CN106021320A - Video based online real-time video search method - Google Patents
Video based online real-time video search method Download PDFInfo
- Publication number
- CN106021320A CN106021320A CN201610294977.8A CN201610294977A CN106021320A CN 106021320 A CN106021320 A CN 106021320A CN 201610294977 A CN201610294977 A CN 201610294977A CN 106021320 A CN106021320 A CN 106021320A
- Authority
- CN
- China
- Prior art keywords
- video
- elements
- gained
- compression
- binary
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/73—Querying
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a video based online real-time video search method which is used for online real-time video search. The method includes the steps: (1) a server end extracts feature data of video streams: the server end receives a plurality of video streams in real time, extracts videos during recent 5-1000 seconds of each video stream, and extracts video feature data of each video as feature data of the corresponding video stream; (2) the step (1) is executed repeatedly till the server end receives a search request initiated by a client end, and then a step (3) starts, wherein the process that the client end initiates the search request includes that the client end shoots a video, extracts one or more frames of images from the video, calculates out binary feature data of a video frame, and uploads the feature data of the image to the sever end; and (3) a search process is performed, feature data of images acquired from the step (2) and feature data of each video stream obtained from the step (1) are matched, and then the step (1) starts.
Description
Technical field
The present invention relates to the search method of a kind of online real-time video based on video.
Background technology
Recently more than ten years, computer vision technique develops rapidly, as David G. Lowe 2004 exists
" the Distinctive Image Features delivered on International Journal of Computer Vision
From Scale-Invariant Keypoints. " so that the search of image content-based becomes practical, and then make based on
The video search of the content of video becomes possibility.Add last decade, the computing capability of computer hardware equipment improve with
And the decline of hardware cost, and the appearance of various cloud computing platform, such as the E2C of Amazon, the cost of calculating is also greatly reduced,
This makes search based on video content the most gradually become practical.The technical scheme of main flow is used for reference mostly from image retrieval at present
Method, and the standard being correlated with has occurred, specifically can be found in ISO/IEC DIS 15938-13.But current mainstream scheme
All consider retrieves for offline video, and first all videos in video library is indexed by this type of scheme, the general bag of this index
Include two parts: extract video frequency feature data and index characteristic.Wherein content based video retrieval system, its video features number
According to generally not only including local description, also include global description's gathering local description, the amount of calculation both this
The least;Meanwhile, global description's after the index of the characteristic in the scheme of main flow is typically based on binarization sets up MBIT
(Multi-block Index Table), is referred to the test model Test Model 13 of ISO/IEC JTC1/SC29:
Compact Descriptors for Visual Search, but the foundation of index can devote a tremendous amount of time.For off-line
The search of video, time delay can be tolerated, but retrieve for online real-time video, to ensure real-time, in order to tackle
Big amount of calculation, it will the cost that hardware is put into remains high;Application scenario yet with Online Video retrieval is limited,
Such as traffic camera search, station synchronization interaction etc., in these occasions, the quantity of live video stream is not big especially, builds
Lithol draws can't bring the biggest performance boost.Therefore, the most general video retrieval technology is not particularly suited for online real-time
The retrieval of video.
Summary of the invention
It is an object of the invention to overcome the deficiencies in the prior art, it is provided that the inspection of a kind of online real-time video based on video
Suo Fangfa, the method omission Index process, do not use global description's, real-time is effectively ensured, the method calculates simply, is prone to
Realize.
It is an object of the invention to be achieved through the following technical solutions: the retrieval side of online real-time video based on video
Method, it comprises the steps:
S1: server end extracts and includes multiple video flowing in real time, takes the video of nearest 5~1000 seconds in described each video flowing,
Extract the video frequency feature data characteristic as corresponding video flowing of described each video;
S2: circulation performs step S1, until receiving client retrieval request;
S3: client one section of video of shooting, then from this video, take a frame or multiple image, the binary system calculating frame of video is special
Levy, the binary features of image uploaded onto the server end request retrieval according to frame of video sequencing in video;
S4: the binary features of client frame of video is mated with the binary features data of each video flowing, coupling
Value the maximum, if being more than threshold value, then corresponding video flowing is video flowing to be retrieved
S5: output retrieval result.
Described extraction video frequency feature data includes following sub-step:
S11: take the frame in video or multiframe;
S12: extract the SIFT characteristic point of each frame in S11, that chooses in gained SIFT feature point is multiple, selected by calculating
The Even-dimensional SIFT of SIFT feature point describes son;
S13: all for S12 gained Even-dimensional SIFT are described sub-boil down to the binary features of 16~512 binary digits.
The binary features of described calculating frame of video includes following sub-step:
S31: extract SIFT feature point from frame of video, that chooses in gained SIFT feature point is multiple, calculates selected SIFT feature
The SIFT of point describes son;
S32: all for S31 gained SIFT are described sub-boil down to the binary features of 16~512 binary digits.
Record matching value is also wanted while described S4 carries out binary features Data Matching.
The described sub-boil down to that described by Even-dimensional SIFT has the binary features of 16~512 binary digits, specifically side
Method is: the SIFT of N-dimensional describes son can regard N number of number as, to described any combination of two of N number of number, obtains N/2 number pair, this N/2
Number is to arbitrary arrangement, and the every number after arrangement is to corresponding bit, and wherein N is even number, and compresses and will obtain one
Having the binary features of N/2 binary digit, each number, to obtaining a binary features, is specially each number pair
First number and the second number compare, and the first number is more than the second number, then compression obtains the two of corresponding binary digit and enters
System is characterized as 1;First number is less than the second number, then compression obtains the binary features of corresponding binary digit is 0.
The invention has the beneficial effects as follows: the invention provides the search method of a kind of online real-time video based on video,
Owing to without Index process, the most not using global description's, real-time is guaranteed.Present invention further introduces a kind of new
SIFT is described the method that son carries out binaryzation compression, and the method calculates simply, is easily achieved, and the binary system after compressing
Feature, it mates the Hamming distance also used, while ensureing performance, also achieves the high efficiency that low amount of calculation is brought.
Accompanying drawing explanation
Fig. 1 is online real-time video searching method flow chart of the present invention;
Fig. 2 is embodiment system schematic;
Fig. 3 is that 128 elements of 128 dimensions SIFT description are numbered schematic diagram.
Detailed description of the invention
Technical scheme is described in further detail below in conjunction with the accompanying drawings, but protection scope of the present invention is not limited to
The following stated.
Fig. 1 is online real-time video searching method flow chart of the present invention, and the method is implemented by this example, with one or many
Platform computer is server end, and with smart mobile phone as client, server end and client use same SIFT to describe son pressure
Compression method.Fig. 2 is the present embodiment system schematic.
Server end includes 10 video flowings in real time, extracts the characteristic of each video flowing, can implement as follows:
Each video flowing to input, processes the most in accordance with the following methods:
A. the video flowing video of nearest 20 seconds is taken;
B., in the video of 20 seconds of above-mentioned gained, a frame was taken out every 0.5 second;
C. each frame of releasing is handled as follows:
C-1. from frame, extract all SIFT feature points of this frame, be ranked up according to response intensity, choose response intensity
2000 strong SIFT feature points (the most all choosing less than 2000), the SIFT of 128 dimensions calculating them describes son.
C-2. Fig. 3 is that 128 elements that each 128 dimension SIFT describes son are numbered respectively, uses following rule
Each 128 dimension SIFT is described one binary features having 64 binary digits of sub-boil down to and preserves;Below to 128 dimensions
SIFT describes the quoting, all with reference to the numbering of the signal in Fig. 3 of element in son:
If No. 0 element is more than No. 80 elements, compression the 1st binary digit of gained is 1, is otherwise 0;
If No. 1 element is more than No. 81 elements, compression the 2nd binary digit of gained is 1, is otherwise 0;
If No. 2 elements are more than No. 82 elements, compression the 3rd binary digit of gained is 1, is otherwise 0;
If No. 3 elements are more than No. 83 elements, compression the 4th binary digit of gained is 1, is otherwise 0;
If No. 4 elements are more than No. 84 elements, compression the 5th binary digit of gained is 1, is otherwise 0;
If No. 5 elements are more than No. 85 elements, compression the 6th binary digit of gained is 1, is otherwise 0;
If No. 6 elements are more than No. 86 elements, compression the 7th binary digit of gained is 1, is otherwise 0;
If No. 7 elements are more than No. 87 elements, compression the 8th binary digit of gained is 1, is otherwise 0;
If No. 40 elements are more than No. 120 elements, compression the 9th binary digit of gained is 1, is otherwise 0;
If No. 41 elements are more than No. 121 elements, compression the 10th binary digit of gained is 1, is otherwise 0;
If No. 42 elements are more than No. 122 elements, compression the 11st binary digit of gained is 1, is otherwise 0;
If No. 43 elements are more than No. 123 elements, compression the 12nd binary digit of gained is 1, is otherwise 0;
If No. 44 elements are more than No. 124 elements, compression the 13rd binary digit of gained is 1, is otherwise 0;
If No. 45 elements are more than No. 125 elements, compression the 14th binary digit of gained is 1, is otherwise 0;
If No. 46 elements are more than No. 126 elements, compression the 15th binary digit of gained is 1, is otherwise 0;
If No. 47 elements are more than No. 127 elements, compression the 16th binary digit of gained is 1, is otherwise 0;
If No. 24 elements are more than No. 72 elements, compression the 17th binary digit of gained is 1, is otherwise 0;
If No. 25 elements are more than No. 73 elements, compression the 18th binary digit of gained is 1, is otherwise 0;
If No. 26 elements are more than No. 74 elements, compression the 19th binary digit of gained is 1, is otherwise 0;
If No. 27 elements are more than No. 75 elements, compression the 20th binary digit of gained is 1, is otherwise 0;
If No. 28 elements are more than No. 76 elements, compression the 21st binary digit of gained is 1, is otherwise 0;
If No. 29 elements are more than No. 77 elements, compression the 22nd binary digit of gained is 1, is otherwise 0;
If No. 30 elements are more than No. 78 elements, compression the 23rd binary digit of gained is 1, is otherwise 0;
If No. 31 elements are more than No. 79 elements, compression the 24th binary digit of gained is 1, is otherwise 0;
If No. 48 elements are more than No. 96 elements, compression the 25th binary digit of gained is 1, is otherwise 0;
If No. 49 elements are more than No. 97 elements, compression the 26th binary digit of gained is 1, is otherwise 0;
If No. 50 elements are more than No. 98 elements, compression the 27th binary digit of gained is 1, is otherwise 0;
If No. 51 elements are more than No. 99 elements, compression the 28th binary digit of gained is 1, is otherwise 0;
If No. 52 elements are more than No. 100 elements, compression the 29th binary digit of gained is 1, is otherwise 0;
If No. 53 elements are more than No. 101 elements, compression the 30th binary digit of gained is 1, is otherwise 0;
If No. 54 elements are more than No. 102 elements, compression the 31st binary digit of gained is 1, is otherwise 0;
If No. 55 elements are more than No. 103 elements, compression the 32nd binary digit of gained is 1, is otherwise 0;
If No. 8 elements are more than ununbium, compression the 33rd binary digit of gained is 1, is otherwise 0;
If No. 9 elements are more than ununtrium, compression the 34th binary digit of gained is 1, is otherwise 0;
If No. 10 elements are more than ununquadium, compression the 35th binary digit of gained is 1, is otherwise 0;
If No. 11 elements are more than ununpentium, compression the 36th binary digit of gained is 1, is otherwise 0;
If No. 12 elements are more than ununhexium, compression the 37th binary digit of gained is 1, is otherwise 0;
If No. 13 elements are more than ununseptium, compression the 38th binary digit of gained is 1, is otherwise 0;
If No. 14 elements are more than ununoctium, compression the 39th binary digit of gained is 1, is otherwise 0;
If No. 15 elements are more than ununennium, compression the 40th binary digit of gained is 1, is otherwise 0;
If No. 16 elements are more than No. 104 elements, compression the 41st binary digit of gained is 1, is otherwise 0;
If No. 17 elements are more than No. 105 elements, compression the 42nd binary digit of gained is 1, is otherwise 0;
If No. 18 elements are more than No. 106 elements, compression the 43rd binary digit of gained is 1, is otherwise 0;
If No. 19 elements are more than No. 107 elements, compression the 44th binary digit of gained is 1, is otherwise 0;
If No. 20 elements are more than No. 108 elements, compression the 45th binary digit of gained is 1, is otherwise 0;
If No. 21 elements are more than No. 109 elements, compression the 46th binary digit of gained is 1, is otherwise 0;
If No. 22 elements are more than ununnilium, compression the 47th binary digit of gained is 1, is otherwise 0;
If No. 23 elements are more than unununium, compression the 48th binary digit of gained is 1, is otherwise 0;
If No. 32 elements are more than No. 88 elements, compression the 49th binary digit of gained is 1, is otherwise 0;
If No. 33 elements are more than No. 89 elements, compression the 50th binary digit of gained is 1, is otherwise 0;
If No. 34 elements are more than No. 90 elements, compression the 51st binary digit of gained is 1, is otherwise 0;
If No. 35 elements are more than No. 91 elements, compression the 52nd binary digit of gained is 1, is otherwise 0;
If No. 36 elements are more than No. 92 elements, compression the 53rd binary digit of gained is 1, is otherwise 0;
If No. 37 elements are more than No. 93 elements, compression the 54th binary digit of gained is 1, is otherwise 0;
If No. 38 elements are more than No. 94 elements, compression the 55th binary digit of gained is 1, is otherwise 0;
If No. 39 elements are more than No. 95 elements, compression the 56th binary digit of gained is 1, is otherwise 0;
If No. 64 elements are more than No. 56 elements, compression the 57th binary digit of gained is 1, is otherwise 0;
If No. 65 elements are more than No. 57 elements, compression the 58th binary digit of gained is 1, is otherwise 0;
If No. 66 elements are more than No. 58 elements, compression the 59th binary digit of gained is 1, is otherwise 0;
If No. 67 elements are more than No. 59 elements, compression the 60th binary digit of gained is 1, is otherwise 0;
If No. 68 elements are more than No. 60 elements, compression the 61st binary digit of gained is 1, is otherwise 0;
If No. 69 elements are more than No. 61 elements, compression the 62nd binary digit of gained is 1, is otherwise 0;
If No. 70 elements are more than No. 62 elements, compression the 63rd binary digit of gained is 1, is otherwise 0;
If No. 71 elements are more than No. 63 elements, compression the 64th binary digit of gained is 1, is otherwise 0;
D. A above, B, step C are constantly circulated, until receiving the retrieval request that client is sent.After receiving retrieval request, make
Retrieve by following manner:
All binary features that D-1: the client got is sent, the binary system obtained with each each frame flowed respectively
Feature is mated, and uses Hamming distance during coupling, if Hamming distance is less than 4, then it is assumed that two binary features couplings;Statistics
What the binary features that each frame that all binary features that client is sent flow with each obtains matched counts.
D-2: the stream taking that frame that coupling counts most corresponding is the stream mated most;Its match point of most frame if coupling is counted
Number has been more than 20, then retrieve successfully, and the stream mated most is stream to be retrieved, and returns retrieval result and retrieves successful message to objective
Family end;Otherwise retrieve failure, return retrieval failure to client.
D-3: return to step A.
Cell-phone customer terminal can realize as follows:
Cell-phone customer terminal continues recorded video, and in real time it is taken out frame, extracts a frame each second, and to each frame at
Reason, this process will be carried out always, until server returns retrieves successful message, the tool wherein processed each pictures
Body process is as follows:
A. from frame, extract all SIFT feature points of this frame, be ranked up according to response intensity, choose response intensity the strongest
2000 SIFT feature points (the most all choosing less than 2000), calculate they 128 dimension SIFT describe son.B. Fig. 3
Being numbered respectively for each 128 dimension SIFT being described 128 elements of son, using following rule by each 128 dimension
SIFT describes one binary features having 64 binary digits of sub-boil down to and preserves;Hereinafter 128 dimension SIFT are described in son
Quoting of element, all with reference to the numbering of the signal in Fig. 3:
If No. 0 element is more than No. 80 elements, compression the 1st binary digit of gained is 1, is otherwise 0;
If No. 1 element is more than No. 81 elements, compression the 2nd binary digit of gained is 1, is otherwise 0;
If No. 2 elements are more than No. 82 elements, compression the 3rd binary digit of gained is 1, is otherwise 0;
If No. 3 elements are more than No. 83 elements, compression the 4th binary digit of gained is 1, is otherwise 0;
If No. 4 elements are more than No. 84 elements, compression the 5th binary digit of gained is 1, is otherwise 0;
If No. 5 elements are more than No. 85 elements, compression the 6th binary digit of gained is 1, is otherwise 0;
If No. 6 elements are more than No. 86 elements, compression the 7th binary digit of gained is 1, is otherwise 0;
If No. 7 elements are more than No. 87 elements, compression the 8th binary digit of gained is 1, is otherwise 0;
If No. 40 elements are more than No. 120 elements, compression the 9th binary digit of gained is 1, is otherwise 0;
If No. 41 elements are more than No. 121 elements, compression the 10th binary digit of gained is 1, is otherwise 0;
If No. 42 elements are more than No. 122 elements, compression the 11st binary digit of gained is 1, is otherwise 0;
If No. 43 elements are more than No. 123 elements, compression the 12nd binary digit of gained is 1, is otherwise 0;
If No. 44 elements are more than No. 124 elements, compression the 13rd binary digit of gained is 1, is otherwise 0;
If No. 45 elements are more than No. 125 elements, compression the 14th binary digit of gained is 1, is otherwise 0;
If No. 46 elements are more than No. 126 elements, compression the 15th binary digit of gained is 1, is otherwise 0;
If No. 47 elements are more than No. 127 elements, compression the 16th binary digit of gained is 1, is otherwise 0;
If No. 24 elements are more than No. 72 elements, compression the 17th binary digit of gained is 1, is otherwise 0;
If No. 25 elements are more than No. 73 elements, compression the 18th binary digit of gained is 1, is otherwise 0;
If No. 26 elements are more than No. 74 elements, compression the 19th binary digit of gained is 1, is otherwise 0;
If No. 27 elements are more than No. 75 elements, compression the 20th binary digit of gained is 1, is otherwise 0;
If No. 28 elements are more than No. 76 elements, compression the 21st binary digit of gained is 1, is otherwise 0;
If No. 29 elements are more than No. 77 elements, compression the 22nd binary digit of gained is 1, is otherwise 0;
If No. 30 elements are more than No. 78 elements, compression the 23rd binary digit of gained is 1, is otherwise 0;
If No. 31 elements are more than No. 79 elements, compression the 24th binary digit of gained is 1, is otherwise 0;
If No. 48 elements are more than No. 96 elements, compression the 25th binary digit of gained is 1, is otherwise 0;
If No. 49 elements are more than No. 97 elements, compression the 26th binary digit of gained is 1, is otherwise 0;
If No. 50 elements are more than No. 98 elements, compression the 27th binary digit of gained is 1, is otherwise 0;
If No. 51 elements are more than No. 99 elements, compression the 28th binary digit of gained is 1, is otherwise 0;
If No. 52 elements are more than No. 100 elements, compression the 29th binary digit of gained is 1, is otherwise 0;
If No. 53 elements are more than No. 101 elements, compression the 30th binary digit of gained is 1, is otherwise 0;
If No. 54 elements are more than No. 102 elements, compression the 31st binary digit of gained is 1, is otherwise 0;
If No. 55 elements are more than No. 103 elements, compression the 32nd binary digit of gained is 1, is otherwise 0;
If No. 8 elements are more than ununbium, compression the 33rd binary digit of gained is 1, is otherwise 0;
If No. 9 elements are more than ununtrium, compression the 34th binary digit of gained is 1, is otherwise 0;
If No. 10 elements are more than ununquadium, compression the 35th binary digit of gained is 1, is otherwise 0;
If No. 11 elements are more than ununpentium, compression the 36th binary digit of gained is 1, is otherwise 0;
If No. 12 elements are more than ununhexium, compression the 37th binary digit of gained is 1, is otherwise 0;
If No. 13 elements are more than ununseptium, compression the 38th binary digit of gained is 1, is otherwise 0;
If No. 14 elements are more than ununoctium, compression the 39th binary digit of gained is 1, is otherwise 0;
If No. 15 elements are more than ununennium, compression the 40th binary digit of gained is 1, is otherwise 0;
If No. 16 elements are more than No. 104 elements, compression the 41st binary digit of gained is 1, is otherwise 0;
If No. 17 elements are more than No. 105 elements, compression the 42nd binary digit of gained is 1, is otherwise 0;
If No. 18 elements are more than No. 106 elements, compression the 43rd binary digit of gained is 1, is otherwise 0;
If No. 19 elements are more than No. 107 elements, compression the 44th binary digit of gained is 1, is otherwise 0;
If No. 20 elements are more than No. 108 elements, compression the 45th binary digit of gained is 1, is otherwise 0;
If No. 21 elements are more than No. 109 elements, compression the 46th binary digit of gained is 1, is otherwise 0;
If No. 22 elements are more than ununnilium, compression the 47th binary digit of gained is 1, is otherwise 0;
If No. 23 elements are more than unununium, compression the 48th binary digit of gained is 1, is otherwise 0;
If No. 32 elements are more than No. 88 elements, compression the 49th binary digit of gained is 1, is otherwise 0;
If No. 33 elements are more than No. 89 elements, compression the 50th binary digit of gained is 1, is otherwise 0;
If No. 34 elements are more than No. 90 elements, compression the 51st binary digit of gained is 1, is otherwise 0;
If No. 35 elements are more than No. 91 elements, compression the 52nd binary digit of gained is 1, is otherwise 0;
If No. 36 elements are more than No. 92 elements, compression the 53rd binary digit of gained is 1, is otherwise 0;
If No. 37 elements are more than No. 93 elements, compression the 54th binary digit of gained is 1, is otherwise 0;
If No. 38 elements are more than No. 94 elements, compression the 55th binary digit of gained is 1, is otherwise 0;
If No. 39 elements are more than No. 95 elements, compression the 56th binary digit of gained is 1, is otherwise 0;
If No. 64 elements are more than No. 56 elements, compression the 57th binary digit of gained is 1, is otherwise 0;
If No. 65 elements are more than No. 57 elements, compression the 58th binary digit of gained is 1, is otherwise 0;
If No. 66 elements are more than No. 58 elements, compression the 59th binary digit of gained is 1, is otherwise 0;
If No. 67 elements are more than No. 59 elements, compression the 60th binary digit of gained is 1, is otherwise 0;
If No. 68 elements are more than No. 60 elements, compression the 61st binary digit of gained is 1, is otherwise 0;
If No. 69 elements are more than No. 61 elements, compression the 62nd binary digit of gained is 1, is otherwise 0;
If No. 70 elements are more than No. 62 elements, compression the 63rd binary digit of gained is 1, is otherwise 0;
If No. 71 elements are more than No. 63 elements, compression the 64th binary digit of gained is 1, is otherwise 0;
C. the binary features that compression obtains is sent to server end requesting query.
From the description above, relative to general video retrieval method, the method for the present invention is examined for Online Video
This kind of special video frequency searching of rope is optimized.Owing to without Index process, the most not using global description's, real-time obtains
Arrive guarantee.Present invention further introduces a kind of new method that SIFT description is carried out binaryzation compression, the method calculates
Simply, it is easily achieved, and the binary features after compressing, it mates the Hamming distance also used, while ensureing performance,
Also achieve the high efficiency that low amount of calculation is brought.
Claims (5)
1. the search method of online real-time video based on video, it is characterised in that: it comprises the steps:
S1: server end extracts and includes multiple video flowing in real time, takes the video of nearest 5~1000 seconds in described each video flowing,
Extract the video frequency feature data characteristic as corresponding video flowing of described each video;
S2: circulation performs step S1, until receiving client retrieval request;
S3: client one section of video of shooting, then from this video, take a frame or multiple image, the binary system calculating frame of video is special
Levy, the binary features of image uploaded onto the server end request retrieval according to frame of video sequencing in video;
S4: the binary features of client frame of video is mated with the binary features data of each video flowing, coupling
Value the maximum, if being more than threshold value, then corresponding video flowing is video flowing to be retrieved
S5: output retrieval result.
The search method of online real-time video based on video the most according to claim 1, it is characterised in that: described carries
Take video frequency feature data and include following sub-step:
S11: take the frame in video or multiframe;
S12: extract the SIFT characteristic point of each frame in S11, that chooses in gained SIFT feature point is multiple, selected by calculating
The Even-dimensional SIFT of SIFT feature point describes son;
S13: all for S12 gained Even-dimensional SIFT are described sub-boil down to the binary features of 16~512 binary digits.
The search method of online real-time video based on video the most according to claim 1, it is characterised in that: described meter
The binary features calculating frame of video includes following sub-step:
S31: extract SIFT feature point from frame of video, that chooses in gained SIFT feature point is multiple, calculates selected SIFT feature
The SIFT of point describes son;
S32: all for S31 gained SIFT are described sub-boil down to the binary features of 16~512 binary digits.
The search method of online real-time video based on video the most according to claim 1, it is characterised in that: described S4
In carry out binary features Data Matching while also want record matching value.
5. according to the search method of based on video the online real-time video described in Claims 2 or 3, it is characterised in that: described
Even-dimensional SIFT described sub-boil down to have the binary features of 16~512 binary digits, method particularly includes: N-dimensional
SIFT describes son can regard N number of number as, to described any combination of two of N number of number, obtains N/2 number pair, and this N/2 number is to arbitrarily
Arrangement, the every number after arrangement is to corresponding bit, and wherein N is even number, and compresses and have N/2 two to enter by obtaining one
The binary features of position processed, each number to obtaining a binary features, be specially the first number of each number pair and
Second number compares, and the first number is more than the second number, then compression obtains the binary features of corresponding binary digit is 1;
First number is less than the second number, then compression obtains the binary features of corresponding binary digit is 0.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610294977.8A CN106021320B (en) | 2016-05-06 | 2016-05-06 | The search method of online real-time video based on video |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610294977.8A CN106021320B (en) | 2016-05-06 | 2016-05-06 | The search method of online real-time video based on video |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106021320A true CN106021320A (en) | 2016-10-12 |
CN106021320B CN106021320B (en) | 2018-06-29 |
Family
ID=57082265
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610294977.8A Active CN106021320B (en) | 2016-05-06 | 2016-05-06 | The search method of online real-time video based on video |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106021320B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108271049A (en) * | 2017-12-01 | 2018-07-10 | 优视科技有限公司 | The method, apparatus and computer equipment of resource downloading |
CN113449143A (en) * | 2020-03-25 | 2021-09-28 | 海信视像科技股份有限公司 | Video content retrieval method, device, equipment and storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102117313A (en) * | 2010-12-29 | 2011-07-06 | 天脉聚源(北京)传媒科技有限公司 | Video retrieval method and system |
CN102760169A (en) * | 2012-06-13 | 2012-10-31 | 天脉聚源(北京)传媒科技有限公司 | Method for detecting advertising slots in television direct transmission streams |
CN103390040A (en) * | 2013-07-17 | 2013-11-13 | 南京邮电大学 | Video copy detection method |
CN103514293A (en) * | 2013-10-09 | 2014-01-15 | 北京中科模识科技有限公司 | Method for video matching in video template library |
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 |
-
2016
- 2016-05-06 CN CN201610294977.8A patent/CN106021320B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102117313A (en) * | 2010-12-29 | 2011-07-06 | 天脉聚源(北京)传媒科技有限公司 | Video retrieval method and system |
CN102760169A (en) * | 2012-06-13 | 2012-10-31 | 天脉聚源(北京)传媒科技有限公司 | Method for detecting advertising slots in television direct transmission streams |
CN103390040A (en) * | 2013-07-17 | 2013-11-13 | 南京邮电大学 | Video copy detection method |
CN103514293A (en) * | 2013-10-09 | 2014-01-15 | 北京中科模识科技有限公司 | Method for video matching in video template library |
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 |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108271049A (en) * | 2017-12-01 | 2018-07-10 | 优视科技有限公司 | The method, apparatus and computer equipment of resource downloading |
CN108271049B (en) * | 2017-12-01 | 2020-09-04 | 阿里巴巴(中国)有限公司 | Resource downloading method and device and computer equipment |
CN113449143A (en) * | 2020-03-25 | 2021-09-28 | 海信视像科技股份有限公司 | Video content retrieval method, device, equipment and storage medium |
CN113449143B (en) * | 2020-03-25 | 2022-09-30 | 海信视像科技股份有限公司 | Video content retrieval method, device, equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN106021320B (en) | 2018-06-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110134829B (en) | Video positioning method and device, storage medium and electronic device | |
Jeannin et al. | MPEG-7 visual motion descriptors | |
Chary et al. | Feature extraction methods for color image similarity | |
JP2001155169A (en) | Method and system for dividing, classifying and summarizing video image | |
Gharbi et al. | Key frames extraction using graph modularity clustering for efficient video summarization | |
Sony et al. | Video summarization by clustering using euclidean distance | |
TW201907736A (en) | Method and device for generating video summary | |
CN103744974B (en) | Method and device for selecting local interest points | |
US20100246959A1 (en) | Apparatus and method for generating additional information about moving picture content | |
CN106021320A (en) | Video based online real-time video search method | |
CN106021321B (en) | The search method of online real-time video based on picture | |
CN110598622A (en) | Video subtitle positioning method, electronic device, and computer storage medium | |
CN111177436A (en) | Face feature retrieval method, device and equipment | |
Mademlis et al. | Summarization of human activity videos using a salient dictionary | |
CN112529018A (en) | Training method and device for local features of image and storage medium | |
CN107704471A (en) | A kind of information processing method and device and file call method and device | |
CN111598176A (en) | Image matching processing method and device | |
CN104580913A (en) | Picture continuous shooting method and device | |
EP2707836A1 (en) | System and method for compact descriptor for visual search | |
CN110807790A (en) | Image data extraction and compression method for video target trajectory tracking content | |
Abdel-Mottaleb et al. | Multimedia descriptions based on MPEG-7: extraction and applications | |
CN113157962A (en) | Image retrieval method, electronic device, and storage medium | |
CN108764258A (en) | A kind of optimum image collection choosing method being inserted into for group's image | |
Liu et al. | A fast, comprehensive shot boundary determination system | |
CN110188237B (en) | Data aggregation system and method for intelligent manufacturing of events |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |