CN106375850B - A kind of judgment method and device matching video - Google Patents
A kind of judgment method and device matching video Download PDFInfo
- Publication number
- CN106375850B CN106375850B CN201510440028.1A CN201510440028A CN106375850B CN 106375850 B CN106375850 B CN 106375850B CN 201510440028 A CN201510440028 A CN 201510440028A CN 106375850 B CN106375850 B CN 106375850B
- Authority
- CN
- China
- Prior art keywords
- video
- template
- section
- measured
- vector
- 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.)
- Expired - Fee Related
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/43—Processing of content or additional data, e.g. demultiplexing additional data from a digital video stream; Elementary client operations, e.g. monitoring of home network or synchronising decoder's clock; Client middleware
- H04N21/44—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream, rendering scenes according to MPEG-4 scene graphs
- H04N21/44008—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream, rendering scenes according to MPEG-4 scene graphs involving operations for analysing video streams, e.g. detecting features or characteristics in the video stream
Abstract
The invention discloses a kind of judgment methods and device for matching video.The described method includes: obtaining multistage video to be measured from video data stream to be measured;Obtain the target feature vector of every section of video to be measured in multistage video to be measured;Multistage template video is obtained from video template library;Obtain the template characteristic vector of every section of template video;Calculate the vector distance between the target feature vector of every section of video to be measured and the template characteristic vector of every section of template video;When the vector distance between the target feature vector of one section of video to be measured and the template characteristic vector of one section of template video is less than or equal to default vector distance, determine that one section of video to be measured matches with one section of template video;Otherwise, it is determined that one section of video to be measured and one section of template video mismatch.According to the technical solution of the present invention, video repeatability detection can be accurately carried out, advertisement prediction is accurately carried out.
Description
Technical field
The present invention relates to technical field of video processing, in particular to a kind of judgment method and device for matching video.
Background technique
Currently, needing to carry out advertisement retrieval and prediction in advertisement processing system.This is actually to carry out video weight
Renaturation detection.But the artificial task amount for carrying out repeated detection is very big, and efficiency is lower, and the accuracy rate and essence of repeatability detection
It spends all relatively low.
Summary of the invention
The present invention provides a kind of judgment method and device for matching video, to use template video accurately to carry out video
Repeatability detection, and ensure the precision and efficiency of video repeatability detection, to realize to the program played in live video stream
Or advertisement is accurately predicted.
The present invention provides a kind of judgment method for matching video, comprising: from video data stream to be measured, it is to be measured to obtain multistage
Video;
Obtain the target feature vector of every section of video to be measured in multistage video to be measured, wherein the target feature vector
For every section of the unique identification video to be measured;
Multistage template video is obtained from video template library;
Obtain the template characteristic vector of every section of template video;
Calculate every section of video to be measured target feature vector and every section of template video template characteristic vector it
Between vector distance;
When the vector between the target feature vector of one section of video to be measured and the template characteristic vector of one section of template video away from
When from being less than or equal to default vector distance, determine that one section of video to be measured matches with one section of template video;Otherwise,
Determine that one section of video to be measured and one section of template video mismatch.
In one embodiment, in the target feature vector and every section of template video for calculating every section of video to be measured
Template characteristic vector between vector distance before, the method also includes:
The template characteristic vector of template video described in multistage is clustered, to obtain multiclass template characteristic vector;And
Match in judgement one section of video to be measured with one section of template video;Otherwise, it is determined that described one section to be measured
Before video and one section of template video mismatch, the method also includes:
When the vector distance between the target feature vector and a kind of template characteristic vector of one section of video to be measured is less than
Or when being equal to default vector distance, one section of video to be measured target template corresponding with one kind template characteristic vector is determined
Video matches, and starts to judge whether one section of video to be measured matches with one section of template video;Otherwise, it is determined that institute
Either segment video to be measured and the target template video is stated to mismatch, and no longer judge one section of video to be measured whether with it is described
One section of template video matches.
In one embodiment, determining one section of video to be measured target corresponding with one kind template characteristic vector
Before template video matches, the method also includes:
Vector distance between the target feature vector and a kind of template characteristic vector of one section of video to be measured
When less than or equal to the default vector distance, by each of one section of video to be measured picture to be measured and the target template
Each target template picture in video is successively matched;
The vector distance when between the target feature vector and a kind of template characteristic vector of one section of video to be measured
When less than or equal to default vector distance, one section of video to be measured target corresponding with one kind template characteristic vector is determined
Template video matches;Otherwise, it is determined that the either segment video to be measured and the target template video mismatch, comprising:
When the vector distance is closest to the default vector distance, and in one section of video to be measured with the target mould
When the number for the picture to be measured that target template picture in plate video matches is higher than preset number, one section of view to be measured is determined
Frequency target template video corresponding with one kind template characteristic vector matches;Otherwise, it is determined that one section of video to be measured with
The corresponding target template video of one kind template characteristic vector mismatches;And
The method also includes:
Determining one section of video to be measured target template video mismatch corresponding with one kind template characteristic vector
Afterwards, the vector distance between the target feature vector of one section of video to be measured is successively being found in multiclass template characteristic vector
Closest to other class template feature vectors of the default vector distance;
Judge in one section of video to be measured in target template video corresponding with other described class template feature vectors
Whether the number for the picture to be measured that target template picture matches is higher than the preset number, and circuits sequentially, until described one
Target Photo in section video to be measured in target template video corresponding with other described class template feature vectors match to
The number of mapping piece is higher than in every section of video to be measured in the preset number or multistage video to be measured and other described class templates
The number for the picture to be measured that target template picture in the corresponding target template video of feature vector matches is below described pre-
If number.
In one embodiment, the template characteristic vector for obtaining every section of template video includes:
Template icon in every section of template video is subjected to DCT conversion, obtains pixel coordinate after DCT is converted
Lateral coefficient and longitudinal coefficient;
The top N coefficient in the preceding M potential coefficient and longitudinal coefficient in the lateral coefficient is chosen, according to the preceding M
Potential coefficient and the top N coefficient generate the template characteristic vector of every section of template video.
In one embodiment, the method also includes:
Determining that one section of video to be measured matches with one section of template video, and the of one section of video to be measured
When the second duration for being greater than one section of template video long for the moment, by one section of video to be measured with one section of template video
In the picture to be measured that does not match that of template picture be added in one section of template video;Or
It, will be in one section of video to be measured when determining that one section of video to be measured and one section of template video mismatch
All pictures to be measured be added in the template video.
The present invention also provides a kind of judgment means for matching video, comprising:
First obtains module, for obtaining multistage video to be measured from video data stream to be measured;
Second obtains module, for obtaining the target feature vector of every section of video to be measured in multistage video to be measured, wherein
The target feature vector is used for every section of the unique identification video to be measured;
Third obtains module, for obtaining multistage template video from video template library;
4th obtains module, for obtaining the template characteristic vector of every section of template video;
Computing module, for calculating the target feature vector of every section of video to be measured and the mould of every section of template video
Vector distance between plate features vector;
First determination module, for working as the target feature vector of one section of video to be measured and the template characteristic of one section of template video
When vector distance between vector is less than or equal to default vector distance, one section of video to be measured and one section of template are determined
Video matches;Otherwise, it is determined that one section of video to be measured and one section of template video mismatch.
In one embodiment, cluster module, in the target feature vector for calculating every section of video to be measured and often
Before vector distance between the template characteristic vector of Duan Suoshu template video, by the template characteristic of template video described in multistage to
Amount is clustered, to obtain multiclass template characteristic vector;And
Second determination module, for matching in judgement one section of video to be measured with one section of template video;Otherwise,
Before determining that one section of video to be measured and one section of template video mismatch, when the target signature of one section of video to be measured
When vector distance between vector and a kind of template characteristic vector is less than or equal to default vector distance, described one section of judgement to be measured
Video target template video corresponding with one kind template characteristic vector matches, and starts to judge one section of video to be measured
Whether match with one section of template video;Otherwise, it is determined that the either segment video to be measured and the target template video are not
Matching, and no longer judge whether one section of video to be measured matches with one section of template video.
In one embodiment, described device further include:
Matching module, for determining one section of video to be measured target mould corresponding with one kind template characteristic vector
Before plate video matches, when between the target feature vector and a kind of template characteristic vector of one section of video to be measured
Vector distance be less than or equal to the default vector distance when, by each of one section of video to be measured picture to be measured with it is described
Each target template picture in target template video is successively matched;
Second determination module, comprising:
Decision sub-module, for when the vector distance is closest to the default vector distance, and one section of view to be measured
When the number of the picture to be measured to match in frequency with the target template picture in the target template video is higher than preset number, sentence
Fixed one section of video to be measured target template video corresponding with one kind template characteristic vector matches;Otherwise, it is determined that institute
One section of video to be measured target template video corresponding with one kind template characteristic vector is stated to mismatch;And
Described device further include:
Module is found, for determining one section of video to be measured target mould corresponding with one kind template characteristic vector
After plate video mismatches, successively found in multiclass template characteristic vector with the target feature vector of one section of video to be measured it
Between vector distance closest to the default vector distance other class template feature vectors;
Judgment module, for judging target corresponding with other described class template feature vectors in one section of video to be measured
Whether the number for the picture to be measured that the target template picture in template video matches is higher than the preset number, and successively follows
Ring, the target figure in target template video corresponding with other described class template feature vectors in one section of video to be measured
The number for the picture to be measured that piece matches is higher than in every section of video to be measured in the preset number or multistage video to be measured and institute
State the number for the picture to be measured that the target template picture in the corresponding target template video of other class template feature vectors matches
It is below the preset number.
In one embodiment, the 4th acquisition module includes:
Transform subblock, for obtain every section of template video template characteristic vector before, by every section of mould
Template icon in plate video carries out DCT conversion, obtains the lateral coefficient and longitudinal coefficient of pixel coordinate after DCT conversion;
Submodule is chosen, for choosing the top N system in preceding M potential coefficient and longitudinal coefficient in the lateral coefficient
Number;
Submodule is generated, for generating every section of template video according to the preceding M potential coefficient and the top N coefficient
Template characteristic vector.
In one embodiment, described device further include:
First adding module, for matching in judgement one section of video to be measured with one section of template video, and institute
State one section of video to be measured the first duration be greater than one section of template video the second duration when, will be in one section of video to be measured
It is added in one section of template video with the picture to be measured that the template picture in one section of template video does not match that;Or
Second adding module is used for when determining that one section of video to be measured and one section of template video mismatch, will
All pictures to be measured in one section of video to be measured are added in the template video.
The technical scheme provided by this disclosed embodiment can include the following benefits:
It can be according between the target feature vector of every section of video to be measured and the template characteristic vector of every section of template video
Vector distance, determines the similarity of every section of video to be measured and every section of template video, so determine in this section of video to be measured whether with
This section of template video matches, and then realization can accurately carry out video repeatability detection, and ensures the detection of video repeatability
Precision and efficiency, and finally realize and accurately determine specific broadcast start time of the template video in video to be measured, thus real
Now accurately carry out advertisement prediction.
Other features and advantages of the present invention will be illustrated in the following description, also, partly becomes from specification
It obtains it is clear that understand through the implementation of the invention.The objectives and other advantages of the invention can be by written explanation
Specifically noted structure is achieved and obtained in book, claims and attached drawing.
Below by drawings and examples, technical scheme of the present invention will be described in further detail.
Detailed description of the invention
Attached drawing is used to provide further understanding of the present invention, and constitutes part of specification, with reality of the invention
It applies example to be used to explain the present invention together, not be construed as limiting the invention.In the accompanying drawings:
Fig. 1 is a kind of flow chart of judgment method for matching video shown according to an exemplary embodiment.
Fig. 2 is a kind of block diagram of judgment means for matching video shown according to an exemplary embodiment.
Fig. 3 is the block diagram of the judgment means of another matching video shown according to an exemplary embodiment.
Fig. 4 is the block diagram of the judgment means of another matching video shown according to an exemplary embodiment.
Fig. 5 is the block diagram of the judgment means of another matching video shown according to an exemplary embodiment.
Fig. 6 A is the block diagram of the judgment means of another matching video shown according to an exemplary embodiment.
Fig. 6 B is the block diagram of the judgment means of another matching video shown according to an exemplary embodiment.
Specific embodiment
Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings, it should be understood that preferred reality described herein
Apply example only for the purpose of illustrating and explaining the present invention and is not intended to limit the present invention.
In the related technology, it currently, in advertisement processing system, needs to carry out advertisement retrieval and prediction.This is actually
Carry out video repeatability detection.But the artificial task amount for carrying out repeated detection is very big, and efficiency is lower, and repeatability detection
Accuracy rate and precision it is all relatively low.
In order to solve the above-mentioned technical problem, the embodiment of the present disclosure provides a kind of judgment method for matching video, this method
Suitable for the determining program, system or the device that match video, as shown in Figure 1, step S101, from video data stream to be measured,
Obtain multistage video to be measured;
Step S102 obtains the target feature vector of every section of video to be measured in multistage video to be measured, wherein target signature
Vector is used for every section of unique identification video to be measured;Target feature vector is used for every section of unique identification video to be measured, is every section to be measured
The video finger print of the video to be measured of the finerprint of video, that is, every section.
Step S103 obtains multistage template video from video template library;
Video template library can be several derived customized channels, the i.e. view such as most interior in a couple of days history program, advertisement
Frequency evidence is stored in video template library, such as: can be most in a couple of days in several customized channels program, wide
The video datas such as announcement, specifically: therefore the pictures such as all programs, advertisement from program start time to finish time use this
Known video template library (it can be multiple programs or advertisement, and every section of template video can include an independent advertisement
Or program) this program or advertisement can be searched in the real time programme stream of all channels, and searched in video data stream to be measured
When rope is to the program or advertisement, that is, it can determine the specific beginning of the program or advertisement in video data stream to be measured being played on
Time accurately carries out advertisement or the prediction of program to realize.Certainly, if prediction programme information, every section of template video are
The video-frequency band of program head;If predicting advertising information, every section of template video chooses the video-frequency band of advertisement
Step S104 obtains the template characteristic vector of every section of template video;In the template characteristic for obtaining every section of template video
When vector, each template picture in every section of template video can be uniformly reduced into size be W*H=720*480, so as into
Row is uniformly processed;And template characteristic vector is used for every section of template video of unique identification, is the finerprint of every section of template video
The video finger print of i.e. every section template video.
Step S105, calculate every section of video to be measured target feature vector and every section of template video template characteristic vector it
Between vector distance;Wherein, vector distance is used to characterize the similarity between every section of video to be measured and every section of template video, and swears
For span from smaller, similarity is higher;By the target feature vector and every section of template video that successively calculate every section of video to be measured
Vector distance between template characteristic vector can accurately determine similar between every section of video to be measured and every section of template video
Degree, accurately locking with the highest video-frequency band to be measured of every section of template video matching rate, it is ensured that video repeatability detection precision and
Efficiency, and prevent detection and omit, and vector distance is smaller, similarity is higher, illustrates this section of video to be measured and this section of template
The matching rate of video is higher.
Step S106, when between the target feature vector of one section of video to be measured and the template characteristic vector of one section of template video
Vector distance when being less than or equal to default vector distance, determine that one section of video to be measured matches with one section of template video;Otherwise,
Determine that one section of video to be measured and one section of template video mismatch.
When the vector between the target feature vector of this section of video to be measured and the template characteristic vector of this section of template video away from
When from being less than or equal to default vector distance, illustrate that the similarity of this section of video to be measured and this section of template video is high, and then say
The matching rate of bright this section of video to be measured and this section of template video is high, it may be considered that this section of video to be measured and this section of template video
It repeats, and then convenient for determining starting position of this section of template video in video data stream to be measured, thus accurately to view to be measured
Program being played on or advertisement (i.e. the template video) are predicted in frequency data stream.Certainly, it is greater than in the vector distance general
When default vector distance, illustrate that the similarity of this section of video to be measured and this section of template video is extremely low, so this section of video to be measured and
The matching rate of this section of template video is extremely low, and the repetitive rate of two sections of videos is extremely low, it may be considered that this section of video to be measured and the Duan Mo
Plate video mismatches.
In one embodiment, before step S105, method further include:
The template characteristic vector of multistage template video is clustered, to obtain multiclass template characteristic vector, wherein every class
Template characteristic vector is a cluster centre;
Since every section of template video corresponds to a template characteristic vector, and when the number of segment that template video includes is excessive,
The number of template characteristic vector will be excessive, therefore, if directly calculating the target feature vector of every section of video to be measured and every section
Vector distance between the template characteristic vector of template video will necessarily make since template characteristic vector is excessive, and influence arrow
Span from computation rate, increase the computation burden of system, therefore, by by the template of the essentially identical template video of similarity
Feature vector is clustered, can make cluster after template characteristic vector number of vectors significantly reduces (such as: cluster
The quantity of the vector in template characteristic vector afterwards is 5% before cluster).
And
Before above-mentioned steps S106, method further include:
When the vector distance between the target feature vector and a kind of template characteristic vector of one section of video to be measured is less than or waits
When default vector distance, determine that (certain in i.e. multiple cluster centres is a kind of for one section of video to be measured and a kind of template characteristic vector
Cluster centre) corresponding target template video matches, and start to judge one section of video to be measured whether with one section of template video phase
Matching;Otherwise, it is determined that either segment video to be measured and target template video mismatch, and no longer judge one section of video to be measured whether with
One section of template video matches, wherein the corresponding target template video of every class cluster centre is by the higher several segments of similarity
Template video composition, i.e., every class cluster centre are made of several higher template characteristic vectors of similarity, wherein this is several
The corresponding video of a template characteristic vector is the several segments template video, and one section of video to be measured be one kind template characteristic to
Measure one section of video to be measured in corresponding target template video.
Due to judge one by one every section of video to be measured whether with the matched workload of every section of template video very big, the place of system
Reason burden is very heavy, therefore, can be with after the template characteristic vector for the template video that similarity is essentially identical is clustered
The vector distance between the target feature vector of every section of video to be measured and every class template feature vector is first calculated, if the vector
Distance is less than default vector distance, then illustrates that the similarity of this section of video to be measured and the target template video is high, and then can be with
Tentatively judge that this section of video to be measured target template video corresponding with such template characteristic vector repeats, and in order to further accurate
The specific airtime of every section of template video in video data stream to be measured is locked, can further judge that this section of video to be measured is
A certain section of template video matching in no target template video corresponding with such template characteristic vector, and then convenient for accurately locking
Fixed specific location namely specific airtime of the every section of template video in video data stream to be measured;Certainly, if the vector away from
From default vector distance is greater than or equal to, then illustrate that the similarity of this section of video to be measured and the target template video is extremely low, substantially
It is considered that completely mismatch, then be completely unnecessary further judge this section of video to be measured whether in the target template video
Certain section of template video match, thus alleviate system video matching detection and prediction burden.In addition, can be with when cluster
Use K-means clustering algorithm.
In one embodiment, determining one section of video to be measured target template video corresponding with a kind of template characteristic vector
Before matching, method further include:
When the vector distance between the target feature vector and a kind of template characteristic vector of one section of video to be measured is less than or waits
When default vector distance, by each target template in each of one section of video to be measured picture to be measured and target template video
Picture is successively matched;
If the vector distance is less than default vector distance, this section of video to be measured and such template characteristic can only be tentatively judged
The corresponding target template video of vector matches, and in order to further be accurately determined this section of video to be measured and such template characteristic
Whether the corresponding target template video of vector really matches, and needs to go deep into relatively each of this section of video to be measured picture to be measured
Whether matched with each target template picture in the target template video, to ensure that video repeats the precision of detection, and then really
Protect the advertisement precision of prediction in video data stream to be measured.
When the vector distance between the target feature vector and a kind of template characteristic vector of one section of video to be measured is less than or waits
When default vector distance, determine that one section of video to be measured target template video corresponding with a kind of template characteristic vector matches;
Otherwise, it is determined that either segment video to be measured and target template video mismatch, comprising:
Vector distance (i.e. one kind template characteristic vector is immediate cluster centre) is preset when vector distance is closest,
And the number of the picture to be measured to match in one section of video to be measured with the target template picture in target template video is higher than default
When number, determine that one section of video to be measured target template video corresponding with a kind of template characteristic vector matches;Otherwise, it is determined that one
Section video to be measured target template video corresponding with a kind of template characteristic vector mismatches;
When the vector distance between the target feature vector and such template characteristic vector of this section of video to be measured is less than or waits
Preset vector distance in this, and when its vector distance is closest to default vector distance, if in this section of video to be measured picture to be measured with
Corresponding target template picture matches number higher than preset number, then illustrates the phase of this section of video to be measured and target template video
It is higher like spending, and then can be determined that this section of video to be measured matches with the target template video, otherwise, illustrate this section of video to be measured
It is lower with the similarity of the target template video, and then not can be determined that this section of video to be measured and the target template video completely not
Match, certainly, is determining the vector distance between the target feature vector of a certain section of video to be measured and a certain class template feature vector
Whether most when the proximity default vector distance, need to calculate this section of video to be measured target feature vector and multiclass template characteristic to
Then the vector distance between every class template feature vector in amount compares each vector distance and the default vector distance
The immediate vector distance of vector distance is preset with this compared with can therefrom select, is most connect so that it is determined that presetting vector distance with this
The target template video of close a certain class template feature vector.
And
Method further include:
After determining that one section of video to be measured target template video corresponding with a kind of template characteristic vector mismatches, successively exist
It finds in multiclass template characteristic vector and is sweared the vector distance between the target feature vector of one section of video to be measured closest to default
Span from other class template feature vectors (i.e. other nearly cluster centres);
After determining that this section of video to be measured and the target template video mismatch, it can be sought in multiclass template characteristic vector
Look for other nearly class mould of vector distance closest to the default vector distance between the target feature vector of this section of video to be measured
Plate features vector, with judge the corresponding target template video of time close class template feature vector whether with this section of video matching to be measured.
Judge the target template in one section of video to be measured in target template video corresponding with other class template feature vectors
Whether the number for the picture to be measured that picture matches is higher than preset number, and circuits sequentially, until in one section of video to be measured with its
The number for the picture to be measured that Target Photo in the corresponding target template video of his class template feature vector matches is higher than default
In every section of video to be measured in number or multistage video to be measured in target template video corresponding with other class template feature vectors
The number of picture to be measured that matches of target template picture be below preset number.
When the vector between the target feature vector of this section of video to be measured and some other time nearly class template feature vectors away from
When from being less than or equal to default vector distance, if picture to be measured and this other time nearly class template feature vector in this section of video to be measured
Corresponding target template picture matches number higher than preset number, then illustrates this section of video to be measured and other the nearly class template
The similarity that feature vector corresponds to target template video is higher, and then can be determined that this section of video to be measured and this other time nearly class mould
Plate features vector corresponds to target template video and matches, and otherwise, illustrates this section of video to be measured and this other time nearly class template feature
The similarity that vector corresponds to target template video is lower, and then can determine that this section of video to be measured and this other time nearly class template are special
Sign vector corresponds to target template video mismatch;Certainly, if in this section of video to be measured with every other time nearly class template feature
The number for the picture to be measured that target template picture in the corresponding target template video of vector matches is below preset number, then
Illustrate that this section of video to be measured and the multistage template video selected all mismatch, then can jump out circulation, other sections of reselection to
It surveys video to be matched with each target template video, accurately to determine in the video data stream to be measured whether played every section of mould
Plate video, and if playing has a certain section of template video, specifically play the time of this section of template video.
In one embodiment, above-mentioned steps S104 can be performed as:
Template icon in every section of template video is subjected to DCT conversion, obtains the transverse direction of pixel coordinate after DCT conversion
Coefficient and longitudinal coefficient;
The top N coefficient in the preceding M potential coefficient and longitudinal coefficient in lateral coefficient is chosen, according to preceding M potential coefficient and top N
Coefficient generates the template characteristic vector of every section of template video.
When generating template characteristic vector, more forward several potential coefficients are characterizing the template in lateral coefficient and longitudinal coefficient
Shared weight is bigger when picture, the template characteristic vector of generation more can accurately indicate the template picture, and if selection
Coefficient digit is more, and the calculation amount of system will be bigger, and the computation burden of system will be heavier, the spy of calculated template picture
The real-time for levying vector will be lower, therefore, can be with integrated template picture feature vector when generating template picture feature vector
The requirement of real-time of permissible accuracy, the computing capability of system and feature vector is chosen the preceding M potential coefficient in lateral coefficient and is indulged
Top N coefficient into coefficient, to meet required precision, the computing capability of system and the reality of feature vector of feature vector simultaneously
When property requirement, wherein M and N is preset value, it is preferable that the value of M and N is 1.
In one embodiment, method further include:
Match in one section of video to be measured of judgement with one section of template video, and the first duration of one section of video to be measured is greater than one
When the second duration of section template video, will not be matched that with the template picture in one section of template video in one section of video to be measured to
Mapping piece is added in one section of template video;
It, can be by this section of view to be measured when the first duration of this section of video to be measured is greater than the second duration of this section of template video
The picture to be measured not matched that in frequency with the template key frame picture in this section of template video is added in this section of template video, from
And keep this section of template video more abundant, complete the update to template video.Or
It, will be all to mapping in one section of video to be measured when determining that one section of video to be measured and one section of template video mismatch
Piece is added in template video.
When this section of video to be measured is mismatched with this section of template video, illustrate to be not present and the Duan Mo in this section of video to be measured
When the picture to be measured that the template picture in plate video matches, each of this section of video to be measured picture to be measured can be added to
In template video, to update the template video to a greater degree.
As shown in Fig. 2, the present invention also provides a kind of judgment means for matching video, comprising:
First obtains module 201, is configured as from video data stream to be measured, obtains multistage video to be measured;
Second obtain module 202, be configured as obtain multistage video to be measured in every section of video to be measured target signature to
Amount, wherein the target feature vector is configured as every section of the unique identification video to be measured;
Third obtains module 203, is configured as obtaining multistage template video from video template library;
4th obtains module 204, is configured as obtaining the template characteristic vector of every section of template video;
Computing module 205 is configured as calculating the target feature vector and every section of template of every section of video to be measured
Vector distance between the template characteristic vector of video;
First determination module 206 is configured as target feature vector and one section of template video when one section of video to be measured
When vector distance between template characteristic vector is less than or equal to default vector distance, determine one section of video to be measured with it is described
One section of template video matches;Otherwise, it is determined that one section of video to be measured and one section of template video mismatch.
As shown in figure 3, in one embodiment, described device further include: cluster module 301 is configured as calculating often
Vector distance between the target feature vector of Duan Suoshu video to be measured and the template characteristic vector of every section of template video it
Before, the template characteristic vector of template video described in multistage is clustered, to obtain multiclass template characteristic vector;And
Second determination module 302 is configured as determining one section of video to be measured and one section of template video phase
Match;Otherwise, it is determined that before one section of video to be measured and one section of template video mismatch, when one section of video to be measured
When vector distance between target feature vector and a kind of template characteristic vector is less than or equal to default vector distance, described in judgement
One section of video to be measured target template video corresponding with one kind template characteristic vector matches, and starts to judge described one section
Whether video to be measured matches with one section of template video;Otherwise, it is determined that the either segment video to be measured and the target mould
Plate video mismatches, and no longer judges whether one section of video to be measured matches with one section of template video.
As shown in figure 4, in one embodiment, described device further include:
Matching module 401 is configured as determining that one section of video to be measured is corresponding with a kind of template characteristic vector
Target template video match before, when one section of video to be measured target feature vector and a kind of template characteristic to
When vector distance between amount is less than or equal to the default vector distance, by each of described one section of video to be measured to mapping
Piece is successively matched with each target template picture in the target template video;
Second determination module 302, comprising:
Decision sub-module 3021 is configured as when the vector distance is closest to the default vector distance, and described one
The number of the picture to be measured to match in section video to be measured with the target template picture in the target template video is higher than default
When number, determine that one section of video to be measured target template video corresponding with one kind template characteristic vector matches;It is no
Then, determine that one section of video to be measured target template video corresponding with one kind template characteristic vector mismatches;And
Described device further include:
Module 402 is found, is configured as determining that one section of video to be measured is corresponding with a kind of template characteristic vector
Target template video mismatch after, successively found in multiclass template characteristic vector special with the target of one section of video to be measured
Levy vector between vector distance closest to the default vector distance other class template feature vectors;
Judgment module 403 is configured as judging in one section of video to be measured and other described class template feature vectors pair
Whether the number for the picture to be measured that the target template picture in target template video answered matches is higher than the preset number, and
It circuits sequentially, until in target template video corresponding with other described class template feature vectors in one section of video to be measured
The number for the picture to be measured that Target Photo matches is higher than every section of video to be measured in the preset number or multistage video to be measured
In the picture to be measured that matches of target template picture in target template video corresponding with other described class template feature vectors
Number be below the preset number.
As shown in figure 5, in one embodiment, the 4th acquisition module 204 includes:
Transform subblock 2041 is configured as before the template characteristic vector for obtaining every section of template video, will be every
Template icon in Duan Suoshu template video carries out DCT conversion, obtains the lateral coefficient of pixel coordinate after DCT is converted and indulges
To coefficient;
Submodule 2042 is chosen, is configured as choosing in the preceding M potential coefficient and longitudinal coefficient in the lateral coefficient
Top N coefficient;
Submodule 2043 is generated, is configured as generating every section of mould according to the preceding M potential coefficient and the top N coefficient
The template characteristic vector of plate video.
As shown in Figure 6A, in one embodiment, described device further include:
First adding module 601 is configured as determining one section of video to be measured and one section of template video phase
Match, and the first duration of one section of video to be measured be greater than one section of template video the second duration when, by described one section to
It surveys and is added to one section of template view in video with the picture to be measured that the template picture in one section of template video does not match that
In frequency;Or
As shown in Figure 6B, in one embodiment, described device further include:
Second adding module 602 is configured as not determining one section of video to be measured and one section of template video not
All pictures to be measured in one section of video to be measured are added in the template video by timing.
About the device in above-described embodiment, wherein modules execute the concrete mode of operation in related this method
Embodiment in be described in detail, no detailed explanation will be given here.
Finally, the judgment means of the matching video in the present invention are suitable for terminal device.For example, it may be mobile phone,
Computer, digital broadcasting terminal, messaging device, game console, tablet device, Medical Devices, body-building equipment, a number
Word assistant etc..
It should be understood by those skilled in the art that, the embodiment of the present invention can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the present invention
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the present invention, which can be used in one or more,
The shape for the computer program product implemented in usable storage medium (including but not limited to magnetic disk storage and optical memory etc.)
Formula.
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Obviously, various changes and modifications can be made to the invention without departing from essence of the invention by those skilled in the art
Mind and range.In this way, if these modifications and changes of the present invention belongs to the range of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to include these modifications and variations.
Claims (6)
1. a kind of judgment method for matching video characterized by comprising
From video data stream to be measured, multistage video to be measured is obtained;
Obtain the target feature vector of every section of video to be measured in multistage video to be measured, wherein the target feature vector is used for
Every section of the unique identification video to be measured;
Multistage template video is obtained from video template library;
Obtain the template characteristic vector of every section of template video;
It calculates between the target feature vector of every section of video to be measured and the template characteristic vector of every section of template video
Vector distance;
When the vector distance between the target feature vector of one section of video to be measured and the template characteristic vector of one section of template video is small
When the first default vector distance, determine that one section of video to be measured matches with one section of template video;Otherwise,
Determine that one section of video to be measured and one section of template video mismatch;
Between the template characteristic vector of target feature vector and every section of template video for calculating every section of video to be measured
Vector distance before, the method also includes:
The template characteristic vector of template video described in multistage is clustered, to obtain multiclass template characteristic vector;And
Match in judgement one section of video to be measured with one section of template video;Otherwise, it is determined that one section of video to be measured
Before being mismatched with one section of template video, the method also includes:
When the vector distance between the target feature vector and a kind of template characteristic vector of one section of video to be measured is less than or waits
When the second default vector distance, one section of video to be measured target template corresponding with one kind template characteristic vector is determined
Video matches, and starts to judge whether one section of video to be measured matches with one section of template video;Otherwise, it is determined that institute
It states one section of video to be measured and the target template video to mismatch, and no longer judges one section of video to be measured whether with described one
Section template video matches;
Before determining that one section of video to be measured target template video corresponding with one kind template characteristic vector matches,
The method also includes:
When the vector distance between the target feature vector and a kind of template characteristic vector of one section of video to be measured is less than
Or when being equal to the second default vector distance, by each of one section of video to be measured picture to be measured and the target template
Each target template picture in video is successively matched;
The vector distance when between the target feature vector and a kind of template characteristic vector of one section of video to be measured is less than
Or when being equal to the second default vector distance, one section of video to be measured target corresponding with one kind template characteristic vector is determined
Template video matches;Otherwise, it is determined that one section of video to be measured and the target template video mismatch, comprising:
When the vector distance between the target feature vector and a kind of template characteristic vector of one section of video to be measured is closest to institute
State the second default vector distance, and in one section of video to be measured with the target template picture phase in the target template video
When the number for the picture to be measured matched is higher than preset number, one section of video to be measured and a kind of template characteristic vector pair are determined
The target template video answered matches;Otherwise, it is determined that one section of video to be measured is corresponding with a kind of template characteristic vector
Target template video mismatches;And
The method also includes:
After determining that one section of video to be measured target template video corresponding with one kind template characteristic vector mismatches, according to
The secondary vector distance in multiclass template characteristic vector between searching and the target feature vector of one section of video to be measured most connects
Other class template feature vectors of the nearly second default vector distance;
Judge the target in one section of video to be measured in target template video corresponding with other described class template feature vectors
Whether the number for the picture to be measured that template picture matches is higher than the preset number, and circuits sequentially, until described one section to
Survey that the Target Photo in video in target template video corresponding with other described class template feature vectors matches to mapping
The number of piece is higher than in every section of video to be measured in the preset number or multistage video to be measured and other described class template features
The number for the picture to be measured that target template picture in the corresponding target template video of vector matches is below the present count
Mesh.
2. the method according to claim 1, wherein it is described obtain every section of template video template characteristic to
Amount includes:
Template icon in every section of template video is subjected to DCT conversion, obtains the transverse direction of pixel coordinate after DCT conversion
Coefficient and longitudinal coefficient;
The top N coefficient in the preceding M potential coefficient and longitudinal coefficient in the lateral coefficient is chosen, according to the preceding M system
The several and top N coefficient generates the template characteristic vector of every section of template video.
3. method according to claim 1 or 2, which is characterized in that the method also includes:
Determining that one section of video to be measured matches with one section of template video, and one section of video to be measured first when
It, will be in one section of video to be measured and in one section of template video when long the second duration for being greater than one section of template video
The picture to be measured that template picture does not match that is added in one section of template video;Or
When determining that one section of video to be measured and one section of template video mismatch, by the institute in one section of video to be measured
Need mapping piece to be added in the template video.
4. a kind of judgment means for matching video characterized by comprising
First obtains module, for obtaining multistage video to be measured from video data stream to be measured;
Second obtains module, for obtaining the target feature vector of every section of video to be measured in multistage video to be measured, wherein described
Target feature vector is used for every section of the unique identification video to be measured;
Third obtains module, for obtaining multistage template video from video template library;
4th obtains module, for obtaining the template characteristic vector of every section of template video;
Computing module, for calculating the target feature vector of every section of video to be measured and the template spy of every section of template video
Levy the vector distance between vector;
First determination module, for working as the target feature vector of one section of video to be measured and the template characteristic vector of one section of template video
Between vector distance when being less than or equal to the first default vector distance, determine one section of video to be measured and one section of template
Video matches;Otherwise, it is determined that one section of video to be measured and one section of template video mismatch;
Described device further include:
Cluster module, for the template in the target feature vector and every section of template video for calculating every section of video to be measured
Before vector distance between feature vector, the template characteristic vector of template video described in multistage is clustered, it is more to obtain
Class template feature vector;And
Second determination module, for matching in judgement one section of video to be measured with one section of template video;Otherwise, it is determined that
Before one section of video to be measured and one section of template video mismatch, when the target feature vector of one section of video to be measured
When vector distance between a kind of template characteristic vector is less than or equal to the second default vector distance, described one section of judgement to be measured
Video target template video corresponding with one kind template characteristic vector matches, and starts to judge one section of video to be measured
Whether match with one section of template video;Otherwise, it is determined that one section of video to be measured and the target template video are not
Match, and no longer judges whether one section of video to be measured matches with one section of template video;
Described device further include:
Matching module, for determining one section of video to be measured target template view corresponding with one kind template characteristic vector
Vector before frequency matches, between the target feature vector and a kind of template characteristic vector of one section of video to be measured
Distance be less than or equal to the second default vector distance when, by each of one section of video to be measured picture to be measured with it is described
Each target template picture in target template video is successively matched;
Second determination module, comprising:
Decision sub-module, for when the arrow between the target feature vector of one section of video to be measured and a kind of template characteristic vector
Span from closest to the described second default vector distance, and in one section of video to be measured with the mesh in the target template video
When the number for the picture to be measured that mark template picture matches is higher than preset number, one section of video to be measured and described one kind are determined
The corresponding target template video of template characteristic vector matches;Otherwise, it is determined that one section of video to be measured and a class template
The corresponding target template video of feature vector mismatches;And
Described device further include:
Module is found, for determining one section of video to be measured target template view corresponding with one kind template characteristic vector
After frequency mismatches, successively found between the target feature vector of one section of video to be measured in multiclass template characteristic vector
Other class template feature vectors of vector distance closest to the described second default vector distance;
Judgment module, for judging target template corresponding with other described class template feature vectors in one section of video to be measured
Whether the number for the picture to be measured that the target template picture in video matches is higher than the preset number, and circuits sequentially, directly
Target Photo phase into one section of video to be measured in target template video corresponding with other described class template feature vectors
The number of matched picture to be measured be higher than in every section of video to be measured in the preset number or multistage video to be measured with it is described its
The number for the picture to be measured that target template picture in the corresponding target template video of his class template feature vector matches is low
In the preset number.
5. device according to claim 4, which is characterized in that the described 4th, which obtains module, includes:
Transform subblock, for before the template characteristic vector for obtaining every section of template video, every section of template to be regarded
Template icon in frequency carries out DCT conversion, obtains the lateral coefficient and longitudinal coefficient of pixel coordinate after DCT conversion;
Submodule is chosen, for choosing the top N coefficient in preceding M potential coefficient and longitudinal coefficient in the lateral coefficient;
Submodule is generated, for generating the template of every section of template video according to the preceding M potential coefficient and the top N coefficient
Feature vector.
6. device according to claim 4 or 5, which is characterized in that described device further include:
First adding module, for matching in judgement one section of video to be measured with one section of template video, and described one
When first duration of section video to be measured is greater than the second duration of one section of template video, by one section of video to be measured with institute
The picture to be measured that the template picture in one section of template video does not match that is stated to be added in one section of template video;Or
Second adding module is used for when determining that one section of video to be measured and one section of template video mismatch, will be described
All pictures to be measured in one section of video to be measured are added in the template video.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510440028.1A CN106375850B (en) | 2015-07-23 | 2015-07-23 | A kind of judgment method and device matching video |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510440028.1A CN106375850B (en) | 2015-07-23 | 2015-07-23 | A kind of judgment method and device matching video |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106375850A CN106375850A (en) | 2017-02-01 |
CN106375850B true CN106375850B (en) | 2019-09-13 |
Family
ID=57880277
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510440028.1A Expired - Fee Related CN106375850B (en) | 2015-07-23 | 2015-07-23 | A kind of judgment method and device matching video |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106375850B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110324660B (en) | 2018-03-29 | 2021-01-19 | 北京字节跳动网络技术有限公司 | Method and device for judging repeated video |
CN114422841B (en) * | 2021-12-17 | 2024-01-02 | 北京达佳互联信息技术有限公司 | Subtitle generation method and device, electronic equipment and storage medium |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101359368A (en) * | 2008-09-09 | 2009-02-04 | 华为技术有限公司 | Video image clustering method and system |
CN101398854A (en) * | 2008-10-24 | 2009-04-01 | 清华大学 | Video fragment searching method and system |
CN101833650A (en) * | 2009-03-13 | 2010-09-15 | 清华大学 | Video copy detection method based on contents |
CN101853377A (en) * | 2010-05-13 | 2010-10-06 | 复旦大学 | Method for identifying content of digital video |
CN102693299A (en) * | 2012-05-17 | 2012-09-26 | 西安交通大学 | System and method for parallel video copy detection |
CN103593464A (en) * | 2013-11-25 | 2014-02-19 | 华中科技大学 | Video fingerprint detecting and video sequence matching method and system based on visual features |
CN103617233A (en) * | 2013-11-26 | 2014-03-05 | 烟台中科网络技术研究所 | Method and device for detecting repeated video based on semantic content multilayer expression |
CN103761252A (en) * | 2013-12-25 | 2014-04-30 | 北京航天测控技术有限公司 | Video retrieval method |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8063295B2 (en) * | 2002-10-03 | 2011-11-22 | Polyphonic Human Media Interface, S.L. | Method and system for video and film recommendation |
US8195689B2 (en) * | 2009-06-10 | 2012-06-05 | Zeitera, Llc | Media fingerprinting and identification system |
-
2015
- 2015-07-23 CN CN201510440028.1A patent/CN106375850B/en not_active Expired - Fee Related
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101359368A (en) * | 2008-09-09 | 2009-02-04 | 华为技术有限公司 | Video image clustering method and system |
CN101398854A (en) * | 2008-10-24 | 2009-04-01 | 清华大学 | Video fragment searching method and system |
CN101833650A (en) * | 2009-03-13 | 2010-09-15 | 清华大学 | Video copy detection method based on contents |
CN101853377A (en) * | 2010-05-13 | 2010-10-06 | 复旦大学 | Method for identifying content of digital video |
CN102693299A (en) * | 2012-05-17 | 2012-09-26 | 西安交通大学 | System and method for parallel video copy detection |
CN103593464A (en) * | 2013-11-25 | 2014-02-19 | 华中科技大学 | Video fingerprint detecting and video sequence matching method and system based on visual features |
CN103617233A (en) * | 2013-11-26 | 2014-03-05 | 烟台中科网络技术研究所 | Method and device for detecting repeated video based on semantic content multilayer expression |
CN103761252A (en) * | 2013-12-25 | 2014-04-30 | 北京航天测控技术有限公司 | Video retrieval method |
Also Published As
Publication number | Publication date |
---|---|
CN106375850A (en) | 2017-02-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106375781B (en) | A kind of judgment method and device repeating video | |
US10147018B2 (en) | Image processing apparatus, image processing method, and storage medium | |
CN101038592B (en) | Method and apparatus for representing a group of images | |
US20140205158A1 (en) | Information processing apparatus, information processing method, and program | |
CN100499778C (en) | Method and device for camera model parameter estimation using iterative center of mass | |
CN106649855B (en) | A kind of adding method and add-on system of video tab | |
US20110170600A1 (en) | Image processing apparatus, image processing method and image processing program | |
US9679380B2 (en) | Emotion modification for image and video content | |
US20200374526A1 (en) | Method, device, apparatus for predicting video coding complexity and storage medium | |
CN109409321B (en) | Method and device for determining lens movement mode | |
CN106375849B (en) | A kind of method, apparatus, the update method of video and device generating template | |
CN104918060A (en) | Method and device for selecting position to insert point in video advertisement | |
CN106375850B (en) | A kind of judgment method and device matching video | |
KR20220098312A (en) | Method, apparatus, device and recording medium for detecting related objects in an image | |
CN100593948C (en) | Method and device for jointing video | |
CN107426610B (en) | Video information synchronization method and device | |
US8837595B2 (en) | Motion estimation method | |
CN106354736A (en) | Judgment method and device of repetitive video | |
CN110348367A (en) | Video classification methods, method for processing video frequency, device, mobile terminal and medium | |
CN106372092A (en) | Method and device for generating template, and video update method and device | |
KR20200053543A (en) | Cold matching by automatic content recognition | |
CN104301805A (en) | Method and device for estimating time span of video | |
JP4997179B2 (en) | Image processing apparatus, method, and program | |
US8391365B2 (en) | Motion estimator and a motion estimation method | |
CN106412690B (en) | Video playing determination method and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: A judgment method and device for matching video Effective date of registration: 20210104 Granted publication date: 20190913 Pledgee: Inner Mongolia Huipu Energy Co.,Ltd. Pledgor: WUXI TVMINING MEDIA SCIENCE & TECHNOLOGY Co.,Ltd. Registration number: Y2020990001517 |
|
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190913 Termination date: 20210723 |