CN104125430B - Video moving object detection method, device and video monitoring system - Google Patents
Video moving object detection method, device and video monitoring system Download PDFInfo
- Publication number
- CN104125430B CN104125430B CN201310157346.8A CN201310157346A CN104125430B CN 104125430 B CN104125430 B CN 104125430B CN 201310157346 A CN201310157346 A CN 201310157346A CN 104125430 B CN104125430 B CN 104125430B
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- mtd
- con
- video
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the present invention provides a kind of video moving object detection method, device and video monitoring system, wherein, method includes：Obtain the motion vector MV and integer cosine transformation ICT coefficients of moving image block in video to be detected；The moving image block is screened according to the ICT coefficients, Candidate Motion block is obtained；The Candidate Motion block is screened according to the MV of Candidate Motion block characteristic information, the target moving mass of the video is obtained, to carry out video frequency motion target detection according to the target moving mass.Video moving object detection method of the embodiment of the present invention and device, can take into account amount of calculation and accuracy of detection.
Description
Technical field
The present embodiments relate to image processing techniques, more particularly to a kind of video moving object detection method, device with
And video monitoring system.
Background technology
Video frequency motion target detection is intelligent video processing and a key technology in analysis, and it is subsequently to carry out target
Tracking and the basis of behavioural analysis, have played in the field such as video monitoring, video frequency searching and man-machine interaction based on video and have focused on
The effect wanted.
With the popularization of high-definition camera, the amount of video in large-scale monitoring system is huge, video moving object detection method
Middle need data volume to be processed increases, and the method for prior art video frequency motion target detection is more, can be largely classified into following two
Class：Optical flow method and motion tracking method.Wherein, optical flow method needs to combine light stream estimation, and combine the color of object video, brightness,
The space clustering feature such as edge carries out Video segmentation, in order to reach enough precision, it is necessary to computationally intensive, and iteration convergence
Speed is not known；The basis of motion tracking method is the characteristic matching of the picture frame of video, is reduced by the algorithm of characteristic matching
Data volume to be processed is needed, but its precision is relatively low, easily occurs missing inspection and flase drop.That is, the video motion of prior art
Object detection method is difficult to reach balance between amount of calculation and precision.
The content 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.
In a first aspect, the embodiment of the present invention provides a kind of video moving object detection method, including：
Obtain the motion vector MV and integer cosine transformation ICT coefficients of moving image block in video to be detected；
The moving image block is screened according to the ICT coefficients, Candidate Motion block is obtained；
The Candidate Motion block is screened according to the MV of Candidate Motion block characteristic information, the video is obtained
Target moving mass, with according to the target moving mass carry out video frequency motion target detection.
In the first possible implementation of first aspect, it is described according to the ICT coefficients to the moving image
Block is screened, and obtains Candidate Motion block, including：
Obtain the ICT coefficient matrixes of each moving image block；
DC component in the ICT coefficient matrixes of each moving image block and low frequency AC components are summed, obtained and each
The corresponding average energy of individual moving image block；
Determine that average energy is more than the first moving image block of preset value, first motion diagram in each moving image block
As block is the Candidate Motion block.
According to the first possible implementation of first aspect or first aspect, in second of possible implementation
In, the characteristic information of the MV according to the Candidate Motion block is screened to the Candidate Motion block, obtains the video
Target moving mass, including：
Medium filtering is carried out in spatial domain to the MV of each Candidate Motion block, the MV after medium filtering is obtained；
Using formula（1）To i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn time domain
MV mould homogeneity measures Conm：
Using formula（2）To i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn time domain
MV orientation consistencies estimate Cond：
Wherein, TjJ-th of sequential in time domain is represented, N represents taken sequential number, B is represented respectivelyi
In sequential TjOn MViAmplitude and deflection；
According to the MV mould homogeneity measures Con of each Candidate Motion blockmEstimate Con with MV orientation consistenciesdAnd MV mould
Whether value, it is target moving mass to determine each Candidate Motion block.
According to second of possible implementation of first aspect, in the third possible implementation, the basis
The MV mould homogeneity measures Con of each Candidate Motion blockmEstimate Con with MV orientation consistenciesdAnd MV modulus value, determine each
Whether Candidate Motion block is target moving mass, including：
Judge i-th of Candidate Motion block BiWhether formula is met（3）Or formula（4）：
Wherein, α is default ConmThreshold value, β be default CondThreshold value,For default weight threshold,
The weight thresholdFor MV mould homogeneity measures ConmEstimate Con with MV orientation consistenciesdFunction；
If i-th of Candidate Motion block BiMeet formula（3）Or formula（4）, then i-th of Candidate Motion block BiMoved for target
Block.
According to the first any one into the third possible implementation of first aspect, first aspect,
In four kinds of possible implementations, the motion vector MV for obtaining moving image block in video to be detected, including：
Obtain the original motion vector RMV of the moving image block；
The RMV of the moving image block is pre-processed, motion vector MV is obtained；
Wherein, the pretreatment includes at least one in following processing procedure：
The normalization of moving image block size is spatially carried out to RMV；
The RMV of the moving image block of I types and the moving image block of P_SKIP types is set to without motion vector；
Using formula（5）Normalization in sequential is carried out to RMV：
Wherein,For i-th of moving image block in present frame, c is the call number of present frame, and r is the index of reference frame
Number.
According to the first any one into the 4th kind of possible implementation of first aspect, first aspect,
In five kinds of possible implementations, the motion vector MV and integer cosine transformation for obtaining moving image block in video to be detected
Before ICT coefficients, in addition to：
Half decoding is carried out to the compressed bit stream of video to be detected, the moving image block is obtained.
Second aspect, the embodiment of the present invention provides a kind of video frequency motion target detection means, including：
Acquisition module, motion vector MV and integer cosine transformation ICT for obtaining moving image block in video to be detected
Coefficient；
Coarse sizing module, for being screened according to the ICT coefficients to the moving image block, obtains Candidate Motion
Block；
Fine screening module, the characteristic information for the MV according to the Candidate Motion block is sieved to the Candidate Motion block
Choosing, obtains the target moving mass of the video, to carry out video frequency motion target detection according to the target moving mass.
In the first possible implementation of second aspect, the coarse sizing module, including：
ICT coefficient matrix extraction units, the ICT coefficient matrixes for obtaining each moving image block；
Average energy computing unit, for the DC component and low frequency in the ICT coefficient matrixes to each moving image block
AC compounent is summed, and obtains average energy corresponding with each moving image block；
Candidate Motion block determining unit, for determining first fortune of the average energy more than preset value in each moving image block
Motion video block, the first moving image block is the Candidate Motion block.
The third aspect, the embodiment of the present invention provides a kind of video monitoring system, including：Head end video collecting device, processing
Server, storage device and client device,
Wherein, the head end video collecting device is used for collection video at the scene and by the video compression coding into code stream
To be transferred to the processing server；
The processing server is connected with the head end video collecting device, is used for：Set according to head end video collection
The video data that preparation is sent, obtains the motion vector MV and integer cosine transformation ICT coefficients of moving image block in video to be detected；
The moving image block is screened according to the ICT coefficients, Candidate Motion block is obtained；According to the MV of the Candidate Motion block
Characteristic information the Candidate Motion block is screened, the target moving mass of the video is obtained, to be transported according to the target
Motion block carries out video frequency motion target detection；And the video data is stored in the storage device；
The client device is connected with the processing server, for asking video data to the processing server,
And show the video data.
In the first possible implementation of the third aspect, the processing server specifically for：
Obtain the ICT coefficient matrixes of each moving image block；
DC component in the ICT coefficient matrixes of each moving image block and low frequency AC components are summed, obtained and each
The corresponding average energy of individual moving image block；
Determine that average energy is more than the first moving image block of preset value, first motion diagram in each moving image block
As block is the Candidate Motion block.
According to the first possible implementation of the third aspect or the third aspect, in second of possible implementation
In, the processing server specifically for：
Medium filtering is carried out in spatial domain to the MV of each Candidate Motion block, the MV after medium filtering is obtained；
Using formula（1）To i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn time domain
MV mould homogeneity measures Conm：
Using formula（2）To i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn time domain
MV orientation consistencies estimate Cond：
Wherein, TjJ-th of sequential in time domain is represented, N represents taken sequential number,Represent respectively
BiIn sequential TjOn MViAmplitude and deflection；
According to the MV mould homogeneity measures Con of each Candidate Motion blockmEstimate Con with MV orientation consistenciesdAnd MV mould
Whether value, it is target moving mass to determine each Candidate Motion block.
According to second of possible implementation of the third aspect, in the third possible implementation, the processing
Server specifically for：
Judge i-th of Candidate Motion block BiWhether formula is met（3）Or formula（4）：
Wherein, α is default ConmThreshold value, β be default CondThreshold value,For default weight threshold,
The weight thresholdFor MV mould homogeneity measures ConmEstimate Con with MV orientation consistenciesdFunction；
If i-th of Candidate Motion block BiMeet formula（3）Or formula（4）, then i-th of Candidate Motion block BiMoved for target
Block.
Video moving object detection method of the embodiment of the present invention, device and video monitoring system, it is to be detected by obtaining
The motion vector MV of moving image block and integer cosine transformation ICT coefficients in video, and respectively according to ICT coefficients and described
MV carries out the screening of two levels to moving image block in video to be detected, is removed by the thicker screening of first layer granularity to be checked
The ambient noise surveyed in video, makes the second layer only to be carried out when screening in Candidate Motion block, realizes the reduction of amount of calculation；Lead to again
The thinner screening of second layer granularity is crossed, target moving mass is more accurately determined, it is ensured that precision；Specifically, first layer is screened
To be screened according to ICT coefficients to moving image block, Candidate Motion block is obtained；Second layer screening is according to Candidate Motion block
MV characteristic information is screened to the Candidate Motion block, is obtained the target moving mass of the video, is realized video motion mesh
Mark detection, and taken into account the precision and amount of calculation of video frequency motion target detection.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
There is the accompanying drawing used required in technology description to be briefly described, it should be apparent that, drawings in the following description are this hairs
Some bright embodiments, for those of ordinary skill in the art, without having to pay creative labor, can be with
Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 is the flow chart of video moving object detection method embodiment one of the present invention；
Fig. 2 is the flow chart of video moving object detection method embodiment two of the present invention；
Fig. 3 is the schematic diagram of the ICT coefficient matrixes of moving image block；
Fig. 4 is the flow chart of video moving object detection method embodiment three of the present invention；
Fig. 5 is the structural representation of video frequency motion target detection means embodiment one of the present invention；
Fig. 6 is the structural representation of video frequency motion target detection means embodiment two of the present invention；
Fig. 7 is the structural representation of video frequency motion target detection means embodiment three of the present invention；
Fig. 8 is the structural representation of video frequency motion target detection means example IV of the present invention；
Fig. 9 is the structural representation of video monitoring system embodiment one of the present invention.
Embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
A part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art
The every other embodiment obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.
One important application of intelligent video processing and analytical technology is video monitoring system, before video monitoring system includes
Hold video capture device, processing server, storage device and client device.Wherein, head end video collecting device can be prison
Control video camera, such as web camera, analog video camera, digital video recorder（Digital Video Recorder, referred to as：
DVR）Etc. equipment, be responsible for collection video image at the scene and by the video image compression coding into code stream so as to network transmission, from
And it is easy to remote monitoring；Processing server can be management server, media server, be set for receiving from front end video acquisition
Standby code stream is simultaneously handled and analyzed, and bit stream data or video data are recorded is stored on disk array, and to client
End equipment forwards video code flow to be played for program request；Storage device can be disk array, and storage is handled after server process
Video data；Disk array is responsible for the storage of video data, can use network attached storage（Network Attached
Storage, referred to as：NAS）, storage area network（Storage Area Network, referred to as：SAN）Or server itself is deposited
Storage；Processing server can also include management function, including be responsible for the functions, this part such as login, authentication, the traffic scheduling of user
Function can also be realized by independent management server；Client device is responsible for running client software, is connected to processing clothes
It is engaged in after device, video data can be asked, and decoded and shown, the video image at scene is checked for user.Processing server can
Receive and pass through network connection, such as IP network between the access of multiple client equipment, each system.
With the popularization of high-definition camera, processing server needs the treating capacity of video data to be processed to increase, therefore, this
Inventive embodiments provide a kind of new video moving object detection method and video moving object detection method device, it is possible to
The video frequency motion target detection means is arranged on processing server, a kind of new video monitoring system is formed, is ensureing inspection
Amount of calculation is reduced on the premise of surveying precision, detection efficiency is improved.
Fig. 1 is the flow chart of video moving object detection method embodiment one of the present invention, as shown in figure 1, the present embodiment
Method can include：
Step 101, the motion vector MV and integer cosine transformation ICT coefficients for obtaining moving image block in video to be detected.
The segmentation granularity of moving image block can be 4x4,8x8,8x16,16x16 etc., and the ICT coefficients are by DC component
With AC compounent composition.H.264 conventional Video coding mode has at present（A kind of video high compression techniques）And MPEG（Moving
Picture Experts Group, Chinese is：Motion Picture Experts Group）, had using the video of H.264 form more new
Characteristic, such as macro block (mb) type and cut size are more diversified, for example, macro block can be uniformly divided into 4 with minimum cut size
× 4 macro block, reference frame number is more, and the ability that it compresses is stronger, but for conventional moving target detecting method, its
Detection difficulty is bigger.
For the video of H.264 form, the code stream analysis of half decoding can be carried out to video to be detected, then from through more than half solutions
The original motion vector of moving image block is can extract in the H.264 code stream of code（Raw Motion Vector, referred to as：RMV）With it is whole
Number cosine transform（Integer Cosine Transform, referred to as：ICT）Coefficient.Pre-processed, can be obtained for RMV again
The motion vector of the moving image block of subsequent treatment must be available for（Motion Vector, referred to as：MV）.
Step 102, according to the ICT coefficients moving image block is screened, obtain Candidate Motion block.
Prior art is screened generally according to the size of MV estimates to moving image block, to obtain Candidate Motion block,
The MV of usual macro block is bigger, and the macro block is bigger for the possibility of target moving mass.But, if video to be detected is not filled in light
Shoot and obtain in the case of foot, for example, obtained video is shot under night environment, its amount of image information is not enough, causes MV to estimate
Value is unreliable, is easily caused missing inspection or flase drop.Further, since easily there is exception in the MV in the texture-free region in video, therefore,
The Candidate Motion block for carrying out screening acquisition according to MV estimates may be the pseudo-motion block in texture-free region, and this is that another is led
Cause the factor of missing inspection.
The present embodiment then obtains Candidate Motion block using a kind of new thinking.Specifically, due to for H.264 form
Video, its ICT coefficient can reflect energy value, generally, and the energy value of some macro block is higher, and the macro block can for target moving mass
Energy property is bigger.Therefore, the step 102 of the present embodiment can be screened according to ICT coefficients to the moving image block, by energy
The macro block that value is more than preset value is defined as Candidate Motion block.
The method that step 102 is screened according to ICT coefficients to moving image block, can pass through relatively simple calculating
Texture-free region is filtered, and generally in a video, the data volume shared by texture-free region is very big, even greater than candidate transports
Data volume shared by motion block, therefore, the method for the step can effectively reduce amount of calculation, and not influence accuracy of detection.
Step 103, according to the MV of the Candidate Motion block characteristic information Candidate Motion block is screened, obtained
The target moving mass of the video, to carry out video frequency motion target detection according to the target moving mass.
Specifically, MV characteristic information can include：The mould of MV size, i.e. MV, and MV direction, and in time domain
MV mould homogeneity measures ConmEstimate Con with the MV orientation consistencies in time domaind, wherein, the MV mould homogeneity measures Con of time domainm
The uniformity of MV sizes between the frame different in time domain for describing a macro block, the MV mould homogeneity measures Con of time domaind
The uniformity in the MV directions between the frame different in time domain for describing a macro block.
Moving target is usually rigid objects, it is believed that the core of moving target has Movement consistency, i.e. MV
Size and Orientation uniformity it is higher；The marginal portion of moving target there may be different motion morphologies, so as to cause
The MV of the MV of the marginal portion of moving target size and direction and core size and Orientation may have larger gap, because
We can suppress MV noises in spatial domain by the way of medium filtering for this, can both retain MV details, can filter out and make an uproar again
Sound.
It is same grand and in time domain, because the time interval of adjacent interframe is extremely short, moving target can be approximately Uniform Movement
The MV of block size and Orientation has higher uniformity, it is therefore possible to use the MV moulds uniformity in time domain of a macro block is surveyed
Spend ConmEstimate Con with the MV orientation consistencies in time domaind, the MV modulus value with the macro block, one reliability constraint of formation,
Candidate Motion block is screened according to the reliability constraint, the macro block for meeting the reliability constraint is defined as mesh
Mark moving mass.
The present embodiment, by the motion vector MV and integer cosine transformation ICT that obtain moving image block in video to be detected
Coefficient, and the screening according to the ICT coefficients and the MV to two levels of moving image block progress in video to be detected respectively,
The ambient noise in video to be detected is removed by the thicker screening of first layer granularity, the second layer is only transported when screening in candidate
Carried out in motion block, realize the reduction of amount of calculation；Again by the thinner screening of second layer granularity, target motion is more accurately determined
Block, it is ensured that precision；Specifically, first layer screening obtains Candidate Motion to be screened according to ICT coefficients to moving image block
Block；Second layer screening is that the Candidate Motion block is screened according to the MV of Candidate Motion block characteristic information, obtains described
The target moving mass of video, realizes that video frequency motion target is detected, and taken into account the precision and amount of calculation of video frequency motion target detection.
Several specific embodiments are used below, and the technical scheme to embodiment of the method shown in Fig. 1 is described in detail.
Fig. 2 is the flow chart of video moving object detection method embodiment two of the present invention, and the present embodiment is in the base of embodiment one
On plinth, further describe and the specific method that screening obtains Candidate Motion block is carried out to moving image block.As shown in Fig. 2 this reality
Applying the method for example can include：
Step 201, the motion vector MV and integer cosine transformation ICT coefficients for obtaining moving image block in video to be detected.
Step 202, the ICT coefficient matrixes for obtaining each moving image block.
H.264 the minimum particle size of macro block is 4 × 4 in the video of form, therefore can split the data of video to be detected
For 4 × 4 matrix be unit moving image block, Fig. 3 for moving image block ICT coefficient matrixes schematic diagram, such as Fig. 3 institutes
Show, the ICT coefficients of a moving image block are located at matrix by a direct current DC coefficient for being located at the matrix upper left corner and 15
The exchange AC coefficients composition of other positions.In general, the average energy of DC coefficient major embodiment moving image blocks, AC coefficient masters
The grain distribution of moving image block is embodied, also, the size distribution of the numerical value of the coefficient of each in matrix is, closer to the upper left corner
Coefficient it is bigger.
If video to be detected is extended formatting, it can split by the minimum particle size of the video to be detected for the unit of macro block
The data of video to be detected.
Step 203, the ICT coefficient matrixes according to each moving image block, calculate being averaged for each moving image block respectively
Energy.
ICT coefficient matrixes are made up of DC component and AC compounent, generally, and the average energy of moving image block can pass through
Summation important in ICT coefficient matrixes is obtained.In view of in ICT coefficient matrixes, DC component and low-frequency ac divide
The influence measured to average energy is larger, therefore, it can directly to the direct current in the ICT coefficient matrixes to each moving image block point
Amount is summed with low frequency AC components, average energy of the numerical value that summation is obtained as each moving image block.
Generally, DC component is located at the upper left corner in ICT coefficient matrixes, and low frequency AC components are and the DC component phase
Three adjacent positions, i.e. DC component and four elements that low frequency AC components are the upper left corner in ICT coefficient matrixes.
Therefore, step 203 can specifically include：
The first step：Four coefficient values in the ICT coefficient matrixes of each moving image block are obtained, four coefficients include straight
Flow coefficient and three ac coefficients adjacent with the DC coefficient.Three adjacent with the DC coefficient in the step hand over
Stream coefficient refers to three ac coefficients in matrix near the upper left corner, i.e. AC1、AC4、AC5。
Second step：Four coefficient values of each moving image block are summed respectively, obtained corresponding with each moving image block
Average energy.
That is, using formula（6）Calculate and obtain the corresponding average energy ICT of each moving image block：
ICT=DC+AC1+AC4+AC5 (6)
It should be noted that if video to be detected is extended formatting, the size of moving image block can not be 4 × 4, for example
For 8 × 8, then more coefficient values can also be used by calculating average energy herein, and such as moving image block for 8 × 8 also may be used
So that using 9 coefficient value sums in matrix near the upper left corner, as the average energy of a moving image block, the present invention is right
This is not construed as limiting.
Step 204, determine that average energy in each moving image block is more than the first moving image block of preset value, described the
One moving image block is the Candidate Motion block.
Preset value can be set according to the empirical value of the average energy of ambient noise, when average energy in moving image block
During less than preset value, it is believed that the moving image block belongs to ambient noise, the moving image block in subsequent steps can be with
Any processing no longer is carried out to it.
Step 205, according to the MV of the Candidate Motion block characteristic information Candidate Motion block is screened, obtained
The target moving mass of the video, to carry out video frequency motion target detection according to the target moving mass.
The present embodiment, by obtaining the ICT coefficient matrixes of each moving image block, the ICT according to each moving image block
Coefficient matrix, calculates the average energy of each moving image block respectively, then by the way that average energy to be more than to the first fortune of preset value
Motion video block, is defined as Candidate Motion block, realizes the filtering to ambient noise, reduces the amount of calculation for obtaining target moving mass.
Fig. 4 is the flow chart of video moving object detection method embodiment three of the present invention, and the present embodiment is in the He of embodiment one
On the basis of embodiment two, further describe and the specific method that screening obtains target moving mass is carried out to Candidate Motion block.Such as
Shown in Fig. 4, the method for the present embodiment can include：
Step 401, the motion vector MV and integer cosine transformation ICT coefficients for obtaining moving image block in video to be detected.
Step 402, according to the ICT coefficients moving image block is screened, obtain Candidate Motion block.
Step 403, in spatial domain medium filtering is carried out to the MV of each Candidate Motion block, obtain the MV after medium filtering.
Specifically, if i-th of candidate fortune block in Candidate Motion block is Bi, to BiMV spatial domain carry out medium filtering knot
Fruit MViFormula can be passed through（7）Obtain：
Wherein,Represent i-th of candidate's fortune block BiThe spatially MV set of 4 neighborhoods, the set includes i-th of candidate
Transport block BiLeft side, right side, upside, adjacent four macro block and B of downsideiThe MV of itself, median represent to ask the set
Medium filtering, that is, it is used as i-th of candidate's fortune block B with the average value of the MV of five macro blocks in the setiMedium filtering knot
Fruit MVi。
Step 404, using formula（1）To the MV after i-th of Candidate Motion block Bi medium filteringiCalculated, obtain Bi
In the MV mould homogeneity measures Con of time domainm：
Using formula（2）To i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn time domain
MV orientation consistencies estimate Cond：
Wherein, TjJ-th of sequential in time domain is represented,I-th of candidate's fortune block B is represented respectivelyiWorking as
Preceding sequential TjOn MV amplitude and deflection, they are i-th candidate fortune block BiIn sequential TjOn rear orientation projection, N represents institute
The sequential number taken.
By formula（1）、（2）As can be seen that MV moulds and orientation consistency estimate Conm, CondIt can detect that i-th of candidate's fortune
Block is in TjSequential and Tj-1、Tj-2... and and Tj-NConsistent degree of the sequential compared to the motion state of the macro block.
Step 405, the MV mould homogeneity measures Con according to each Candidate Motion blockmEstimate Con with MV orientation consistenciesdWith
And MV modulus value, whether determine each Candidate Motion block is target moving mass.
Further specifically, step 405 can specifically include：
The first step：Judge i-th of Candidate Motion block BiWhether formula is met（3）Or formula（4）：
Wherein, α is default ConmThreshold value, β be default CondThreshold value,For default weight threshold,
The weight thresholdFor MV mould homogeneity measures ConmEstimate Con with MV orientation consistenciesdFunction.
Formula（3）Represent, i-th of Candidate Motion block BiMV mould homogeneity measures ConmEstimate with MV orientation consistencies
CondDefault requirement is satisfied by, and the current MV sizes of the Candidate Motion block meet default threshold value, i.e. the Candidate Motion block
It is currently motion state, then by Candidate Motion block BiIt is defined as target moving mass.
Formula（4）Represent, i-th of Candidate Motion block BiMV mould homogeneity measures ConmEstimate with MV orientation consistencies
CondMiddle only one of which meets default requirement, and the current MV sizes of the Candidate Motion block are multiplied by a weight thresholdAfter meet default threshold value, then by Candidate Motion block BiIt is defined as target moving mass.Mould uniformity and orientation consistency are
Weigh Candidate Motion block whether be target moving mass two dimensions.It can be expressed asGinseng
Number a, b value can be adjusted according in practical engineering application.If for example, orientation consistency has preferably difference fortune in engineering
The effect of motion block and non-athletic block, then increase its weight, that is, increase b value.
Second step：If i-th of Candidate Motion block BiMeet formula（3）Or formula（4）, then i-th of Candidate Motion block BiFor mesh
Mark moving mass.
The present embodiment, carries out medium filtering in spatial domain by the MV to each Candidate Motion block, obtains after medium filtering
MV, and according to the MV mould homogeneity measures Con of each Candidate Motion blockmEstimate Con with MV orientation consistenciesdAnd MV modulus value,
Judge whether the above-mentioned MV characteristic informations of each Candidate Motion block meet reliability constraint, to determine each Candidate Motion block
Whether it is target moving mass, realizes the fine screening carried out to Candidate Motion block.
Further, in each above-mentioned embodiment, the motion vector MV for obtaining moving image block in video to be detected,
It can include：
Step one：Obtain the original motion vector RMV of the moving image block.
Further, the moving image block can carry out half decoding acquisition by the compressed bit stream to video to be detected.
Generally, the moving image block in the step can carry out decoding acquisition by the compressed bit stream to video to be detected,
In embodiments of the present invention, half can be carried out to the compressed bit stream of video to be detected to decode, it is possible to which acquisition can extract original
The moving image block of the ICT coefficients needed in the method for motion vector RMV and the various embodiments described above, therefore inspection can be reduced
The overall amount of calculation of survey process.
Step 2：The RMV of the moving image block is pre-processed, motion vector MV is obtained.
Wherein, the pretreatment includes at least one in following processing procedure：
The normalization of moving image block size is spatially carried out to RMV；
The RMV of the moving image block of I types and the moving image block of P_SKIP types is set to without motion vector；
Using formula（5）Normalization in sequential is carried out to RMV：
Wherein,For i-th of moving image block in present frame, c is the call number of present frame, and r is the index of reference frame
Number.
In order that the effect of video moving object detection method of the present invention is more preferably, the preprocessing process can include above-mentioned
All processing procedures, its specific method is as follows：
It is possible, firstly, to spatially carry out the normalization of macroblock size to RMV, it is therefore an objective to obtain in the same size and uniform
Motion vector field, need not repeat to consider the size of each block in subsequent treatment.Because cut size minimum RMB is 4 × 4,
Therefore 4 × 4 pieces of directly duplication RMV that 4 × 4 macro block is covered by it can be will be greater than.
Secondly, RMV that can be by type for I macro blocks or P_SKIP macro block is set to without motion vector, as 0.Reason
Be P_SKIP types macro block without pixel residual error, without motion vector residual error, often the texture-free region in background block occur, this
Often there are multiple optimal match points in the pixel in class region, cause the RMV values obtained inaccurate.
Again, the normalization in sequential can be carried out to RMB, it is therefore an objective to which the reference frame of all macro blocks of present frame is unified etc.
Imitate as former frame, validity during judgement so as to improve the reliability constraint constituted in the characteristic information by MV.
Fig. 5 is the structural representation of video frequency motion target detection means embodiment one of the present invention, as shown in figure 5, this implementation
The device 500 of example can include：Acquisition module 1, coarse sizing module 2 and fine screening module 3, wherein, acquisition module 1 can be used for
Obtain the motion vector MV and integer cosine transformation ICT coefficients of moving image block in video to be detected；Coarse sizing module 2 can be used
The moving image block is screened according to the ICT coefficients, Candidate Motion block is obtained；Fine screening module 3 can be used for
The Candidate Motion block is screened according to the MV of Candidate Motion block characteristic information, the target fortune of the video is obtained
Motion block, to carry out video frequency motion target detection according to the target moving mass.
The device of the present embodiment, can be used for the technical scheme for performing embodiment of the method shown in Fig. 1, possesses corresponding function
Module, its realization principle is similar, and here is omitted.
The technique effect of the present embodiment is, by the motion vector MV and integer that obtain moving image block in video to be detected
Cosine transform ICT coefficients, and two are carried out to moving image block in video to be detected according to the ICT coefficients and the MV respectively
The screening of level, the ambient noise in video to be detected is removed by the thicker screening of first layer granularity, when screening the second layer
It need to only be carried out in Candidate Motion block, realize the reduction of amount of calculation；Again by the thinner screening of second layer granularity, more precisely
Determine target moving mass, it is ensured that precision；Specifically, first layer screening is that moving image block is screened according to ICT coefficients,
Obtain Candidate Motion block；Second layer screening is that the Candidate Motion block is sieved according to the MV of Candidate Motion block characteristic information
Choosing, obtains the target moving mass of the video, realizes that video frequency motion target is detected, and taken into account the essence of video frequency motion target detection
Degree and amount of calculation.
Fig. 6 is the structural representation of video frequency motion target detection means embodiment two of the present invention, as shown in fig. 6, this implementation
The device 600 of example is on the basis of Fig. 5 shown device structures, and further, the coarse sizing module can also include：ICT systems
Matrix number extraction unit 21, average energy computing unit 22 and Candidate Motion block determining unit 23, wherein, ICT coefficient matrixes are carried
Unit 21 is taken, can be used for the ICT coefficient matrixes for obtaining each moving image block；Average energy computing unit 22, can be used for
According to the ICT coefficient matrixes of each moving image block, the average energy of each moving image block is calculated respectively；Candidate Motion block is true
Order member 23, is determined for the first moving image block that average energy in each moving image block is more than preset value, described
First moving image block is the Candidate Motion block.
Further, the average energy computing unit 22 specifically can be used for：Obtain the ICT systems of each moving image block
Four coefficient values in matrix number, four coefficients include DC coefficient and the three exchange systems adjacent with the DC coefficient
Number；Four coefficient values of each moving image block are summed respectively, average energy corresponding with each moving image block is obtained.
The device of the present embodiment, can be used for the technical scheme for performing embodiment of the method shown in Fig. 2, its realization principle and skill
Art effect is similar, and here is omitted.
Fig. 7 is the structural representation of video frequency motion target detection means embodiment three of the present invention, as shown in fig. 7, this implementation
The device 700 of example is on the basis of Fig. 6 shown device structures, and further, fine screening module 3 can include：Medium filtering list
Member 31, uniformity computing unit 32 and target moving mass determining unit 33, wherein,
Median filter unit 31, can be used for carrying out medium filtering in spatial domain to the MV of each Candidate Motion block, in obtaining
MV after value filtering；
Uniformity computing unit 32, can be used for using formula（1）To i-th of Candidate Motion block BiMedium filtering after
MViCalculated, obtain BiIn the MV mould homogeneity measures Con of time domainm：
Using formula（2）To i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn time domain
MV orientation consistencies estimate Cond：
Wherein, TjJ-th of sequential in time domain is represented, N represents taken sequential number,Represent respectively
BiIn sequential TjOn MViAmplitude and deflection；
Target moving mass determining unit 33, can be used for the MV mould homogeneity measures Con according to each Candidate Motion blockmWith
MV orientation consistencies estimate CondAnd MV modulus value, whether determine each Candidate Motion block is target moving mass.
Further, the target moving mass determining unit 33, specifically for：
Judge i-th of Candidate Motion block BiWhether formula is met（3）Or formula（4）：
Wherein, α is default ConmThreshold value, β be default CondThreshold value,For default weight threshold,
The weight thresholdFor MV mould homogeneity measures ConmEstimate Con with MV orientation consistenciesdFunction；
If i-th of Candidate Motion block BiMeet formula（3）Or formula（4）, then i-th of Candidate Motion block BiMoved for target
Block.
The device of the present embodiment, can be used for the technical scheme for performing embodiment of the method shown in Fig. 4, its realization principle and skill
Art effect is similar, and here is omitted.
Fig. 8 is the structural representation of video frequency motion target detection means example IV of the present invention, as shown in figure 8, this implementation
The device 800 of example is on the basis of above-mentioned any device structure, and further, acquisition module 1 can include：RMV acquiring units
11 and pretreatment unit 12, wherein, RMV acquiring units 11 can be used for the original motion vector for obtaining the moving image block
RMV；Pretreatment unit 12, can be used for pre-processing the RMV of the moving image block, obtain motion vector MV；Wherein,
The pretreatment includes at least one in following processing procedure：
The normalization of moving image block size is spatially carried out to RMV；
The RMV of the moving image block of I types and the moving image block of P_SKIP types is set to without motion vector；
Using formula（5）Normalization in sequential is carried out to RMV：
Wherein,For i-th of moving image block in present frame, c is the call number of present frame, and r is the index of reference frame
Number.
Further, the device of the present embodiment can also include：Half decoder module 4, half decoder module 4 can be used for pair
The compressed bit stream of video to be detected carries out half and decoded, and obtains the moving image block.
The device of the present embodiment, can be used for the technical scheme for performing any means embodiment of the present invention, its realization principle
Similar with technique effect, here is omitted.
Fig. 9 is the structural representation of video monitoring system embodiment one of the present invention, as shown in figure 9, the video of the present embodiment
Monitoring system 900 can include：Head end video collecting device 901, processing server 902, storage device 903 and client device
904, wherein, the processing server 902 can include the video frequency motion target detection means described in any embodiment of the present invention,
The processing server 902 is connected with the head end video collecting device 901, for the head end video collecting device 901
The video data of collection is handled, and the video data is stored in the storage device 903；
Wherein, the head end video collecting device 901 be used at the scene collection video and by the video compression coding into
Code stream is to be transferred to the processing server 902；
The processing server 902 can be used for：The video data sent according to the head end video collecting device 901,
Obtain the motion vector MV and integer cosine transformation ICT coefficients of moving image block in video to be detected；According to the ICT coefficients pair
The moving image block is screened, and obtains Candidate Motion block；According to the MV of Candidate Motion block characteristic information to described
Candidate Motion block is screened, and obtains the target moving mass of the video, to carry out video motion according to the target moving mass
Target detection；And the video data is stored in the storage device 903；
The client device 904 is connected with the processing server 902, for being asked to the processing server 902
Video data, and show the video data.
Further, the processing server 902 specifically for：
Obtain the ICT coefficient matrixes of each moving image block；
DC component in the ICT coefficient matrixes of each moving image block and low frequency AC components are summed, obtained and each
The corresponding average energy of individual moving image block；
Determine that average energy is more than the first moving image block of preset value, first motion diagram in each moving image block
As block is the Candidate Motion block.
Further, the processing server 902 specifically for：
Medium filtering is carried out in spatial domain to the MV of each Candidate Motion block, the MV after medium filtering is obtained；
Using formula（1）To i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn time domain
MV mould homogeneity measures Conm：
Using formula（2）To i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn time domain
MV orientation consistencies estimate Cond：
Wherein, TjJ-th of sequential in time domain is represented, N represents taken sequential number,Represent respectively
BiIn sequential TjOn MViAmplitude and deflection；
According to the MV mould homogeneity measures Con of each Candidate Motion blockmEstimate Con with MV orientation consistenciesdAnd MV mould
Whether value, it is target moving mass to determine each Candidate Motion block.
Further, the processing server 902 specifically for：
Judge i-th of Candidate Motion block BiWhether formula is met（3）Or formula（4）：
Wherein, α is default ConmThreshold value, β be default CondThreshold value,For default weight threshold,
The weight thresholdFor MV mould homogeneity measures ConmEstimate Con with MV orientation consistenciesdFunction；
If i-th of Candidate Motion block BiMeet formula（3）Or formula（4）, then i-th of Candidate Motion block BiMoved for target
Block.
Processing server, can be arranged to be enough in the execution such as present invention arbitrarily by the video monitoring system of the present embodiment
Method described in video moving object detection method embodiment, therefore, with the technology effect described in above-mentioned any means embodiment
Really, its technical principle is also similar with above method embodiment, and here is omitted.
One of ordinary skill in the art will appreciate that：Realizing all or part of step of above-mentioned each method embodiment can lead to
The related hardware of programmed instruction is crossed to complete.Foregoing program can be stored in a computer read/write memory medium.The journey
Sequence upon execution, performs the step of including above-mentioned each method embodiment；And foregoing storage medium includes：ROM, RAM, magnetic disc or
Person's CD etc. is various can be with the medium of store program codes.
Finally it should be noted that：Various embodiments above is merely illustrative of the technical solution of the present invention, rather than its limitations；To the greatest extent
The present invention is described in detail with reference to foregoing embodiments for pipe, it will be understood by those within the art that：Its according to
The technical scheme described in foregoing embodiments can so be modified, or which part or all technical characteristic are entered
Row equivalent substitution；And these modifications or replacement, the essence of appropriate technical solution is departed from various embodiments of the present invention technology
The scope of scheme.
Claims (8)
1. a kind of video moving object detection method, it is characterised in that including：
Obtain the motion vector MV and integer cosine transformation ICT coefficients of moving image block in video to be detected；
The moving image block is screened according to the ICT coefficients, Candidate Motion block is obtained；
The Candidate Motion block is screened according to the MV of Candidate Motion block characteristic information, the mesh of the video is obtained
Moving mass is marked, to carry out video frequency motion target detection according to the target moving mass；
The characteristic information of the MV according to the Candidate Motion block is screened to the Candidate Motion block, obtains the video
Target moving mass, including：
Medium filtering is carried out in spatial domain to the MV of each Candidate Motion block, the MV after medium filtering is obtained；
Using formula (1) to i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn the MV moulds of time domain
Homogeneity measure Conm：
<mrow>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mfrac>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mn>1</mn>
<mo>&le;</mo>
<mi>j</mi>
<mo><</mo>
<mi>N</mi>
</mrow>
</munder>
<mrow>
<mo>|</mo>
<mrow>
<mo>|</mo>
<msubsup>
<mi>MV</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>N</mi>
</msub>
</msubsup>
<mo>|</mo>
<mo>-</mo>
<mo>|</mo>
<msubsup>
<mi>MV</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
</msubsup>
<mo>|</mo>
</mrow>
<mo>|</mo>
</mrow>
</mrow>
<mi>N</mi>
</mfrac>
</mtd>
<mtd>
<mtable>
<mtr>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>(</mo>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mn>1</mn>
<mo>&le;</mo>
<mi>j</mi>
<mo><</mo>
<mi>N</mi>
</mrow>
</munder>
<mo>|</mo>
<msubsup>
<mi>MV</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
</msubsup>
<mo>|</mo>
<mo>&NotEqual;</mo>
<mn>0</mn>
</mrow>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
</mtable>
</mtd>
</mtr>
<mtr>
<mtd>
<mi>&infin;</mi>
</mtd>
<mtd>
<mrow>
<mi>e</mi>
<mi>l</mi>
<mi>s</mi>
<mi>e</mi>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
Using formula (2) to i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn the MV side of time domain
To homogeneity measure Cond：
<mrow>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>=</mo>
<mfrac>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mn>1</mn>
<mo>&le;</mo>
<mi>j</mi>
<mo><</mo>
<mi>N</mi>
</mrow>
</munder>
<mo>|</mo>
<msubsup>
<mi>ang</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>N</mi>
</msub>
</msubsup>
<mo>-</mo>
<msubsup>
<mi>ang</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
</msubsup>
<mo>|</mo>
</mrow>
<mi>N</mi>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, TjJ-th of sequential in time domain is represented, N represents taken sequential number,B is represented respectivelyiWhen
Sequence TjOn MViAmplitude and deflection；
Judge i-th of Candidate Motion block BiWhether formula (3) or formula (4) are met：
<mrow>
<mfenced open = "{" close = "}">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>></mo>
<mi>&alpha;</mi>
<mo>&cap;</mo>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>></mo>
<mi>&beta;</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>|</mo>
<msub>
<mi>MV</mi>
<mi>i</mi>
</msub>
<mo>|</mo>
<mo>></mo>
<msub>
<mi>T</mi>
<mrow>
<mi>m</mi>
<mi>v</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mfenced open = "{" close = "}">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>></mo>
<mi>&alpha;</mi>
<mo>&cup;</mo>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>></mo>
<mi>&beta;</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>&lambda;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>,</mo>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>)</mo>
</mrow>
</msub>
<mo>&CenterDot;</mo>
<mo>|</mo>
<msub>
<mi>MV</mi>
<mi>t</mi>
</msub>
<mo>|</mo>
<mo>></mo>
<msub>
<mi>T</mi>
<mrow>
<mi>m</mi>
<mi>v</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, α is default ConmThreshold value, β be default CondThreshold value,It is described for default weight threshold
Weight thresholdFor MV mould homogeneity measures ConmEstimate Con with MV orientation consistenciesdFunction；
If i-th of Candidate Motion block BiFormula (3) or formula (4) are met, then i-th of Candidate Motion block BiFor target moving mass.
2. according to the method described in claim 1, it is characterised in that it is described according to the ICT coefficients to the moving image block
Screened, obtain Candidate Motion block, including：
Obtain the ICT coefficient matrixes of each moving image block；
DC component in the ICT coefficient matrixes of each moving image block and low frequency AC components are summed, obtained and each fortune
The corresponding average energy of motion video block；
Determine that average energy is more than the first moving image block of preset value, the first moving image block in each moving image block
For the Candidate Motion block.
3. the method according to any one of claim 1~2, it is characterised in that motion diagram in the acquisition video to be detected
As the motion vector MV of block, including：
Obtain the original motion vector RMV of the moving image block；
The RMV of the moving image block is pre-processed, motion vector MV is obtained；
Wherein, the pretreatment includes at least one in following processing procedure：
The normalization of moving image block size is spatially carried out to RMV；
The RMV of the moving image block of I types and the moving image block of P_SKIP types is set to without motion vector；
The normalization in sequential is carried out to RMV using formula (5)：
<mrow>
<mi>N</mi>
<mi>M</mi>
<mi>V</mi>
<mrow>
<mo>(</mo>
<msubsup>
<mi>B</mi>
<mi>c</mi>
<mi>i</mi>
</msubsup>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<mi>R</mi>
<mi>M</mi>
<mi>V</mi>
<mrow>
<mo>(</mo>
<msubsup>
<mi>B</mi>
<mi>c</mi>
<mi>i</mi>
</msubsup>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>c</mi>
<mo>-</mo>
<mi>r</mi>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>5</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein,For i-th of moving image block in present frame, c is the call number of present frame, and r is the call number of reference frame.
4. according to method according to any one of claims 1 to 2, it is characterised in that moved in the acquisition video to be detected
Before the motion vector MV and integer cosine transformation ICT coefficients of image block, in addition to：
Half decoding is carried out to the compressed bit stream of video to be detected, the moving image block is obtained.
5. a kind of video frequency motion target detection means, it is characterised in that including：
Acquisition module, motion vector MV and integer cosine transformation ICT coefficients for obtaining moving image block in video to be detected；
Coarse sizing module, for being screened according to the ICT coefficients to the moving image block, obtains Candidate Motion block；
Fine screening module, the characteristic information for the MV according to the Candidate Motion block is screened to the Candidate Motion block,
The target moving mass of the video is obtained, to carry out video frequency motion target detection according to the target moving mass；
The fine screening module, including：
Median filter unit, carries out medium filtering in spatial domain for the MV to each Candidate Motion block, obtains after medium filtering
MV；
Uniformity computing unit, for using formula (1) to i-th of Candidate Motion block BiMedium filtering after MViCounted
Calculate, obtain BiIn the MV mould homogeneity measures Con of time domainm：
<mrow>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mfrac>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mn>1</mn>
<mo>&le;</mo>
<mi>j</mi>
<mo><</mo>
<mi>N</mi>
</mrow>
</munder>
<mrow>
<mo>|</mo>
<mrow>
<mo>|</mo>
<msubsup>
<mi>MV</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>N</mi>
</msub>
</msubsup>
<mo>|</mo>
<mo>-</mo>
<mo>|</mo>
<msubsup>
<mi>MV</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
</msubsup>
<mo>|</mo>
</mrow>
<mo>|</mo>
</mrow>
</mrow>
<mi>N</mi>
</mfrac>
</mtd>
<mtd>
<mtable>
<mtr>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>(</mo>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mn>1</mn>
<mo>&le;</mo>
<mi>j</mi>
<mo><</mo>
<mi>N</mi>
</mrow>
</munder>
<mo>|</mo>
<msubsup>
<mi>MV</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
</msubsup>
<mo>|</mo>
<mo>&NotEqual;</mo>
<mn>0</mn>
</mrow>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
</mtable>
</mtd>
</mtr>
<mtr>
<mtd>
<mi>&infin;</mi>
</mtd>
<mtd>
<mrow>
<mi>e</mi>
<mi>l</mi>
<mi>s</mi>
<mi>e</mi>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
Using formula (2) to i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn the MV side of time domain
To homogeneity measure Cond：
<mrow>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>=</mo>
<mfrac>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mn>1</mn>
<mo>&le;</mo>
<mi>j</mi>
<mo><</mo>
<mi>N</mi>
</mrow>
</munder>
<mo>|</mo>
<msubsup>
<mi>ang</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>N</mi>
</msub>
</msubsup>
<mo>-</mo>
<msubsup>
<mi>ang</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
</msubsup>
<mo>|</mo>
</mrow>
<mi>N</mi>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, TjJ-th of sequential in time domain is represented, N represents taken sequential number,B is represented respectivelyiWhen
Sequence TjOn MViAmplitude and deflection；
Target moving mass determining unit, for judging i-th of Candidate Motion block BiWhether formula (3) or formula (4) are met：
<mrow>
<mfenced open = "{" close = "}">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>></mo>
<mi>&alpha;</mi>
<mo>&cap;</mo>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>></mo>
<mi>&beta;</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>|</mo>
<msub>
<mi>MV</mi>
<mi>i</mi>
</msub>
<mo>|</mo>
<mo>></mo>
<msub>
<mi>T</mi>
<mrow>
<mi>m</mi>
<mi>v</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
2
<mrow>
<mfenced open = "{" close = "}">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>></mo>
<mi>&alpha;</mi>
<mo>&cup;</mo>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>></mo>
<mi>&beta;</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>&lambda;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>,</mo>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>)</mo>
</mrow>
</msub>
<mo>&CenterDot;</mo>
<mo>|</mo>
<msub>
<mi>MV</mi>
<mi>t</mi>
</msub>
<mo>|</mo>
<mo>></mo>
<msub>
<mi>T</mi>
<mrow>
<mi>m</mi>
<mi>v</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, α is default ConmThreshold value, β be default CondThreshold value,It is described for default weight threshold
Weight thresholdFor MV mould homogeneity measures ConmEstimate Con with MV orientation consistenciesdFunction；
If i-th of Candidate Motion block BiFormula (3) or formula (4) are met, then i-th of Candidate Motion block BiFor target moving mass.
6. device according to claim 5, it is characterised in that the coarse sizing module, including：
ICT coefficient matrix extraction units, the ICT coefficient matrixes for obtaining each moving image block；
Average energy computing unit, for the DC component and low-frequency ac in the ICT coefficient matrixes to each moving image block
Component is summed, and obtains average energy corresponding with each moving image block；
Candidate Motion block determining unit, for determining first motion diagram of the average energy more than preset value in each moving image block
As block, the first moving image block is the Candidate Motion block.
7. a kind of video monitoring system, it is characterised in that including：Head end video collecting device, processing server, storage device and
Client device,
Wherein, the head end video collecting device be used at the scene collection video and by the video compression coding into code stream so as to
It is transferred to the processing server；
The processing server is connected with the head end video collecting device, is used for：Sent out according to the head end video collecting device
The video data sent, obtains the motion vector MV and integer cosine transformation ICT coefficients of moving image block in video to be detected；According to
The ICT coefficients are screened to the moving image block, obtain Candidate Motion block；According to the MV of Candidate Motion block spy
Reference breath is screened to the Candidate Motion block, the target moving mass of the video is obtained, with according to the target moving mass
Carry out video frequency motion target detection；And the video data is stored in the storage device；
The client device is connected with the processing server, for asking video data to the processing server, and is shown
Show the video data；
The processing server specifically for：
Medium filtering is carried out in spatial domain to the MV of each Candidate Motion block, the MV after medium filtering is obtained；
Using formula (1) to i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn the MV moulds of time domain
Homogeneity measure Conm：
<mrow>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mfrac>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mn>1</mn>
<mo>&le;</mo>
<mi>j</mi>
<mo><</mo>
<mi>N</mi>
</mrow>
</munder>
<mrow>
<mo>|</mo>
<mrow>
<mo>|</mo>
<msubsup>
<mi>MV</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>N</mi>
</msub>
</msubsup>
<mo>|</mo>
<mo>-</mo>
<mo>|</mo>
<msubsup>
<mi>MV</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
</msubsup>
<mo>|</mo>
</mrow>
<mo>|</mo>
</mrow>
</mrow>
<mi>N</mi>
</mfrac>
</mtd>
<mtd>
<mtable>
<mtr>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>(</mo>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mn>1</mn>
<mo>&le;</mo>
<mi>j</mi>
<mo><</mo>
<mi>N</mi>
</mrow>
</munder>
<mo>|</mo>
<msubsup>
<mi>MV</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
</msubsup>
<mo>|</mo>
<mo>&NotEqual;</mo>
<mn>0</mn>
</mrow>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
</mtable>
</mtd>
</mtr>
<mtr>
<mtd>
<mi>&infin;</mi>
</mtd>
<mtd>
<mrow>
<mi>e</mi>
<mi>l</mi>
<mi>s</mi>
<mi>e</mi>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
Using formula (2) to i-th of Candidate Motion block BiMedium filtering after MViCalculated, obtain BiIn the MV side of time domain
To homogeneity measure Cond：
<mrow>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>=</mo>
<mfrac>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mn>1</mn>
<mo>&le;</mo>
<mi>j</mi>
<mo><</mo>
<mi>N</mi>
</mrow>
</munder>
<mo>|</mo>
<msubsup>
<mi>ang</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>N</mi>
</msub>
</msubsup>
<mo>-</mo>
<msubsup>
<mi>ang</mi>
<mi>i</mi>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
</msubsup>
<mo>|</mo>
</mrow>
<mi>N</mi>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, TjJ-th of sequential in time domain is represented, N represents taken sequential number,B is represented respectivelyi
Sequential TjOn MViAmplitude and deflection；
Judge i-th of Candidate Motion block BiWhether formula (3) or formula (4) are met：
<mrow>
<mfenced open = "{" close = "}">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>></mo>
<mi>&alpha;</mi>
<mo>&cap;</mo>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>></mo>
<mi>&beta;</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>|</mo>
<msub>
<mi>MV</mi>
<mi>i</mi>
</msub>
<mo>|</mo>
<mo>></mo>
<msub>
<mi>T</mi>
<mrow>
<mi>m</mi>
<mi>v</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mfenced open = "{" close = "}">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>></mo>
<mi>&alpha;</mi>
<mo>&cup;</mo>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>></mo>
<mi>&beta;</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>&lambda;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>Con</mi>
<mi>m</mi>
</msub>
<mo>,</mo>
<msub>
<mi>Con</mi>
<mi>d</mi>
</msub>
<mo>)</mo>
</mrow>
</msub>
<mo>&CenterDot;</mo>
<mo>|</mo>
<msub>
<mi>MV</mi>
<mi>t</mi>
</msub>
<mo>|</mo>
<mo>></mo>
<msub>
<mi>T</mi>
<mrow>
<mi>m</mi>
<mi>v</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, α is default ConmThreshold value, β be default CondThreshold value,It is described for default weight threshold
Weight thresholdFor MV mould homogeneity measures ConmEstimate Con with MV orientation consistenciesdFunction；
If i-th of Candidate Motion block BiFormula (3) or formula (4) are met, then i-th of Candidate Motion block BiFor target moving mass.
8. video monitoring system according to claim 7, it is characterised in that the processing server specifically for：
Obtain the ICT coefficient matrixes of each moving image block；
DC component in the ICT coefficient matrixes of each moving image block and low frequency AC components are summed, obtained and each fortune
The corresponding average energy of motion video block；
Determine that average energy is more than the first moving image block of preset value, the first moving image block in each moving image block
For the Candidate Motion block.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310157346.8A CN104125430B (en) | 2013-04-28 | 2013-04-28 | Video moving object detection method, device and video monitoring system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310157346.8A CN104125430B (en) | 2013-04-28 | 2013-04-28 | Video moving object detection method, device and video monitoring system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104125430A CN104125430A (en) | 2014-10-29 |
| CN104125430B true CN104125430B (en) | 2017-09-12 |
Family
ID=51770679
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310157346.8A Active CN104125430B (en) | 2013-04-28 | 2013-04-28 | Video moving object detection method, device and video monitoring system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104125430B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106910200B (en) * | 2015-12-23 | 2019-11-08 | 中国石油天然气股份有限公司 | A kind of moving Object Segmentation method based on phase information |
| CN107229620B (en) * | 2016-03-23 | 2019-11-22 | 杭州海康威视数字技术股份有限公司 | A kind of storage method and device of video data |
| CN106331644A (en) * | 2016-09-07 | 2017-01-11 | 中车唐山机车车辆有限公司 | Video monitoring method, device and train monitoring system |
| CN110083742B (en) * | 2019-04-29 | 2022-12-06 | 腾讯科技（深圳）有限公司 | Video query method and device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1960491A (en) * | 2006-09-21 | 2007-05-09 | 上海大学 | Real time method for segmenting motion object based on H.264 compression domain |
| CN101505426A (en) * | 2009-03-17 | 2009-08-12 | 北京中星微电子有限公司 | Motion detection method and apparatus for compression domain |
| CN101582063A (en) * | 2008-05-13 | 2009-11-18 | 华为技术有限公司 | Video service system, video service device and extraction method for key frame thereof |
| CN102196259A (en) * | 2010-03-16 | 2011-09-21 | 北京中星微电子有限公司 | Moving object detection system and method suitable for compression domain |
| CN102761737A (en) * | 2011-04-28 | 2012-10-31 | 财团法人工业技术研究院 | Systems and methods for processing shadows in compressed video images |
-
2013
- 2013-04-28 CN CN201310157346.8A patent/CN104125430B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1960491A (en) * | 2006-09-21 | 2007-05-09 | 上海大学 | Real time method for segmenting motion object based on H.264 compression domain |
| CN101582063A (en) * | 2008-05-13 | 2009-11-18 | 华为技术有限公司 | Video service system, video service device and extraction method for key frame thereof |
| CN101505426A (en) * | 2009-03-17 | 2009-08-12 | 北京中星微电子有限公司 | Motion detection method and apparatus for compression domain |
| CN102196259A (en) * | 2010-03-16 | 2011-09-21 | 北京中星微电子有限公司 | Moving object detection system and method suitable for compression domain |
| CN102761737A (en) * | 2011-04-28 | 2012-10-31 | 财团法人工业技术研究院 | Systems and methods for processing shadows in compressed video images |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104125430A (en) | 2014-10-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104519401B (en) | Video segmentation point preparation method and equipment | |
| CN105100789B (en) | A kind of method for evaluating video quality | |
| CN103678299B (en) | A kind of method and device of monitor video summary | |
| CN110300977A (en) | Method for image procossing and video compress | |
| CN104125430B (en) | Video moving object detection method, device and video monitoring system | |
| CN105791774A (en) | Surveillance video transmission method based on video content analysis | |
| CN105976383A (en) | Power transmission equipment fault diagnosis method based on limit learning machine image recognition | |
| JP6016332B2 (en) | Image processing apparatus and image processing method | |
| Yu et al. | Exposing frame deletion by detecting abrupt changes in video streams | |
| TW200828176A (en) | Apparatus and method for processing video data | |
| CN110944200B (en) | Method for evaluating immersive video transcoding scheme | |
| JP2004350283A (en) | Method for segmenting compressed video into 3-dimensional objects | |
| CN101316362A (en) | Movie action scene detection method based on story line development model analysis | |
| CN106651901A (en) | Object tracking method and device | |
| CN108921876A (en) | Method for processing video frequency, device and system and storage medium | |
| Akbari et al. | A new forensic video database for source smartphone identification: Description and analysis | |
| CN108875587A (en) | Target distribution detection method and equipment | |
| CN108765405A (en) | A kind of image authenticating method and system | |
| CN110415260A (en) | Smog image segmentation and recognition methods based on dictionary and BP neural network | |
| CN110008102B (en) | Server performance testing method and system based on intelligent video application | |
| Bakas et al. | Mpeg double compression based intra-frame video forgery detection using cnn | |
| AU2021218192B1 (en) | Video anomaly detection method and system based on generative assistant discriminative network | |
| CN109859196A (en) | A kind of image Focus field emission array implementation method based on Modular PCA | |
| CN107492085A (en) | Stereo image quality evaluation method based on dual-tree complex wavelet transform | |
| CN108960346B (en) | Building similarity analysis platform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |