wherein

Median filter unit 31, can, for the MV of each candidate's moving mass is carried out to medium filtering in spatial domain, obtain the MV after medium filtering;

Consistency computing unit 32, can be for adopting formula (1) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain mould homogeneity measure Con _m:

${\mathrm{Con}}_m=\left\{\begin{array}{cc}\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}\left|\right|{\mathrm{MV}}_i^{T_N}|-|{\mathrm{MV}}_i^{T_j}\left|\right|}{N}& \mathrm{if}\left(\underset{1\&le;j\&le;N}{\mathrm{\&Sigma;}}\right|{\mathrm{MV}}_i^{T_j}|\&NotEqual;0)\\ \&infin;& \mathrm{else}\end{array}\right.---\left(1\right)$

Adopt formula (2) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain direction homogeneity measure Con _d:

${\mathrm{Con}}_d=\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}|{\mathrm{ang}}_i^{T_N}-{\mathrm{ang}}_i^{T_j}|}{N}---\left(2\right)$

Wherein, T _jrepresent j sequential in time domain, N represents got sequential number, represent respectively B _iat sequential T _jon MV _iamplitude and deflection;

Target travel piece determining unit 33, can be for according to the MV mould homogeneity measure Con of each candidate's moving mass _mwith MV direction homogeneity measure Con _dand the mould value of MV, determine whether each candidate's moving mass is target travel piece.

Further, described target travel piece determining unit 33, specifically for:

Judge i candidate's moving mass B _iwhether meet formula (3) or formula (4):

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cap;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ \left|{\mathrm{MV}}_i\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(3\right)$

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cup;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ {\mathrm{\&lambda;}}_{({\mathrm{Con}}_m,{\mathrm{Con}}_d)}\&CenterDot;\left|{\mathrm{MV}}_t\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(4\right)$

Wherein, α is default Con _mthreshold value, β is default Con _dthreshold value, for default weight threshold, described weight threshold for MV mould homogeneity measure Con _mwith MV direction homogeneity measure Con _dfunction;

If i candidate's moving mass B _imeet formula (3) or formula (4), i candidate's moving mass B _ifor target travel piece.

The device of the present embodiment, can be for the technical scheme of embodiment of the method shown in execution graph 4, and it realizes principle and technique effect is similar, repeats no more herein.

Fig. 8 is the structural representation of video frequency motion target checkout gear embodiment tetra-of the present invention, as shown in Figure 8, the device 800 of the present embodiment is on the basis of above-mentioned any device structure, further, acquisition module 1 can comprise: RMV acquiring unit 11 and pretreatment unit 12, wherein, RMV acquiring unit 11, can be for obtaining the original motion vector RMV of described moving image piece; Pretreatment unit 12, can, for the RMV of described moving image piece is carried out to preliminary treatment, obtain motion vector MV; Wherein, described preliminary treatment comprises at least one in following processing procedure:

RMV is spatially carried out to the normalization of moving image piece size;

The RMV of the moving image piece of I type and the moving image piece of P_SKIP type is set to without motion vector;

Adopt formula (5) to carry out the normalization in sequential to RMV:

$\mathrm{NMV}\left(B_c^i\right)=\frac{\mathrm{RMV}\left(B_c^i\right)}{c-r}---\left(5\right)$

Wherein, for i moving image piece in present frame, the call number that c is present frame, the call number that r is reference frame.

Further, the device of the present embodiment can also comprise: half decoder module 4, the compressed bit stream that this half decoder module 4 can detect for treating video carries out half decoding, obtains described moving image piece.

The device of the present embodiment, can be for carrying out the technical scheme of any means embodiment of the present invention, and it realizes principle and technique effect is similar, repeats no more herein.

Fig. 9 is the structural representation of video monitoring system embodiment mono-of the present invention, as shown in Figure 9, the video monitoring system 900 of the present embodiment can comprise: front end video capture device 901, processing server 902, memory device 903 and client device 904, wherein, described processing server 902 can comprise the video frequency motion target checkout gear described in any embodiment of the present invention, described processing server 902 is connected with described front end video capture device 901, for the video data that described front end video capture device 901 is gathered, process, and described video data is left on described memory device 903,

Wherein, described front end video capture device 901 is for gathering at the scene video and described video compression coding being become to code stream to be transferred to described processing server 902;

Described processing server 902 can be for: the video data sending according to described front end video capture device 901, obtains motion vector MV and the integer cosine transformation ICT coefficient of moving image piece in video to be detected; According to described ICT coefficient, described moving image piece is screened, obtain candidate's moving mass; According to the characteristic information of the MV of described candidate's moving mass, described candidate's moving mass is screened, obtain the target travel piece of described video, to carry out video frequency motion target detection according to described target travel piece; And described video data is left on described memory device 903;

Described client device 904 is connected with described processing server 902, for asking video datas to described processing server 902, and shows described video data.

Further, described processing server 902 specifically for:

Obtain the ICT coefficient matrix of each moving image piece;

To the DC component in the ICT coefficient matrix of each moving image piece and the summation of low-frequency ac component, obtain the average energy corresponding with each moving image piece;

Determine that average energy in each moving image piece is greater than the first moving image piece of preset value, described the first moving image piece is described candidate's moving mass.

Further, described processing server 902 specifically for:

The MV of each candidate's moving mass is carried out to medium filtering in spatial domain, obtain the MV after medium filtering;

Adopt formula (1) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain mould homogeneity measure Con _m:

${\mathrm{Con}}_m=\left\{\begin{array}{cc}\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}\left|\right|{\mathrm{MV}}_i^{T_N}|-|{\mathrm{MV}}_i^{T_j}\left|\right|}{N}& \mathrm{if}\left(\underset{1\&le;j\&le;N}{\mathrm{\&Sigma;}}\right|{\mathrm{MV}}_i^{T_j}|\&NotEqual;0)\\ \&infin;& \mathrm{else}\end{array}\right.---\left(1\right)$

Adopt formula (2) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain direction homogeneity measure Con _d:

${\mathrm{Con}}_d=\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}|{\mathrm{ang}}_i^{T_N}-{\mathrm{ang}}_i^{T_j}|}{N}---\left(2\right)$

Wherein, T _jrepresent j sequential in time domain, N represents got sequential number, represent respectively B _iat sequential T _jon MV _iamplitude and deflection;

According to the MV mould homogeneity measure Con of each candidate's moving mass _mwith MV direction homogeneity measure Con _dand the mould value of MV, determine whether each candidate's moving mass is target travel piece.

Further, described processing server 902 specifically for:

Judge i candidate's moving mass B _iwhether meet formula (3) or formula (4):

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cap;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ \left|{\mathrm{MV}}_i\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(3\right)$

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cup;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ {\mathrm{\&lambda;}}_{({\mathrm{Con}}_m,{\mathrm{Con}}_d)}\&CenterDot;\left|{\mathrm{MV}}_t\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(4\right)$

Wherein, α is default Con _mthreshold value, β is default Con _dthreshold value, for default weight threshold, described weight threshold for MV mould homogeneity measure Con _mwith MV direction homogeneity measure Con _dfunction;

If i candidate's moving mass B _imeet formula (3) or formula (4), i candidate's moving mass B _ifor target travel piece.

The video monitoring system of the present embodiment, can processing server be set to can be used in the method for carrying out as described in video moving object detection method embodiment as any in the present invention, therefore, there is the technique effect described in above-mentioned any means embodiment, its know-why is also similar with said method embodiment, repeats no more herein.

One of ordinary skill in the art will appreciate that: all or part of step that realizes above-mentioned each embodiment of the method can complete by the relevant hardware of program command.Aforesaid program can be stored in a computer read/write memory medium.This program, when carrying out, is carried out the step that comprises above-mentioned each embodiment of the method; And aforesaid storage medium comprises: various media that can be program code stored such as ROM, RAM, magnetic disc or CDs.

Finally it should be noted that: each embodiment, only in order to technical scheme of the present invention to be described, is not intended to limit above; Although the present invention is had been described in detail with reference to aforementioned each embodiment, those of ordinary skill in the art is to be understood that: its technical scheme that still can record aforementioned each embodiment is modified, or some or all of technical characterictic is wherein equal to replacement; And these modifications or replacement do not make the essence of appropriate technical solution depart from the scope of various embodiments of the present invention technical scheme.

Claims (12)

1. a video moving object detection method, is characterized in that, comprising:

Obtain motion vector MV and the integer cosine transformation ICT coefficient of moving image piece in video to be detected;

According to described ICT coefficient, described moving image piece is screened, obtain candidate's moving mass;

According to the characteristic information of the MV of described candidate's moving mass, described candidate's moving mass is screened, obtain the target travel piece of described video, to carry out video frequency motion target detection according to described target travel piece.

2. method according to claim 1, is characterized in that, describedly according to described ICT coefficient, described moving image piece is screened, and obtains candidate's moving mass, comprising:

Obtain the ICT coefficient matrix of each moving image piece;

To the DC component in the ICT coefficient matrix of each moving image piece and the summation of low-frequency ac component, obtain the average energy corresponding with each moving image piece;

Determine that average energy in each moving image piece is greater than the first moving image piece of preset value, described the first moving image piece is described candidate's moving mass.

3. method according to claim 1 and 2, is characterized in that, describedly according to the characteristic information of the MV of described candidate's moving mass, described candidate's moving mass is screened, and obtains the target travel piece of described video, comprising:

The MV of each candidate's moving mass is carried out to medium filtering in spatial domain, obtain the MV after medium filtering;

Adopt formula (1) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain mould homogeneity measure Con _m:

${\mathrm{Con}}_m=\left\{\begin{array}{cc}\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}\left|\right|{\mathrm{MV}}_i^{T_N}|-|{\mathrm{MV}}_i^{T_j}\left|\right|}{N}& \mathrm{if}\left(\underset{1\&le;j\&le;N}{\mathrm{\&Sigma;}}\right|{\mathrm{MV}}_i^{T_j}|\&NotEqual;0)\\ \&infin;& \mathrm{else}\end{array}\right.---\left(1\right)$

Adopt formula (2) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain direction homogeneity measure Con _d:

${\mathrm{Con}}_d=\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}|{\mathrm{ang}}_i^{T_N}-{\mathrm{ang}}_i^{T_j}|}{N}---\left(2\right)$

Wherein, T _jrepresent j sequential in time domain, N represents got sequential number, represent respectively B _iat sequential T _jon MV _iamplitude and deflection;

According to the MV mould homogeneity measure Con of each candidate's moving mass _mwith MV direction homogeneity measure Con _dand the mould value of MV, determine whether each candidate's moving mass is target travel piece.

4. method according to claim 3, is characterized in that, described according to the MV mould homogeneity measure Con of each candidate's moving mass _mwith MV direction homogeneity measure Con _dand the mould value of MV, determine that whether each candidate's moving mass is target travel piece, comprising:

Judge i candidate's moving mass B _iwhether meet formula (3) or formula (4):

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cap;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ \left|{\mathrm{MV}}_i\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(3\right)$

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cup;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ {\mathrm{\&lambda;}}_{({\mathrm{Con}}_m,{\mathrm{Con}}_d)}\&CenterDot;\left|{\mathrm{MV}}_t\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(4\right)$

Wherein, α is default Con _mthreshold value, β is default Con _dthreshold value, for default weight threshold, described weight threshold for MV mould homogeneity measure Con _mwith MV direction homogeneity measure Con _dfunction;

If i candidate's moving mass B _imeet formula (3) or formula (4), i candidate's moving mass B _ifor target travel piece.

5. according to the method described in claim 1～4 any one, it is characterized in that, described in obtain the motion vector MV of moving image piece in video to be detected, comprising:

Obtain the original motion vector RMV of described moving image piece;

The RMV of described moving image piece is carried out to preliminary treatment, obtain motion vector MV;

Wherein, described preliminary treatment comprises at least one in following processing procedure:

RMV is spatially carried out to the normalization of moving image piece size;

The RMV of the moving image piece of I type and the moving image piece of P_SKIP type is set to without motion vector;

Adopt formula (5) to carry out the normalization in sequential to RMV:

$\mathrm{NMV}\left(B_c^i\right)=\frac{\mathrm{RMV}\left(B_c^i\right)}{c-r}---\left(5\right)$

Wherein, for i moving image piece in present frame, the call number that c is present frame, the call number that r is reference frame.

6. according to the method described in any one in claim 1～5, it is characterized in that, described in obtain the motion vector MV and integer cosine transformation ICT coefficient of moving image piece in video to be detected before, also comprise:

The compressed bit stream of video to be detected is carried out to half decoding, obtain described moving image piece.

7. a video frequency motion target checkout gear, is characterized in that, comprising:

Acquisition module, for obtaining motion vector MV and the integer cosine transformation ICT coefficient of video moving image piece to be detected;

Coarse sizing module, for according to described ICT coefficient, described moving image piece being screened, obtains candidate's moving mass;

Fine screening module, for described candidate's moving mass being screened according to the characteristic information of the MV of described candidate's moving mass, obtains the target travel piece of described video, to carry out video frequency motion target detection according to described target travel piece.

8. device according to claim 7, is characterized in that, described coarse sizing module, comprising:

ICT coefficient matrix extraction unit, for obtaining the ICT coefficient matrix of each moving image piece;

Average energy computing unit, for DC component and the summation of low-frequency ac component of the ICT coefficient matrix to each moving image piece, obtains the average energy corresponding with each moving image piece;

Candidate's moving mass determining unit, for determining that each moving image piece average energy is greater than the first moving image piece of preset value, described the first moving image piece is described candidate's moving mass.

9. a video monitoring system, is characterized in that, comprising: front end video capture device, processing server, memory device and client device,

Wherein, described front end video capture device is for gathering video and described video compression coding is become to code stream to be transferred to described processing server at the scene;

Described processing server is connected with described front end video capture device, for: according to the video data of described front end video capture device transmission, obtain motion vector MV and the integer cosine transformation ICT coefficient of moving image piece in video to be detected; According to described ICT coefficient, described moving image piece is screened, obtain candidate's moving mass; According to the characteristic information of the MV of described candidate's moving mass, described candidate's moving mass is screened, obtain the target travel piece of described video, to carry out video frequency motion target detection according to described target travel piece; And described video data is left on described memory device;

Described client device is connected with described processing server, for to described processing server request video data, and shows described video data.

10. video monitoring system according to claim 9, is characterized in that, described processing server specifically for:

Obtain the ICT coefficient matrix of each moving image piece;

To the DC component in the ICT coefficient matrix of each moving image piece and the summation of low-frequency ac component, obtain the average energy corresponding with each moving image piece;

Determine that average energy in each moving image piece is greater than the first moving image piece of preset value, described the first moving image piece is described candidate's moving mass.

11. according to the video monitoring system described in claim 9 or 10, it is characterized in that, described processing server specifically for:

The MV of each candidate's moving mass is carried out to medium filtering in spatial domain, obtain the MV after medium filtering;

Adopt formula (1) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain mould homogeneity measure Con _m:

${\mathrm{Con}}_m=\left\{\begin{array}{cc}\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}\left|\right|{\mathrm{MV}}_i^{T_N}|-|{\mathrm{MV}}_i^{T_j}\left|\right|}{N}& \mathrm{if}\left(\underset{1\&le;j\&le;N}{\mathrm{\&Sigma;}}\right|{\mathrm{MV}}_i^{T_j}|\&NotEqual;0)\\ \&infin;& \mathrm{else}\end{array}\right.---\left(1\right)$

Adopt formula (2) to i candidate's moving mass B _imedium filtering after MV _icalculate, obtain B _iat the MV of time domain direction homogeneity measure Con _d:

${\mathrm{Con}}_d=\frac{\underset{1\&le;j<N}{\mathrm{\&Sigma;}}|{\mathrm{ang}}_i^{T_N}-{\mathrm{ang}}_i^{T_j}|}{N}---\left(2\right)$

Wherein, T _jrepresent j sequential in time domain, N represents got sequential number, represent respectively B _iat sequential T _jon MV _iamplitude and deflection;

According to the MV mould homogeneity measure Con of each candidate's moving mass _mwith MV direction homogeneity measure Con _dand the mould value of MV, determine whether each candidate's moving mass is target travel piece.

12. video monitoring systems according to claim 11, is characterized in that, described processing server specifically for:

Judge i candidate's moving mass B _iwhether meet formula (3) or formula (4):

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cap;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ \left|{\mathrm{MV}}_i\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(3\right)$

$\left\{\begin{array}{c}{\mathrm{Con}}_m>\mathrm{\&alpha;}\&cup;{\mathrm{Con}}_d>\mathrm{\&beta;}\\ {\mathrm{\&lambda;}}_{({\mathrm{Con}}_m,{\mathrm{Con}}_d)}\&CenterDot;\left|{\mathrm{MV}}_t\right|>T_{\mathrm{mv}}\end{array}\right\}---\left(4\right)$

Wherein, α is default Con _mthreshold value, β is default Con _dthreshold value, for default weight threshold, described weight threshold for MV mould homogeneity measure Con _mwith MV direction homogeneity measure Con _dfunction;

If i candidate's moving mass B _imeet formula (3) or formula (4), i candidate's moving mass B _ifor target travel piece.