CN1719909A - Method for measuring audio-video frequency content change - Google Patents
Method for measuring audio-video frequency content change Download PDFInfo
- Publication number
- CN1719909A CN1719909A CN 200510027786 CN200510027786A CN1719909A CN 1719909 A CN1719909 A CN 1719909A CN 200510027786 CN200510027786 CN 200510027786 CN 200510027786 A CN200510027786 A CN 200510027786A CN 1719909 A CN1719909 A CN 1719909A
- Authority
- CN
- China
- Prior art keywords
- program
- feature
- receiving terminal
- video
- audio
- 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.)
- Granted
Links
Abstract
This invention relate to a method for testing content variance of audio and video programs, which designs a test server at the playing end and designs a character computing unit and a character sending unit at the program receiving end, in which, the character computing unit extracts the character streams from the current received programs automatically, the character sending unit sends the character streams to the test server via a backward transmission channel, at the same time, the test server extracts the character streams automatically from all the programs at the playing end to form a character stream database of programs. After receiving the character streams fed back by a receive end of users, the test server compares and matches them with those in the database to finish the monitor and statistics to the content variance of watching programs at a receiving end.
Description
Technical field
The invention belongs to the audio frequency and video technical field, be specifically related to a kind of method that audio-video frequency content changes of measuring.The change-detection that can be used for audio-video frequency content under the application scenario such as radio and television, accurately investigation such as user watched and audience rating.
Application background
Broadcasting, TV and website etc. have become the most general media format, are the informed sources of major event, and society is had profound influence.Along with Digital Television (DTV) is universal, the continuous popularization [1] that Web broadcast, Web TV (IPTV) are used, significant for the real-time monitoring that audio-video frequency contents such as broadcasting and TV programme change with measurement.Two kinds of application the most representative are exactly the investigation of rating (listening) rate and prevent content tampering monitoring etc.
Audience ratings (audience rating) has the final and decisive juncture meaning for a TV (broadcasting) medium.In fact the competition of TV media is exactly the competition of television program quality, and its form of expression is exactly the height of audience ratings.Audience rating data derives from investigation.Present traditional audience rating investigation method has three kinds: telephone poll method, diary investigation method and measuring instrument investigation method.Different investigation methods has the different qualities of data and cost.In addition, in the collection of data, the size of sample size exerts an influence to its accuracy.Sample size is the investigation precision height greatly certainly, but the expense of corresponding input is also high.When sample size was increased to a certain degree, the increase rate of its precision reduced.In view of the foregoing, media research company determines generally that in provincial capital the sample family is 300 familys, is about 900 people mostly.In addition, in order to guarantee the representativeness of sample, need rotate sample.Can see that traditional audience rating investigation method all is to watch situation by the sample of users record television, gathers and finishes.So just caused the uneven of the increase of cost and the quality of data.Sample simultaneously user's management and also further increased the cost expense by turns.
In the transmission course of broadcasting and TV, may have some program and be maliciously tampered, disseminate illegal or bad media information to the audient, for example " Falun Gong " jammer satellite TV is exactly the example of giving prominence to.So, be necessary each receiving terminal that audio/video program arrived of transmission is monitored in real time, so that control measure are illegally distorted and taked in discovery in time.Present anti-tamper monitoring all is artificial, and promptly the professional observes each TV programme.Along with the appearance of the continuous increase, particularly Web TV (IP TV) of broadcasting and the transmission form of TV programme and number of programs, personal monitoring's cost expense is huge, or even infeasible.
Automatically monitor if can change the audio-video frequency content of receiving terminal program, audience rating investigating and anti-tamper monitoring etc. just can be analyzed in monitoring automatically so.The automatic analysis of audio-video frequency content in conjunction with audio/video information retrieval technique [2,3], is carried out extraction, transmission and the coupling of content characteristic for the audio frequency and video of program, has both significantly reduced feedback information volume, is highly resistant to again because the quality that transmission problem causes descends.
For the structured analysis of video, at first be that unit is cut apart with the camera lens.The camera lens of video sequence is cut apart and is also claimed the shot change detection, is one of key technology in the video frequency searching [4,5,6,7].Shot change is meant scene content change in the video sequence.The research of camera lens dividing method has had for many years, the present common method cut apart of camera lens comprises gray scale split plot design [8], edge split plot design [9], color histogram split plot design [10], compression domain video stream dividing method [11], piece coupling camera lens dividing method, statistical decision camera lens dividing method [2], based on the camera lens dividing method [12] of cluster, detection [13] of gradual shot or the like.Wherein, the computation complexity that the compression domain camera lens is cut apart is low, can finely satisfy in real time and use.Aspect audio segmentation, then simple relatively, can cut apart [14] to audio frequency by detecting long quiet fragment.
At receiving terminal, need send the programs feature stream information to the measurement server of transmitting terminal.Server just can be finished automatic monitoring and statistics that each receiving terminal programme content changes by comparing the programs feature of each receiving terminal and the feature of broadcast program fast.Traditional broadcasting and TV, it all is unidirectional passive reception, need to increase extra reverse transfer channel, could measure the characteristic information of server transmission program receiving (under present condition to transmitting terminal, it is very easy increasing backward channel, for example can set up backward channel by communications such as GSM or CDMA).For Digital Television, Web broadcast or Web TV, its transfer channel itself is exactly two-way, so the reverse transfer channel is ready-made available.
List of references
1.Waterman D.“the economics of internet TV：new niches vs mass audiences”，Info-Thejournal of policy，regulation and strategy for telecommunications，June 2001，vol.3，no.3，pp.215-229(15)
2.D.Narasimhalu.“Multimedia databases”Multimedia Systems 4(5)，1996
3.F Idris，S Panchanathan.“Review of image and video indexing techniques”Journal of VisualCommunication and Image Representation，1997
4.A hanger G，Little TDC.“A survey of technologies for parsing and Indexing Digital Video”Journal of Visual Communication and Image Representation，1996
5.Uillas Gargi，Rangacha Kasturi，Susan H Stryer.“Performance Characterization ofVideo-Shot-Change Detection Methods”IEEE trans on CSVT，2000
6.Boreczky JS，Rowe L.“Comparison of Video shot boundary detection techniques”Proceedings of Visual Communication and Image Representation，1996
7.Lienhart R.“Comparison of automatic shot boundary detection algorithms”Proceedings ofIS&T/SPEI Storage and Retrieval for Image and Video Databases VII，1999
8.Kikukawa T，Kawafuchi S.“Development of an automatic summary editing system for theaudio visual resources”Transactions of the Institute of Electronics，Information andCommunication Engineers，1992
9.Mai K，Miller J，Zabih R.“A robust method for detecting cuts and dissolves in videosequences”Proceedings of ACM Multimedia 95，San Francisco，1995
10.A Nagasaka，Y Tanaka.“Automatic video indexing and full-video search for objectappearances”Proc IFIP 2nd Working Conf Visual database Systems，1992
11.H J Zhang.“Video parsing using compressed data”Proc SPIE Symp Electronic ImagingScience and technology：Image and Video Processing II，1994
12.MR Naphade，R Mehrotra，AM Ferman.“A High-performance Shot boundary DetectionAlgorithm Using Multiple Cues”Proc of Int Conf on Image Processing，1998
13.Zhang H，Kankanhalli A，Smoliar SW.“Automatic partitioning of full motion videoMultimedia System”，1993
14.L Lu，HJ Zhang，H Jiang.“Content Analysis for Audio Classification and Segmentation”IEEE Transactions on Speech and Audio Processing，2002
Summary of the invention
The object of the present invention is to provide a kind of audio/video program characteristic information that utilizes to measure and the method that detects the audio/video program content change.
The measurement audio-video frequency content changing method that the present invention proposes, its basic technical scheme is: (1) receiving terminal is when receiving audio/video program (as broadcast listening, teleview), extract current characterization information stream of watching program: (2) utilize reverse transfer channel (if there is no backward channel at receiving terminal, also can set up the reverse transfer channel by channels such as landline telephone, mobile phones), real-time or non real-time to the characterization information stream of the measurement server feedback institute program receiving of program transmitting terminal; (3) the measurement server of transmitting terminal will extract the characterization information stream of program equally, and the characterization information stream of receiving terminal feedback bundle is mated, thereby realizes measurement or the monitoring that programme content changes.
The present invention is on the basis that content change is measured or monitored, also can listen to or audience rating investigating, anti-programme content are distorted automatic detection, the mass measurement of receiving terminal audio frequency and video etc., need not to investigate as traditional rating (listening) rate again, to listen to or viewership at the program receiving end manual record, thereby increased automaticity greatly, reduced cost of labor.In addition, distort detection automatically and reduced cost of labor too, also can realize instant alarming simultaneously.
Content of the present invention comprises: (1) proposes the feedback frame model of real-time (or non real-time) monitoring of (broadcasting, TV etc.) audio/video program content change, i.e. concrete function, effect and the detailed collaboration process of TV station's program transmitting terminals such as (radio station), measurement server and program receiving end; (2) extraction of the programs feature stream of measurement server is promptly carried out feature stream to all programs of playing and is extracted, and the feature stream that is used for watching with some receiving terminals mates; (3) at program receiving end equipment or module are arranged, it carries out the feature stream extraction to the current program of watching, and sends out by the reverse transfer channel simultaneously and is fed to the measurement server; (4) measure server and carry out the feature stream coupling for each receiving terminal, promptly when server is received the feature stream that receiving terminal sends, compare with the feature stream of broadcast program, determine the programme content that receiving terminal is watched, finish audience rating investigating and distorted program detection etc.
The feedback frame model that the present invention proposes is seen shown in Figure 1.The system that provides in this framework forms module, and key modules has: feature calculation unit, feature retransmission unit, feature receiving element, program matching unit etc.
At the transmitting terminal of (radio station or TV station's side) audio/video program, server is measured in configuration one, is used to support the measurement to the program of large-scale consumer, measures server and can be made up of multiple servers, carries out parallel processing.Measure and to be provided with feature calculation unit, program matching unit, feature receiving element, feature stream buffer area and output unit as a result in the server.Measure server and at first handle all broadcast programs of transmitting terminal, feature calculation unit is extracted the feature stream of each road program automatically; The measurement server will propose the receiving terminal service of measurement requirement for each, therefore measure the characteristic information of each receiving terminal feedback bundle of feature receiving element automatic reception in the server, carry out matching ratio by the program matching unit according to relative time information and characteristic information etc. in the feature stream then, thereby the acquisition measurement result, for example the user watches whether situation, the programme content of program are distorted etc.Measurement result is by output unit output as a result.Between feature calculation unit and program matching unit, the feature stream buffer area is set, is used for temporary characteristic information.
The receiving terminal of audio/video program (user side) feature calculation unit of configuration and a feature retransmission unit, feature calculation unit is carried out feature extraction to the audio/video program that the user watches, to obtain characterization information stream, because the user changes programme channel or frequency of radio station etc., so characterization information stream may be that different program characteristic of correspondence streams by the different periods are formed by connecting; The feature retransmission unit will be extracted by feature calculation unit and obtain characterization information stream sends to transmitting terminal by the reverse transfer channel measurement server.
From the said frame model as can be seen, receiving terminal must possess the reverse transfer channel, and communicates with the measurement server.The reverse transfer channel all exists soon entering on a large scale in the Digital Television and Web TV of application.For traditional television reception mode, can solve by increasing extra reverse transfer channel equally, such as utilizing ready-made telephone wire, Internet network, even can be wireless network etc.
Among the present invention, feature calculation unit is carried out feature extraction to playing (at transmitting terminal) and receiving (at receiving terminal) audio/video program.For video frequency program, be that base unit is cut apart with the video lens, to choose K (1≤K≤3) width of cloth key frame in each camera lens, from key frame, extract features such as dominant hue, histogram or texture, flow as video features.For audio program,, audio stream is divided into by the quiet and non-quiet segment of forming earlier to its quiet detection; To non-quiet segment, calculate its audio frequency characteristics then.Usually select zero-crossing rate, short-time energy, tone, frequency spectrum barycenter or migration etc. one or more, as audio frequency characteristics.Concrete which kind of feature of selecting is determined according to practical application.
Description of drawings
Fig. 1 is system framework model diagram of the present invention.
Number in the figure: 1 is transmitting terminal, and 2 are the program transmission channel, and 3 is receiving terminal, and 4 is the reverse transfer channel, 5 for measuring server, and 6 is feature calculation unit, and 7 is the program matching unit, and 8 is the feature receiving element, 9 is the feature stream buffer memory, and 10 is output unit as a result, and 11 is the feature transmitter unit.
Embodiment
Explanation according to the system framework model of front further describes embodiment below.
1, feature calculation unit
At the audio/video program transmitting terminal, measure server for the N road TV programme that TV station plays, need calculate the feature stream of N road TV programme respectively, and be stored in the measurement server.Feature stream is lacking beginning and end in theory, therefore generally can only preserve a period of time the program and the feature thereof of (for example a couple of days).Need extract the feature of institute's program receiving equally at receiving terminal, but as long as handle 1 road program of current reception.
When extracting video stream characteristics, be that base unit is cut apart and extracted with the video lens.So-called video lens is meant that camera takes the video-frequency band of gained from start to end.On camera lens, carry out feature extraction, both reduced the characteristic information amount greatly, considered the content of video itself again, avoided even time-domain sampling may cause feature to repeat.
The basic principle that extracts video stream characteristics is: feature should characterize video content, and data volume must lack as far as possible, resists the quality that error of transmission causes simultaneously again and descends.Adopt lens length can average out preferably as feature.The length of general camera lens is about the hundreds of frame, just can represent with two bytes, and is very low.Lens length comprises time stab information again simultaneously, helps the program coupling.
To measure and statistical accuracy in order improving, to carry out after camera lens cuts apart, each camera lens is chosen K width of cloth key frame (1≤K≤3 K=1) generally speaking usually, key frame is the representative of camera lens, then each key frame is carried out feature extraction.We can extract features such as dominant hue, color histogram or texture from key frame, the data volume of these features is also few, can select according to practical application, can not cause increasing considerably of characteristic amount.
Suppose (TV station) transmitting terminal broadcast N road program, the buffering area queue size is a S camera lens, and then the length of each camera lens is L
s nN=1,2 ..., N, s=1,2 ..., S} chooses K key frame to each camera lens, and then every frame is characterized as F
Sk nN=1,2 ..., N, s=1,2 ..., S, k=1,2 ..., K}.So just can generate the feature stream of N road program, can be expressed as:
Here subscript 1,2...N represent different programs, subscript 1,2 ... S represents the video lens order of a program.
In order to carry out the program coupling, the time started that needs to write down each camera lens simultaneously is because the frame per second of each program is certain, so only need write down the zero-time of each first camera lens of program and the frame per second of this program.Here (n is a program number for n, k) expression, and k is the camera lens serial number to use function t.
At the receiving terminal of TV programme, in reception and broadcast program, feature calculation unit is carried out the feature extraction of above-mentioned steps equally, and to measuring the server feedback feature.It should be noted that at program receiving end the user may be switched program channel at any time, so add the synchronous code that program switches at switching point.
Suppose that it is # that program switches synchronous code, camera lens is numbered S ', and each lens length is L
S', then receiving terminal is transmitted to the feature stream of measuring server and is: ... #L
1F
1L
2F
2... L
S' F
S' #....In the practical application, the number (such as 0) that synchronous code can adopt lens length can not get, lens length adopts the integer of fixed length, and characteristic vector adopts the decimal of the fixed length of fixed number.Directly send by feedback channel like this and get final product.
For audio stream, earlier it is carried out silence detection, audio stream can be divided into by the quiet and non-quiet fragment of forming, to non-quiet fragment, calculate its audio frequency characteristics then, for example features such as zero-crossing rate, short-time energy, tone, frequency spectrum barycenter or migration.Concrete which kind of feature of selecting will depend on practical application.Very similar with the expression of video features stream, audio stream is expressed as characterization information stream, and takes very similar step, to the characterization information stream of measuring the server feedback audio frequency.
2, programme content coupling
The measurement server of transmitting terminal receives the feature stream that a certain receiving terminal sends, and is resolved to program and switches synchronous code, carries out the program coupling by the characteristic matching unit.
Matching process is to find camera lens to switch the program number at some camera lenses place the most similar after the synchronous code earlier, revises the current program number of watching of this receiving terminal then.The first camera lens that causes when considering the program switching is imperfect.So coupling should be switched second camera lens after the synchronous code from camera lens.During the coupling program, need to revise the time stamp of this receiving terminal simultaneously, promptly time stamp is increased progressively the length of the camera lens that matches.
Program coupling step is: suppose the number of P for the coupling camera lens, (n k) is the timestamp function of getting the station synchronization camera lens to t, t
k' be the timestamp that receives the lens features vector in the feature stream, T
DelayBe the maximum delay time of receiving terminal feedback, T is the timestamp information that current this receiving terminal has mated, and D is an Euclidean distance, and α, β are weights.Then can determine the program number n that mates with following formula.
N is constant before switching synchronous code at next program.T receives that whenever a lens length feature all will increase accordingly.If a certain section feature stream of watching program is #L
1F
1L
2F
2... L
S' F
S' #, time stamp before are T, then pass through this section program after, time stamp increases to T+L
1+ L
2+ ...+L
S'
Consider that certain receiving terminal begins when measured, T does not have initialization, does not consider for the constraint of T during the program coupling.
If do not receive the feedback of certain receiving terminal for a long time, then this receiving terminal should be set to and be measured state.
Complete matching process is expressed as follows with the false code form:
(1) receive the feature stream that certain receiving terminal sends, if this receiving-end state is not measure, T=0 then if current what receive is synchronous code, then changes 2), otherwise change 3);
(2) configuration state is in the coupling, and i=0 changes 1);
(3) if the state of this receiving terminal is in measuring, T=T+L changes 1).If the state of this receiving terminal is in the coupling, i=i+1 is if i==p then changes 4), otherwise change 1);
(4) if T==0, search satisfy t (n, k+p) 〉=t
K+p'-T
DelayN, k; Otherwise search satisfied
N, k.For all possible n, k, calculate
Get and make d n, k hour, T=t (n k+p), puts this receiving-end state in measuring, and changes 1);
3, programme content is distorted detection
Distorted the program detection and found that promptly the program that some receiving terminal receives is not one of program of TV station.Distort and be divided into two classes, a class is for whole the replacing of road program in the multi-channel program, and another kind of doing for the video pictures of one road program wherein transmitted after sprite inserts editor again.
For the first kind:, then report to the police if the lens features distance is higher than certain threshold T h continuously:
ifα|(L
n K+i，L
K′+1)|+βD(F
n K+i，F
K′+i)＞Th，i＝1，2，...，Q
For second class: need to introduce a monitoring algorithm, the characteristic distance that promptly detects certain piece in the piecemeal whether is very big, and the characteristic distance of other a part of piece is very little.
4, rating (listening) rate investigation method
After the program coupling, can obtain the information of the arbitrary moment program receiving of certain receiving terminal.Judge whether certain receiving terminal is watched that a certain program can adopt and searched following program and switch T
0 p, T
1 pThe initial time stamp and the concluding time that are program p are stabbed T
k, T
K+1Be the time stamp that continuous two programs switch, t be the threshold value that program begins and ending place allows not watch (on the occasion of) T
0 p-T
k＞-t, T
K+1-T
1 p＞-t.
Based on audience rating investigating sampling principle, utilize communication channel to specify the feedback receiving terminal of some, in order to generate the result of audience ratings sample investigation.Equally also can using by turns, strategy improves the quality of data of sample investigation.
Simulated experiment
In order to verify the feasibility of the inventive method, carried out simulated experiment.TV programme transmission and feature are transmitted and are all carried out in local area network (LAN), and transmission and backward channel all are IP network, and be consistent with common Internet network.Transmitting terminal (TV station) and measurement server (comprising feature calculation unit, program matching unit and feature receiving element) all are disposed in the same station server, and receiving terminal (comprising feature calculation and retransmission unit) is disposed among the interior PC of same local area network (LAN).
Transmitting terminal simultaneously sends 10 tunnel different TV programme to network and measurement server, simulates 10 different channels, every section program duration 30 minutes.All programs all are the true TV programme of recording from satellite from CCTV, ABC etc., and program is also contained all kinds such as news, physical culture, TV play, MTV.
At first verify the feasibility of program matching algorithm, coupling be characterized as lens length.Receiving terminal is collected 30 minutes programs, and specifies and collect the number of times (0～3 time) that switches in the process, and different switching times is one group of experiment, and every group of experiment carried out 5 times.Switching instant and channel are all generated at random by program.Experimental result is listed as follows:
| Switching times | On average be matched to power |
| 0 | 100％ |
| 1 | 100％ |
| 2 | 93％ |
| 3 | 90％ |
As can be seen from the table, matching algorithm can extraordinaryly carry out the program coupling.Because switching times increases, and can cause mistake once in a while.We are by analyze finding, mistake is because too frequent between twice switching, has only tens seconds even a few second, and the length of a camera lens of less than also is so matching algorithm can't be determined the channel that switched.And, consider in the practical application that this switching is the quick switching of user when channel selection in fact, is not the information of required monitoring and record.
Then carry out being distorted the test experience of program.At extra 1 road program that sends of transmitting terminal, this road program does not send to measuring server, so this road program can be regarded as the program of being distorted by measured server.Same, receiving terminal is collected 30 minutes programs, and specifies and collect the number of times (0～3 time) that switches in the process, and different switching times is one group of experiment, and every group of experiment carried out 5 times.Different is that having the channel that once switches here at least is the extra program that sends.Experimental result is listed as follows:
| Switching times | Programme content is distorted warning |
| 0 | 100％ |
| 1 | 100％ |
| 2 | 100％ |
| 3 | 100％ |
As can be seen, each receiving terminal is collected extra program from experiment, measures server and can both in time find.So whether according to algorithm of the present invention, it is very effective being distorted for the detection programme content.
Claims (4)
1, a kind of being used to measured the method that audio-video frequency content changes, and it is characterized in that designing the feedback frame model of the real-time or non real-time monitoring of audio/video program content change, wherein:
At the transmitting terminal of audio/video program, server is measured in configuration one, is used to support the measurement to the program of large-scale consumer, measures server and is made up of multiple servers, carries out parallel processing; Measure and to be provided with feature calculation unit, program matching unit, feature receiving element, feature stream buffer area and output unit as a result in the server; Measure server and at first handle all broadcast programs of transmitting terminal, feature calculation unit is extracted the feature stream of each road program automatically; Measure server and propose the receiving terminal service of measurement requirement for each, each receiving terminal of feature receiving element automatic reception feeds back characteristic information in the measurement server, carry out matching ratio by the program matching unit according to relative time information in the feature stream and characteristic information then, obtain measurement result; Measurement result is by output unit output as a result; Between feature calculation unit and the program matching unit feature stream buffer area is set, is used for temporary characteristic information;
At a receiving terminal feature calculation unit of configuration and a feature retransmission unit of audio/video program, feature calculation unit is carried out feature extraction to the audio/video program that the user watches, to obtain characterization information stream; The feature retransmission unit will be extracted by feature calculation unit and obtain characterization information stream sends to transmitting terminal by the reverse transfer channel measurement server.
2, method according to claim 1 is characterized in that said feature calculation unit is play transmitting terminal and the audio/video program of receiving terminal reception carries out feature extraction, wherein:
For video frequency program, be that base unit is cut apart with the video lens, from each camera lens, choose 1-3 width of cloth key frame again, from key frame, extract dominant hue, color histogram or textural characteristics, flow as video features;
For audio program, detect sound is quiet earlier, be by the quiet and non-quiet segment of forming with audio segmentation; To non-quiet segment, calculate its audio frequency characteristics then, audio frequency characteristics is selected one or more of zero-crossing rate, short-time energy, tone, frequency spectrum barycenter or migration for use.
3, method according to claim 2 is characterized in that for video frequency program, and the step that feature calculation unit is extracted programs feature is as follows: suppose a transmitting terminal broadcast N road program, the buffering area queue size is a S camera lens, and then the length of each camera lens is L
s nN=1,2 ..., N, s=1,2 ..., S} chooses K key frame to each camera lens, and then every frame is characterized as F
Sk nN=1,2 ..., N, s=1,2 ..., S, k=1,2 ..., K}; So just generate the feature stream of N road program, be expressed as:
Here subscript 1,2 ... N represents different programs, subscript 1,2 ... S represents the video lens order of a program;
Write down the zero-time of each first camera lens of program and the frame per second of this program simultaneously, (n is a program number for n, k) expression, and k is the camera lens serial number with function t;
At program receiving end, feature calculation unit is carried out the feature extraction of above-mentioned steps equally, and adds the synchronous code that program switches at switching point when switching program channel:
Suppose that it is # that program switches synchronous code, camera lens is numbered S ', and each lens length is L
S ', then receiving terminal is transmitted to the feature stream of measuring server and is: ... #L
1F
1L
2F
2... L
S 'F
S '#....
4, method according to claim 3 is characterized in that the measurement server of transmitting terminal receives the feature stream that a certain receiving terminal sends, and is resolved to program and switches synchronous code, carries out the program coupling by the characteristic matching unit; The process of program coupling is to find camera lens to switch the program number at some camera lenses place the most similar after the synchronous code earlier, revises the current program number of watching of this receiving terminal then; Revise the time stamp of this receiving terminal simultaneously, time stamp is increased progressively the length that matches camera lens, the false code form of its matching process is as follows:
(1) receive the feature stream that certain receiving terminal sends, if this receiving-end state is not measure, T=0 then if current what receive is synchronous code, then changes (2), otherwise changes (3);
(2) configuration state is in the coupling, and i=0 changes (1);
(3) if the state of this receiving terminal is in measuring, T=T+L changes (1).If the state of this receiving terminal is in the coupling, i=i+1 is if i=p then changes 4), otherwise change (1);
(4) if T=0, search satisfy t (n, k+p) 〉=t
K+p'-T
DelayN, k; Otherwise search satisfy t (n, k+1) 〉=T, t (n, k+p) 〉=t
K+p'-T
DelayN, k.For all possible n, k, calculate
Get and make d n, k hour, (n k+p), puts this receiving-end state in measuring to T=t, changes (1);
Wherein, suppose the number of P for the coupling camera lens, (n k) is the timestamp function of getting the station synchronization camera lens to t, t
k' be the timestamp that receives the lens features vector in the feature stream, T
DelayBe the maximum delay time of receiving terminal feedback, T is the timestamp information that current this receiving terminal has mated, and D is an Euclidean distance, and α, β are weights; N is for determining the program number of coupling.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100277867A CN100477809C (en) | 2005-07-15 | 2005-07-15 | Method for measuring audio-video frequency content change |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100277867A CN100477809C (en) | 2005-07-15 | 2005-07-15 | Method for measuring audio-video frequency content change |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1719909A true CN1719909A (en) | 2006-01-11 |
| CN100477809C CN100477809C (en) | 2009-04-08 |
Family
ID=35931602
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100277867A Expired - Fee Related CN100477809C (en) | 2005-07-15 | 2005-07-15 | Method for measuring audio-video frequency content change |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100477809C (en) |
Cited By (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100531279C (en) * | 2006-09-04 | 2009-08-19 | 中兴通讯股份有限公司 | A method to determine the number of subscriber of mobile multimedia broadcast by quasi-real-time statistic |
| WO2009143668A1 (en) * | 2008-05-26 | 2009-12-03 | Yuvad Technologies Co., Ltd. | A method for automatically monitoring viewing activities of television signals |
| CN101771687A (en) * | 2009-12-29 | 2010-07-07 | 北京牡丹电子集团有限责任公司 | Method for monitoring illegal internet videos and image programs |
| CN101938582A (en) * | 2010-08-26 | 2011-01-05 | 广州杰赛科技股份有限公司 | Digital audio monitor, monitoring method and monitoring system |
| CN101714905B (en) * | 2008-10-08 | 2011-08-10 | 中兴通讯股份有限公司 | Method, device and system for counting program information and user equipment |
| CN101572778B (en) * | 2008-04-28 | 2011-11-30 | 北大方正集团有限公司 | Method and system for broadcasting programs |
| WO2012126406A2 (en) * | 2012-04-24 | 2012-09-27 | 华为技术有限公司 | Method and system for researching viewership |
| CN101365102B (en) * | 2008-10-14 | 2012-12-05 | 北京中星微电子有限公司 | Audience rating statistical method and system based on video content recognition |
| US8370382B2 (en) | 2008-05-21 | 2013-02-05 | Ji Zhang | Method for facilitating the search of video content |
| CN103079054A (en) * | 2012-12-30 | 2013-05-01 | 广东九联科技股份有限公司 | Video signal monitoring system and monitoring method |
| US8437555B2 (en) | 2007-08-27 | 2013-05-07 | Yuvad Technologies, Inc. | Method for identifying motion video content |
| US8488835B2 (en) | 2008-05-21 | 2013-07-16 | Yuvad Technologies Co., Ltd. | System for extracting a fingerprint data from video/audio signals |
| CN103260051A (en) * | 2007-04-23 | 2013-08-21 | 冲电气工业株式会社 | Communication device and communication method |
| US8548192B2 (en) | 2008-05-22 | 2013-10-01 | Yuvad Technologies Co., Ltd. | Method for extracting a fingerprint data from video/audio signals |
| CN103366753A (en) * | 2013-06-28 | 2013-10-23 | 宁波大学 | Moving picture experts group audio layer-3 (MP3) audio double-compression detection method under same code rate |
| US8577077B2 (en) | 2008-05-22 | 2013-11-05 | Yuvad Technologies Co., Ltd. | System for identifying motion video/audio content |
| US8611701B2 (en) | 2008-05-21 | 2013-12-17 | Yuvad Technologies Co., Ltd. | System for facilitating the search of video content |
| CN101512657B (en) * | 2006-08-29 | 2014-03-12 | 艾科星科技公司 | Method and apparatus for receiving, storing, and presenting multimedia programming without indexing prior to storage |
| CN103780924A (en) * | 2012-10-26 | 2014-05-07 | 合肥康东福双信息科技有限公司 | Adaptive intelligent television synchronous broadcasting monitoring system on dynamic delay condition |
| CN104463864A (en) * | 2014-12-05 | 2015-03-25 | 华南师范大学 | Multistage parallel key frame cloud extraction method and system |
| CN105228004A (en) * | 2015-09-07 | 2016-01-06 | 崔俊生 | A kind of digital TV contents monitoring method and digital TV contents monitoring system thereof |
| CN108235137A (en) * | 2018-03-22 | 2018-06-29 | 青岛海信电器股份有限公司 | It is a kind of that method, apparatus and the television set that platform acts are cut by sound waveform judgement |
| CN105611326B (en) * | 2015-12-28 | 2019-01-29 | 上海昌视网络科技有限公司 | Streaming media on demand checks method of calibration |
| WO2021027846A1 (en) * | 2019-08-12 | 2021-02-18 | 南京中兴新软件有限责任公司 | Method, apparatus, and system for detecting video code stream |
| CN116939197A (en) * | 2023-09-15 | 2023-10-24 | 海看网络科技(山东)股份有限公司 | Live program head broadcasting and replay content consistency monitoring method based on audio and video |
-
2005
- 2005-07-15 CN CNB2005100277867A patent/CN100477809C/en not_active Expired - Fee Related
Cited By (35)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101512657B (en) * | 2006-08-29 | 2014-03-12 | 艾科星科技公司 | Method and apparatus for receiving, storing, and presenting multimedia programming without indexing prior to storage |
| CN100531279C (en) * | 2006-09-04 | 2009-08-19 | 中兴通讯股份有限公司 | A method to determine the number of subscriber of mobile multimedia broadcast by quasi-real-time statistic |
| CN103260051A (en) * | 2007-04-23 | 2013-08-21 | 冲电气工业株式会社 | Communication device and communication method |
| US8452043B2 (en) | 2007-08-27 | 2013-05-28 | Yuvad Technologies Co., Ltd. | System for identifying motion video content |
| US8437555B2 (en) | 2007-08-27 | 2013-05-07 | Yuvad Technologies, Inc. | Method for identifying motion video content |
| CN101572778B (en) * | 2008-04-28 | 2011-11-30 | 北大方正集团有限公司 | Method and system for broadcasting programs |
| US8370382B2 (en) | 2008-05-21 | 2013-02-05 | Ji Zhang | Method for facilitating the search of video content |
| US8611701B2 (en) | 2008-05-21 | 2013-12-17 | Yuvad Technologies Co., Ltd. | System for facilitating the search of video content |
| US8488835B2 (en) | 2008-05-21 | 2013-07-16 | Yuvad Technologies Co., Ltd. | System for extracting a fingerprint data from video/audio signals |
| US8548192B2 (en) | 2008-05-22 | 2013-10-01 | Yuvad Technologies Co., Ltd. | Method for extracting a fingerprint data from video/audio signals |
| US8577077B2 (en) | 2008-05-22 | 2013-11-05 | Yuvad Technologies Co., Ltd. | System for identifying motion video/audio content |
| WO2009143668A1 (en) * | 2008-05-26 | 2009-12-03 | Yuvad Technologies Co., Ltd. | A method for automatically monitoring viewing activities of television signals |
| CN101714905B (en) * | 2008-10-08 | 2011-08-10 | 中兴通讯股份有限公司 | Method, device and system for counting program information and user equipment |
| CN101365102B (en) * | 2008-10-14 | 2012-12-05 | 北京中星微电子有限公司 | Audience rating statistical method and system based on video content recognition |
| CN101771687A (en) * | 2009-12-29 | 2010-07-07 | 北京牡丹电子集团有限责任公司 | Method for monitoring illegal internet videos and image programs |
| CN101938582B (en) * | 2010-08-26 | 2013-04-10 | 广州杰赛科技股份有限公司 | Digital audio monitor, monitoring method and monitoring system |
| CN101938582A (en) * | 2010-08-26 | 2011-01-05 | 广州杰赛科技股份有限公司 | Digital audio monitor, monitoring method and monitoring system |
| WO2012126406A3 (en) * | 2012-04-24 | 2013-03-28 | 华为技术有限公司 | Method and system for researching viewership |
| CN102763427A (en) * | 2012-04-24 | 2012-10-31 | 华为技术有限公司 | Method and system for researching viewership |
| WO2012126406A2 (en) * | 2012-04-24 | 2012-09-27 | 华为技术有限公司 | Method and system for researching viewership |
| CN103780924B (en) * | 2012-10-26 | 2017-12-05 | 合肥康东福双信息科技有限公司 | Self-adapting intelligent television synchronous broadcasts monitoring system under the conditions of dynamic deferred |
| CN103780924A (en) * | 2012-10-26 | 2014-05-07 | 合肥康东福双信息科技有限公司 | Adaptive intelligent television synchronous broadcasting monitoring system on dynamic delay condition |
| CN103079054A (en) * | 2012-12-30 | 2013-05-01 | 广东九联科技股份有限公司 | Video signal monitoring system and monitoring method |
| CN103366753B (en) * | 2013-06-28 | 2015-07-22 | 宁波大学 | Moving picture experts group audio layer-3 (MP3) audio double-compression detection method under same code rate |
| CN103366753A (en) * | 2013-06-28 | 2013-10-23 | 宁波大学 | Moving picture experts group audio layer-3 (MP3) audio double-compression detection method under same code rate |
| CN104463864A (en) * | 2014-12-05 | 2015-03-25 | 华南师范大学 | Multistage parallel key frame cloud extraction method and system |
| CN104463864B (en) * | 2014-12-05 | 2018-08-14 | 华南师范大学 | Multistage parallel key frame cloud extracting method and system |
| CN105228004A (en) * | 2015-09-07 | 2016-01-06 | 崔俊生 | A kind of digital TV contents monitoring method and digital TV contents monitoring system thereof |
| CN105228004B (en) * | 2015-09-07 | 2019-04-19 | 崔俊生 | A kind of digital TV contents monitoring method and its digital TV contents monitor system |
| CN105611326B (en) * | 2015-12-28 | 2019-01-29 | 上海昌视网络科技有限公司 | Streaming media on demand checks method of calibration |
| CN108235137A (en) * | 2018-03-22 | 2018-06-29 | 青岛海信电器股份有限公司 | It is a kind of that method, apparatus and the television set that platform acts are cut by sound waveform judgement |
| CN108235137B (en) * | 2018-03-22 | 2021-03-02 | 海信视像科技股份有限公司 | Method and device for judging channel switching action through sound waveform and television |
| WO2021027846A1 (en) * | 2019-08-12 | 2021-02-18 | 南京中兴新软件有限责任公司 | Method, apparatus, and system for detecting video code stream |
| US11729436B2 (en) | 2019-08-12 | 2023-08-15 | Zte Corporation | Method, apparatus, and system for detecting video code stream |
| CN116939197A (en) * | 2023-09-15 | 2023-10-24 | 海看网络科技(山东)股份有限公司 | Live program head broadcasting and replay content consistency monitoring method based on audio and video |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100477809C (en) | 2009-04-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1719909A (en) | Method for measuring audio-video frequency content change | |
| US11575454B2 (en) | Automated data-matching based on fingerprints | |
| US8453170B2 (en) | System and method for monitoring and recognizing broadcast data | |
| CN1998168B (en) | Method and apparatus for identification of broadcast source | |
| US8639178B2 (en) | Broadcast source identification based on matching broadcast signal fingerprints | |
| CN104023251B (en) | Based on interactive approach and the system of video | |
| DE112015005408T5 (en) | Systems and methods for identifying users who have viewed a television advertisement | |
| US20200396014A1 (en) | Cloud callout identification of unknown broadcast signatures based on previously recorded broadcast signatures | |
| CN102799605A (en) | Method and system for monitoring advertisement broadcast | |
| US20040260682A1 (en) | System and method for identifying content and managing information corresponding to objects in a signal | |
| CN1585947A (en) | Method and system for personal information retrieval, update and presentation | |
| WO2014164728A1 (en) | Methods and systems for identifying information of a broadcast station and information of broadcasted content | |
| CN104584571A (en) | Generating a sequence of audio fingerprints at a set top box | |
| WO2015156842A1 (en) | Methods and apparatus to identify media using hash keys | |
| CN102890950B (en) | Media automatic editing device, method, media transmissions method and its broadcasting system | |
| CN104394436A (en) | Audience rating monitoring method and device of network television live channel | |
| CN108540857B (en) | Television audience rating and user preference control method and system based on mobile phone remote controller | |
| CN103023923A (en) | Information transmission method and information transmission device | |
| Bardeli et al. | Audio fingerprinting for media synchronisation and duplicate detection | |
| CN103309865B (en) | Method and system for realizing video source clustering | |
| JP3983532B2 (en) | Scene extraction device | |
| WO2018039060A1 (en) | Systems and methods for sourcing live streams | |
| CN104202628A (en) | Server, client and client program identification system and method | |
| CN115495600A (en) | Video and audio retrieval method based on features | |
| CN115550699A (en) | Live broadcast content pushing method for different users |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20090408 Termination date: 20140715 |
|
| EXPY | Termination of patent right or utility model |