CN108924586A - A kind of detection method of video frame, device and electronic equipment - Google Patents
A kind of detection method of video frame, device and electronic equipment Download PDFInfo
- Publication number
- CN108924586A CN108924586A CN201810638981.0A CN201810638981A CN108924586A CN 108924586 A CN108924586 A CN 108924586A CN 201810638981 A CN201810638981 A CN 201810638981A CN 108924586 A CN108924586 A CN 108924586A
- Authority
- CN
- China
- Prior art keywords
- detected
- video
- video frame
- frame
- segment
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 135
- 238000000034 method Methods 0.000 claims abstract description 58
- 230000004927 fusion Effects 0.000 claims description 29
- 238000004891 communication Methods 0.000 claims description 19
- 238000004590 computer program Methods 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 12
- 108010001267 Protein Subunits Proteins 0.000 claims description 11
- 230000002123 temporal effect Effects 0.000 claims description 10
- 239000000203 mixture Substances 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 5
- 239000012466 permeate Substances 0.000 claims description 3
- 238000000151 deposition Methods 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000008569 process Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 2
- 238000007500 overflow downdraw method Methods 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000009394 selective breeding Methods 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 239000012634 fragment Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000011068 loading method Methods 0.000 description 1
- 238000002844 melting Methods 0.000 description 1
- 230000008018 melting Effects 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 208000001491 myopia Diseases 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
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]
- H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
- H04N21/23—Processing of content or additional data; Elementary server operations; Server middleware
- H04N21/234—Processing of video elementary streams, e.g. splicing of video streams or manipulating encoded video stream scene graphs
- H04N21/23418—Processing of video elementary streams, e.g. splicing of video streams or manipulating encoded video stream scene graphs involving operations for analysing video streams, e.g. detecting features or characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
-
- 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]
- H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
- H04N21/23—Processing of content or additional data; Elementary server operations; Server middleware
- H04N21/234—Processing of video elementary streams, e.g. splicing of video streams or manipulating encoded video stream scene graphs
- H04N21/23424—Processing of video elementary streams, e.g. splicing of video streams or manipulating encoded video stream scene graphs involving splicing one content stream with another content stream, e.g. for inserting or substituting an advertisement
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Theoretical Computer Science (AREA)
- Medical Informatics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Marketing (AREA)
- Health & Medical Sciences (AREA)
- Television Signal Processing For Recording (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
Abstract
The embodiment of the invention provides a kind of detection method of video frame, device and electronic equipments, belong to video detection technology field.This method includes：Obtain video frame to be detected；Obtain the Hash feature of the video frame to be detected；Pre-stored each Hash feature samples in the Hash feature of the video frame to be detected and database are subjected to characteristic matching respectively；The Hash feature samples stored in the database are a kind of head of pre-stored style and/or each Hash feature samples of run-out；If the Hash feature of the video frame to be detected is consistent with a Hash feature samples characteristic matching in Hash feature samples pre-stored in database, the head frame or piece the tail frame video frame to be detected being determined as in video to be detected.The detection efficiency of the head and/or run-out to video file can be improved using the present invention.
Description
Technical field
The present invention relates to video detection technology fields, more particularly to a kind of detection method of video frame, device and electronics
Equipment.
Background technique
Currently, there is head/run-out effect that a large amount of certain softwares automatically generate in the homemade video that user uploads, these
Short video title/trailer content may occupy the 10% of short video length, the case where not skipping head/run-out option automatically
Under, influence user's viewing experience.
Therefore, it is embodied to promote user, video server end can carry out head/run-out to the homemade video that user uploads
Detection.The principle of head detection is identical as the principle that run-out detects, and head detection and run-out detection are referred to as to mesh below
Mark the detection of video clip.
Currently, the detection to target video segment mainly passes through manually mark head/run-out point, that is, artificial mark
Outpour head/run-out start frame and end frame.However, manually indicating head/tail point since short number of videos is numerous and needing
Excess resource is expended, so that detection efficiency is low.
Summary of the invention
The detection method for being designed to provide a kind of video frame, device and the electronic equipment of the embodiment of the present invention, to improve
The detection efficiency of head and/or run-out to video file.Specific technical solution is as follows：
In a first aspect, a kind of detection method of video frame is provided, the method includes：
Obtain video frame to be detected；
Obtain the Hash feature of the video frame to be detected；
Each Hash feature samples pre-stored in the Hash feature of the video frame to be detected and database are distinguished
Carry out characteristic matching；The Hash feature samples stored in the database are the pre-stored a kind of head and/or piece of style
Each Hash feature samples of tail；
If one in the Hash feature of the video frame to be detected and database in pre-stored Hash feature samples
Hash feature samples characteristic matching is consistent, head frame or the run-out video frame to be detected being determined as in video to be detected
Frame.
Optionally, the Hash feature samples, including：The perceived hash characteristics of each Sample video frame and average Hash
Feature；
The step of Hash feature for obtaining the video frame to be detected, including：
Calculate the perceived hash characteristics and average Hash feature of the video frame to be detected；
Each Hash feature samples pre-stored in the Hash feature of the video frame to be detected and database are distinguished
The step of carrying out characteristic matching, including：
By a kind of head of style pre-stored in the perceived hash characteristics of the video frame to be detected and database and/
Or each perceived hash characteristics sample of run-out carries out first distance calculating respectively；
By a kind of head of style pre-stored in the average Hash feature of the video frame to be detected and database and/
Or each average Hash feature samples of run-out carry out second distance calculating respectively；
If the Hash in the Hash feature of the video frame to be detected and database in pre-stored Hash feature samples
The first distance calculated result of feature samples feature is less than perceptual hash threshold value, and second distance calculated result is less than average Hash
Threshold value, it is determined that pre-stored Hash feature samples characteristic matching in the Hash feature and database of the video frame to be detected
Unanimously, so that the video frame successful match to be detected；
Alternatively, if in the Hash feature of the video frame to be detected and database in pre-stored Hash feature samples
Each Hash feature samples feature matches inconsistent, then it fails to match for the video frame to be detected.
Optionally, the step that the video frame to be detected is determined as to head frame or piece tail frame in video to be detected
After rapid, further include：
It selects the video frame not detected as video frame to be detected, returns to the acquisition video frame to be detected
The step of Hash feature.
Optionally, the step of acquisition video frame to be detected, including：Using all video frames of video to be detected as not
Video frame is detected, is never detected in video frame by the sequence of broadcasting and obtains a frame as video frame to be detected.
Optionally, if carrying out head detection to video to be detected, the step of acquisition video frame to be detected, including：
Video frame in the first preset duration originated in the video to be detected is determined as the first detection range；
Key frame is extracted out of described first detection range, is determined as being used to detect head in the video to be detected
First does not detect video frame；It is not detected in video frame by extraction sequence from first and obtains a frame as the first video frame to be detected；
Pre-stored each Hash feature samples in the Hash feature by the video frame to be detected and database
The step of progress characteristic matching is respectively：By the Hash feature of the first video frame to be detected with it is pre-stored each in database
Head Hash feature samples carry out characteristic matching respectively；
The step of video frame for selecting one not detect is as video frame to be detected for：By the sequential selection one of broadcasting
A first does not detect video frame as the first video frame to be detected；
If carrying out run-out detection to video to be detected, the step of acquisition video frame to be detected, including：
Video frame in the second preset duration before terminating in the video to be detected is determined as the second detection range；
Key frame is extracted out of described second detection range, is determined as being used to detect run-out in the video to be detected
Second does not detect video frame；It is not detected in video frame by extraction sequence from second and obtains a frame as the second video frame to be detected；
Pre-stored each Hash feature samples in the Hash feature by the video frame to be detected and database
The step of progress characteristic matching is respectively：By the Hash feature of the second video frame to be detected with it is pre-stored each in database
Run-out Hash feature samples carry out characteristic matching respectively；
The step of video frame for selecting one not detect is as current video frame to be detected for：It is selected by the sequence of broadcasting
It selects one second and does not detect video frame as the second video frame to be detected.
Optionally, the use for extracting key frame out of described first detection range, being determined as in the video to be detected
In the step of detect head first does not detect video frame, including：
By the first preset interval, multiple first are extracted out of described first detection range and does not detect video frame；
It is described to extract key frame out of described second detection range, it is determined as being used for detection lug in the video to be detected
Not the step of the second of head does not detect video frame, including：
By the second preset interval, multiple second are extracted out of described second detection range and does not detect video frame.
Optionally, by sequential selection one first of broadcasting do not detect video frame as the first video frame to be detected it
Before, this method further includes：
Judge matching result that upper one has carried out the matched first video frame to be detected whether with current matching
The matching result of one video frame to be detected is identical；
If it is not the same, then by upper one carried out the first of the matched first video frame to be detected and current matching to
The video frame not detected between detection video frame is determined as first and does not detect video frame；It executes described by the sequential selection one played
A first does not detect the step of video frame is as the first video frame to be detected；
Alternatively, execution is described not to detect video frame as first by sequential selection one first of broadcasting if identical
The step of video frame to be detected；
Before not detecting video frame as the second video frame to be detected by sequential selection one second of broadcasting, this method
Further include：
Judge matching result that upper one has carried out the matched second video frame to be detected whether with current matching
The matching result of two video frames to be detected is identical；
If it is not the same, then by upper one carried out the second of the matched second video frame to be detected and current matching to
The video frame not detected between detection video frame is determined as second and does not detect video frame；It executes described by the sequential selection one played
A second does not detect the step of video frame is as the second video frame to be detected；
Alternatively, execution is described not to detect video frame as second by sequential selection one second of broadcasting if identical
The step of video frame to be detected.
Optionally, the method also includes：To the head frame or piece tail frame being confirmed as in the video to be detected into
Row fusion, obtains target video segment.
Optionally, the described pair of head frame being confirmed as in the video to be detected or piece tail frame merge, and obtain
The step of target video segment, including：
Obtain the temporal information of the head frame or piece tail frame that are each confirmed as in the video to be detected；
The video frame of Time Continuous is determined as a matching sub-piece；
Judge whether the time difference between adjacent matching sub-piece is less than or equal to preset time of fusion difference threshold value；
The neighbor sub-piece that time difference is less than or equal to preset time of fusion difference threshold value is permeated matching
Segment；
Selected from fused matching segment one as target video segment.
Optionally, it is described selected from fused matching segment one as target video segment the step of, including：
By first or the last one matching segment is determined as target video segment.
Optionally, it is described selected from fused matching segment one as target video segment the step of, including：
It calculates in each matching segment, ratio shared by the head frame or piece tail frame being confirmed as in the video to be detected
Example value；
The ratio value is maximum, and the matching segment for being greater than preset ratio threshold value is determined as current matching segment；
If current matching segment includes the head frame in video to be detected in video to be detected, current matching segment is judged
Next matching segment whether meet preset first splicing condition；
If meeting preset first splicing condition, current matching segment is spliced with next segment that matches,
Spliced matching segment is determined as current matching segment, returns to next matching segment of the judgement current matching segment
The step of whether meeting preset first splicing condition；
If being unsatisfactory for preset first splicing condition, current matching segment is determined as target video segment；
If current matching segment includes the piece tail frame in video to be detected in video to be detected, current matching segment is judged
Upper matching segment whether meet preset second splicing condition；
If meeting preset second splicing condition, current matching segment is matched into segment with upper one and is spliced,
Spliced matching segment is determined as current matching segment, returns to the upper matching segment of the judgement current matching segment
The step of whether meeting preset second splicing condition；
If being unsatisfactory for preset second splicing condition, current matching segment is determined as target video segment.
Optionally, the preset first splicing condition is：If current matching segment and next time difference for matching segment
Less than or equal to default first splicing threshold value and next matching segment is confirmed as the head frame in the video to be detected
Or ratio value >=max (α ss shared by piece tail framei_best,sst), it is determined that current clip and next matching segment
Spliced；
The preset second splicing condition is：If the time difference that current matching segment matches segment with upper one is less than or waits
Splice threshold value in default second and the upper matching segment is confirmed as head frame or run-out in the video to be detected
Ratio value >=max (α ss shared by framei_best,sst), it is determined that segment is matched with described upper one to output segment and is spelled
It connects；
Wherein α is goal-selling threshold value, ssi_bestTo be confirmed as the head frame or run-out in the video to be detected
Shared by the head frame being confirmed as in the video to be detected or piece tail frame of the maximum matching segment of ratio value shared by frame
Ratio value, sstFor preset ratio threshold value.
Optionally, if calculated maximum ratio value is not more than preset ratio threshold value, and current in video to be detected
With the head frame that segment includes in video to be detected, then first matching segment is determined as target video segment；
If calculated maximum ratio value is not more than preset ratio threshold value, and current matching segment packet in video to be detected
The piece tail frame in video to be detected is included, then the last one matching segment is determined as target video segment.
Second aspect, provides a kind of detection device of video frame, and described device includes：
First obtains module, for obtaining video frame to be detected；
Second obtains module, for obtaining the Hash feature of the video frame to be detected；
Matching module, for by pre-stored each Hash in the Hash feature of the video frame to be detected and database
Feature samples carry out characteristic matching respectively；The Hash feature samples stored in the database are a kind of pre-stored style
Each Hash feature samples of head and/or run-out；
Matching result determining module, if for the video frame to be detected Hash feature with it is pre-stored in database
A Hash feature samples characteristic matching in Hash feature samples is consistent, and the video frame to be detected is determined as view to be detected
Head frame or piece tail frame in frequency.
Optionally, the Hash feature samples, including：The perceived hash characteristics of each Sample video frame and average Hash
Feature；
Described second obtains module, is specifically used for：Calculate the video frame to be detected perceived hash characteristics and average Kazakhstan
Uncommon feature；
The matching module, including：First distance computing unit and second distance computing unit；
The first distance computing unit, for will in the perceived hash characteristics of the video frame to be detected and database it is pre-
A kind of head of the style first stored and/or each perceived hash characteristics sample of run-out carry out first distance calculating respectively；
The second distance computing unit, for will in the average Hash feature of the video frame to be detected and database it is pre-
A kind of head of the style first stored and/or each average Hash feature samples of run-out carry out second distance calculating respectively；
The matching result determining module, is specifically used for：If in the Hash feature of the video frame to be detected and database
The first distance calculated result of Hash feature samples feature in pre-stored Hash feature samples is less than perceptual hash threshold value,
And second distance calculated result is less than average Hash threshold value, it is determined that in the Hash feature and database of the video frame to be detected
Pre-stored Hash feature samples characteristic matching is consistent, so that the video frame successful match to be detected；
Alternatively, if in the Hash feature of the video frame to be detected and database in pre-stored Hash feature samples
Each Hash feature samples feature matches inconsistent, then it fails to match for the video frame to be detected.
Optionally, further include selecting module, the video frame to be detected is determined as in video to be detected for described
After the step of head frame or piece tail frame,
It selects the video frame not detected as video frame to be detected, returns to the acquisition video frame to be detected
The step of Hash feature.
Optionally, described first module is obtained, be specifically used for：Using all video frames of video to be detected as not detecting view
Frequency frame is never detected in video frame by the sequence of broadcasting and obtains a frame as current video frame to be detected.
Optionally, described first module is obtained, including：First detection unit and second detection unit；First detection
Unit, including：First range determines that subelement and the first video frame to be detected determine subelement；
First range determines subelement, is used for when carrying out head detection to video to be detected, will be described to be detected
The video frame in the first preset duration originated in video is determined as the first detection range；
First video frame to be detected determines subelement, for extracting key frame out of described first detection range, really
Be set in the video to be detected for detecting the first video frame to be detected of head, by extraction sequence from the first view to be detected
A frame is obtained in frequency frame as the current first video frame to be detected；
The matching module is used for when carrying out head detection to video to be detected, by the current first video frame to be detected
Hash feature and database in pre-stored each head Hash feature samples carry out characteristic matching respectively；
The selecting module, for video to be detected carry out head detection when, by broadcasting sequential selection one not
First video frame to be detected of detection is as the current first video frame to be detected；
The second detection unit, including：Second range determines that subelement and the second video frame to be detected determine subelement；
Second range determines subelement, is used for when carrying out run-out detection to video to be detected, will be described to be detected
The video frame in the second preset duration before terminating in video is determined as the second detection range；
Second video frame to be detected determines subelement, for extracting key frame out of described second detection range, really
Be set in the video to be detected for detecting the second video frame to be detected of run-out, by extraction sequence from the second view to be detected
A frame is obtained in frequency frame as the current second video frame to be detected；
The matching module is used for when carrying out run-out detection to video to be detected, by the current second video frame to be detected
Hash feature and database in pre-stored each run-out Hash feature samples carry out characteristic matching respectively；
The selecting module, for video to be detected carry out run-out detection when, by broadcasting sequential selection one not
Second video frame to be detected of detection is as the current second video frame to be detected.
Optionally, the described first video frame to be detected determines subelement, is specifically used for：By the first preset interval, from described
Multiple first is extracted in first detection range does not detect video frame；
Second video frame to be detected determines subelement, is specifically used for：By the second preset interval, from second detection
Multiple second is extracted in range does not detect video frame.
Optionally, described device further includes：
First judges matching result module, for not detecting in the selecting module by sequential selection one first of broadcasting
Before video frame is as the first video frame to be detected,
Judge matching result that upper one has carried out the matched first video frame to be detected whether with current matching
The matching result of one video frame to be detected is identical；
If it is not the same, then by upper one carried out the first of the matched first video frame to be detected and current matching to
The video frame not detected between detection video frame is determined as first and does not detect video frame；It executes described by the sequential selection one played
A first does not detect the step of video frame is as the first video frame to be detected；
Alternatively, execution is described not to detect video frame as first by sequential selection one first of broadcasting if identical
The step of video frame to be detected；
Described device further includes：
Second judges matching result module, for not detecting in the selecting module by sequential selection one second of broadcasting
Before video frame is as the second video frame to be detected,
Judge matching result that upper one has carried out the matched second video frame to be detected whether with current matching
The matching result of one video frame to be detected is identical；
If it is not the same, then by upper one carried out the second of the matched second video frame to be detected and current matching to
The video frame not detected between detection video frame is determined as second and does not detect video frame；It executes described by the sequential selection one played
A second does not detect the step of video frame is as the second video frame to be detected；
Alternatively, execution is described not to detect video frame as second by sequential selection one second of broadcasting if identical
The step of video frame to be detected.
Optionally, the Fusion Module, including：Temporal information acquiring unit, matching sub-piece determination unit, time of fusion
Poor judging unit, matching segment composition unit and target video Piece Selection unit；
Temporal information acquiring unit, for obtaining the head frame or run-out that are each confirmed as in the video to be detected
The temporal information of frame；
Sub-piece determination unit is matched, for the video frame of Time Continuous to be determined as a matching sub-piece；
Time of fusion difference judging unit, for judging whether the time difference between adjacent matching sub-piece is less than or equal to
Preset time of fusion difference threshold value；
Segment composition unit is matched, for the time difference to be less than or equal to the neighbor of preset time of fusion difference threshold value
Sub-piece permeates a matching segment；
Target video Piece Selection unit, for selected from fused matching segment one as target video piece
Section.
Optionally, the target video Piece Selection unit, is specifically used for：By first or the last one matching segment is true
It is set to target video segment.
Optionally, the target video Piece Selection unit, including：Ratio value computation subunit, current matching segment are true
Stator unit, the first splicing condition judgment sub-unit, the first splicing subelement, first object video clip determine subelement, the
Two splicing condition judgment sub-units, the second splicing subelement and the second target video segment determine subelement；
Ratio value computation subunit, for calculating in each matching segment, the piece being confirmed as in the video to be detected
Ratio value shared by head frame or piece tail frame；
Current matching segment determines subelement, for the ratio value is maximum, and is greater than the matching of preset ratio threshold value
Segment is determined as current matching segment；
First splicing condition judgment sub-unit, if including in video to be detected for current matching segment in video to be detected
Head frame, then judge whether next matching segment of current matching segment meets preset first splicing condition；
First splicing subelement, if for meeting preset first splicing condition, by current matching segment with it is next
A matching segment is spliced, and spliced matching segment is determined as current matching segment, returns to the judgement current matching
Whether next matching segment of segment meets the step of preset first splicing condition；
First object video clip determines subelement, if for being unsatisfactory for preset first splicing condition, it will be current
Matching segment is determined as target video segment；
Second splicing condition judgment sub-unit, if including in video to be detected for current matching segment in video to be detected
Piece tail frame, then judge whether the upper matching segment of current matching segment meets preset second splicing condition；
Second splicing subelement, if for meeting preset second splicing condition, by current matching segment and upper one
A matching segment is spliced, and spliced matching segment is determined as current matching segment, returns to the judgement current matching
Whether the upper matching segment of segment meets the step of preset second splicing condition；
Second target video segment determines subelement, if for being unsatisfactory for preset second splicing condition, it will be current
Matching segment is determined as target video segment.
Optionally, the preset first splicing condition is：If current matching segment and next time for matching segment
Difference is less than or equal to default first splicing threshold value and next matching segment is confirmed as the head in the video to be detected
Ratio value >=max (α ss shared by frame or piece tail framei_best,sst), it is determined that current clip and next matching piece
Duan Jinhang splicing；
The preset second splicing condition is：If the time difference that current matching segment matches segment with upper one be less than or
Equal to default second splicing threshold value and the upper matching segment is confirmed as head frame or piece in the video to be detected
Ratio value >=max (α ss shared by tail framei_best,sst), it is determined that segment is matched with described upper one to output segment and is spelled
It connects；
Wherein α is goal-selling threshold value, ssi_bestTo be confirmed as the head frame or run-out in the video to be detected
The head frame of the maximum matching segment of ratio value shared by frame being confirmed as in the video to be detected or piece tail frame institute
The ratio value accounted for, sstFor preset ratio threshold value.
Optionally, the determining current matching segment subelement, is also used to：
If calculated maximum ratio value is not more than preset ratio threshold value, and current matching segment packet in video to be detected
The head frame in video to be detected is included, then first matching segment is determined as target video segment；
If calculated maximum ratio value is not more than preset ratio threshold value, and current matching segment packet in video to be detected
The piece tail frame in video to be detected is included, then the last one matching segment is determined as target video segment.
The third aspect, provides a kind of electronic equipment, the electronic equipment include processor, communication interface, memory and
Communication bus, wherein processor, communication interface, memory complete mutual communication by communication bus；
Memory, for storing computer program；
Processor when for executing the program stored on memory, realizes any method of claim 1-13
Step.
Present invention implementation additionally provides a kind of computer readable storage medium, storage in the computer readable storage medium
There is the step of computer program, the computer program realizes the detection method of any of the above-described video frame when being executed by processor.
The embodiment of the invention also provides a kind of computer program products comprising instruction, when it runs on computers
When, so that computer executes the detection method of any of the above-described video frame.
The detection method and system of video frame provided in an embodiment of the present invention, can be by the Hash of current video frame to be detected
Pre-stored each Hash feature samples carry out characteristic matching respectively in feature and database, will be with Hash feature in database
Sample characteristics match head frame or the frame that consistent video frame to be detected is determined as in video to be detected.In this way, using this hair
Bright embodiment can be directly with each video frame to be detected in video to be detected and head and/or run-out feature in database
Sample carries out feature comparison, detection head and/or trailer content is carried out to video to be detected based on image consistency, due to this hair
Bright embodiment is directly handled video frame information, thus to picture variation more robust, it is more convenient to use, accurately.
Certainly, implement any of the products of the present invention or method it is not absolutely required at the same reach all the above excellent
Point.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described.
Fig. 1 is a kind of detection method flow chart of video frame of the embodiment of the present invention；
Fig. 2 is a kind of detection method schematic diagram of video frame of the embodiment of the present invention；
Fig. 3 is the video frame fusion method flow chart in a kind of detection method of video frame of the embodiment of the present invention；
Fig. 4 is the method flow of the selection target video clip in a kind of detection method of video frame of the embodiment of the present invention
Figure；
Fig. 5 is a kind of structure of the detecting device schematic diagram of video frame of the embodiment of the present invention；
Fig. 6 is the structural schematic diagram of a kind of electronic equipment provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention is described.
In the prior art, to the mainly artificial mark head/run-out point of the detection of target video segment, that is, people
Work marks out head/run-out start frame and end frame.However, manually indicating head/tail point since short number of videos is numerous
It needs to expend excess resource, so that detection efficiency is low.
Based on above-mentioned consideration, the present invention provides a kind of detection method of video frame, device and electronic equipments.The above method
It can be applied to server, including：Obtain video frame to be detected；Obtain the Hash feature of the video frame to be detected；It will be described
Pre-stored each Hash feature samples carry out characteristic matching respectively in the Hash feature and database of video frame to be detected；Institute
State each Hash feature sample of head and run-out that the Hash feature samples stored in database are a kind of pre-stored style
This；If a Hash spy in the Hash feature of the video frame to be detected and database in pre-stored Hash feature samples
It is consistent to levy sample characteristics matching, the head frame or piece the tail frame video frame to be detected being determined as in video to be detected.Benefit
With method provided in an embodiment of the present invention, target video segment can be identified based on image consistency, it will not be by video quality
With the influence of video time.The prior art that compares is by manually indicating head/run-out point, side provided in an embodiment of the present invention
Method can be improved the efficiency of head frame and/or piece the tail frame detection to video file.
The above method is introduced with specific embodiment below.
Referring to Fig. 1, Fig. 1 is a kind of detection method flow chart of video frame of the embodiment of the present invention, is included the following steps：
Step 101, video frame to be detected is obtained.
In a kind of implementation, the video frame to be detected of video to be detected is obtained by server, video to be detected can be
The short-sighted frequency of self-control uploaded by user.
In a kind of implementation, a frame can be obtained from video to be detected at random as video frame to be detected.
In a kind of implementation, a frame is obtained from video frame to be detected as current video to be detected by the sequence of broadcasting
Frame sequentially chooses video frame to be detected when carrying out head detection backward since first frame, when carrying out run-out detection, from
Last frame starts to choose video frame to be detected to front sequence.
Step 102, the Hash feature of the video frame to be detected is obtained.
Specifically, the Hash feature of the video to be detected includes perceived hash characteristics and average Hash feature, can be
Acquisition is calculated by server.
Step 103, by pre-stored each Hash feature in the Hash feature of the video frame to be detected and database
Sample carries out characteristic matching respectively；The Hash feature samples stored in the database are a kind of pre-stored head of style
And/or each Hash feature samples of run-out.
In a kind of implementation, it only can include each Hash feature samples of head in database, or only include piece
The Hash feature samples of tail, or the Hash feature samples comprising head and run-out；When the Hash for containing only head in database
When feature samples, it can only detect to match consistent head frame with sample characteristics in database；When containing only run-out in database
Hash feature samples when, can only detect to match consistent tail frame with sample characteristics in database；If containing in database
Head and run-out Hash feature samples, then can identify and match consistent head frame and run-out with sample characteristics in database
Frame.
In a kind of implementation, first by artificial judgment video title to be detected and/or the style of run-out, then will be to be checked
The Hash feature samples surveyed in video database corresponding with its style are matched.
For example, a video to be detected is that head content belongs to style 1 by artificial judgment, then by this video and wind to be detected
Sample is matched in the database of lattice 1.
In a kind of implementation, the Hash feature samples stored in database can be obtained in the following way.
Can be with the video of several (style A) head and/or run-out with a kind of specified style of artificial selection, and identify
The frame number of head and/or trailer content start frame in video and end frame, template data set and template as this method
Point information；
According to template point information, each frame data for belonging to head and/or trailer content are concentrated to template data, i.e., it is described
Sample video frame calculates separately perceived hash characteristics and average Hash feature, i.e. joint Hash feature, and deposits to the two
Storage constructs the head and/or run-out property data base of style A.
Wherein, the method for obtaining Hash feature samples can be：
Gray level image is converted by input frame image, computational length is the perceived hash characteristics of 64 dimensions；
Gray level image is converted by input frame image, computational length is the average Hash feature of 64 dimensions；
The calculated result of perceived hash characteristics sample and average Hash feature samples is spliced into feature vector, i.e., joint is breathed out
Uncommon feature, stores to database as the head of style A and/or the feature samples of run-out.
In fact, the perceived hash characteristics of calculating video frame and average Hash feature belong to the prior art, it is no longer superfluous here
It states.
In this step, by pre-stored a kind of piece of style in the Hash feature of each video frame to be detected and database
The Hash feature samples of head and/or run-out carry out can specifically include apart from calculating respectively：
The perceived hash characteristics of Sample video frame in the perceived hash characteristics of video to be detected and database are edited
Distance calculates.
The average Hash feature of Sample video frame in the average Hash feature and database of video to be detected is edited
Distance calculates.
Step 104, if the Hash feature of current video frame to be detected and pre-stored Hash feature samples in database
In a Hash feature samples characteristic matching it is consistent, the head frame video frame to be detected being determined as in video to be detected
Or piece tail frame.
In a kind of embodiment, each perceived hash characteristics sample in the perceived hash characteristics and database of video frame to be detected
Editing distance is less than default perceptual hash threshold value, and each average Hash in the average Hash feature and database of video frame to be detected
Feature samples are less than the video frame to be detected of default average Hash threshold value, then the Hash feature and data of current video frame to be detected
A Hash feature samples characteristic matching in library in pre-stored Hash feature samples is consistent.
Consistent feature samples are matched with video frame Hash character to be detected if can not find in the database, are recognized
Head or trailer content are not belonging to for current video frame.
Specifically, the video frame to be detected is determined as in video to be detected head frame or piece tail frame the step of it
Afterwards, further include：
It selects the video frame to be detected not detected as current video frame to be detected, it is described to be checked to return to the acquisition
The step of surveying the Hash feature of video frame.
It, can be according to video playing sequential selection video frame to be detected in a kind of implementation.
For example, when carrying out head detection, after having detected a video frame to be detected, by playing sequence selection it is next to
Detection video frame continues to test, and completes until all video frames to be detected detect；When carrying out run-out detection, one has been detected
After video frame to be detected, a upper video frame to be detected is selected to continue to test by playing sequence, until all videos to be detected
Frame detection is completed.
As it can be seen that detecting video frame using the embodiment of the present invention, the Hash feature and sample for calculating video frame to be detected are utilized
The editing distance of the Hash feature of video frame detects the video frame in video to be detected based on image consistency, can be quickly quasi-
Really detect the head and/or trailer content of video to be detected, it is easy to use.
Specifically, referring to fig. 2, Fig. 2 is a kind of detection method schematic diagram of video frame provided in an embodiment of the present invention, packet
It includes：First process and the second process.
First process：Establish the database of a kind of head of style and/or each Hash feature samples of run-out.
As shown in Fig. 2, it is possible, firstly, to be directed to each specified style artificial selection multiple template video, such as：For wind
Lattice A, style A can be the video for automatically generating head or run-out using video template by certain software for editing, select template view
Frequently 1 and template video 2.
Then, for the template video of selection 1 and template video 2, the frame model of manual identification head and/or run-out is carried out
It encloses.
In this step, with the start frame of manual identification's head and/or tail content in video and frame information can be terminated, will risen
Beginning frame and end frame information are as template point information.
Again for each video frame in frame range, according to the point information being manually entered, to belonging in head or run-out
Each frame data held calculate separately perceptual hash and average Hash feature, and calculated result is spliced into a feature vector, as wind
The head of lattice A or a Hash feature samples of run-out, store to database.
Specifically, the perceived hash characteristics and average Hash feature of each template video frame can be calculated, the two is spliced, is obtained
To the joint Hash feature of each template video frame, it is stored in head and/or run-out property data base.
In a kind of implementation, the perceived hash characteristics of each template video frame and the method for average Hash feature, packet are calculated
It includes：
Gray level image is converted by input frame image, computational length is the perceived hash characteristics of 64 dimensions；
Gray level image is converted by input frame image, computational length is the average Hash feature of 64 dimensions.
Second process：Video to be detected is detected.
As described in Figure 2, it for the detection of video to be detected, may include steps of：
Firstly, extracting the video frame that key frame is detected as needs from video to be detected.
Here it is possible to all regard all video frames of video frame to be detected as key frame.
It is of course also possible to first determine detection range, then extract key frame.
Specifically, if current matching is head matching, only using the preceding T second data of video as detection range.If current
Matching is that run-out matches, then only using the last T second data of video as detection range.
For the video data in detection range, 1 frame key frame is extracted every K frame.
Wherein, the typical set-up of T is 30-60, and the representative value of K is 5.
Such as：Typical head market is 5~20s, as long as then detecting first 1 minute or preceding half point of video to be detected
The video of clock whether have head can, do not need to detect all videos content.Specifically, can detecte view to be detected
First 30 seconds of frequency, the frame per second of general video are one second 30 frame, can extract a frame key frame every 5 frames according to playing sequence.
Specifically, if characteristic matching one of the time upper two adjacent key frames all with the sample Hash feature in database
It causes, that is, belongs to head or trailer content, then this two adjacent key frame is all determined as head frame or piece tail frame；If phase
The characteristic matching of adjacent two key frames and the Hash feature samples in database is inconsistent, i.e. a video frame belong to head or
Trailer content, one is not belonging to head or trailer content, then the video all detected the video frame between this two frame as needs
Frame does not detect video frame, is detected frame by frame.
Specifically, be directed to each key frame as video to be detected, frame 1~frame N, respectively with the sample that is stored in database
This video frame carries out image consistency matching, including：
The perceived hash characteristics and average Hash feature for calculating each frame, obtain joint Hash feature.It is specific to calculate
Method is identical as database creation process, is not repeated herein.
By a kind of head of style pre-stored in the Hash feature of video frame to be detected and database and/or run-out
Each Hash feature samples are carried out apart from calculating respectively.
Distance value is less than to the video frame to be detected of preset threshold, is determined as head in the video to be detected or run-out
Video frame.Namely obtain the matching result of each video frame to be detected.
Specific calculation is as follows：
Calculate each perceived hash characteristics sample in the perceived hash characteristics and database of video frame to be detected first is compiled
Distance is collected, and calculates the average Hash feature of video frame to be detected and the second of average Hash feature samples each in database and compiles
Collect distance.
If the first editing distance<T1, and the second editing distance<T2, then it is assumed that this video frame to be detected and database are Sino-Kazakhstan
Uncommon feature samples characteristic matching is consistent, that is, has image consistency.If the first editing distance<T1 and the second editing distance >=T2,
First editing distance >=T1 and the second editing distance<T2 or the first editing distance >=T1 and the second editing distance >=T2, then recognize
Video frame to be detected and Hash feature samples characteristic matching in database are inconsistent thus, that is, do not have image consistency.
Wherein, T1, T2 can be configured according to the Stringency that detection requires image consistency, it is desirable that stringenter, threshold
It is worth smaller, representative value can be between the 5%~25% of Hash characteristic length.
It optionally, can be by the head in video frame to be detected and database if carrying out head detection to video to be detected
Hash feature samples carry out characteristic matching, can be by video frame to be detected and data if carrying out run-out detection to video to be detected
Run-out Hash feature samples in library carry out characteristic matching.
In a kind of implementation, Hash feature samples are with markd in database, and head Hash feature samples have
Piece labeling head, run-out Hash feature samples are marked with run-out, so when video to be detected carries out head matching, need and data
Head Hash feature samples carry out characteristic matching in library, similarly, when video to be detected carries out run-out matching, only need to in database
Run-out Hash feature samples carry out characteristic matching.
Optionally, it will test that result belongs to head or the video frame of trailer content is denoted as fm, will test result and be not belonging to piece
The video frame of head or trailer content is denoted as fn。
Specifically, the video frame for the head or run-out that can be obtained according to matching result is merged and is filtered.
Specifically, being merged to the video frame of head or run-out in the video to be detected is confirmed as, then to fusion
Segment afterwards is filtered, and is obtained the video frame of head and/or run-out in video to be detected, is exported as testing result.
As it can be seen that detecting target video segment using the embodiment of the present invention, server can be based on image consistency, to be checked
It surveys video and carries out frame level detection, the accuracy of detection is substantially increased, and by the way of extracting key frame, to video to be detected
It is sampled detection, shortens detection time, detection speed is fast.
Specifically, Fig. 3 is the video frame in a kind of detection method of video frame provided in an embodiment of the present invention referring to Fig. 3
Fusion method flow chart, includes the following steps：
Step 301, obtain each be confirmed as the video frame of head frame or piece tail frame in the video to be detected when
Between information.
In a kind of implementation, target video segment in the video to be detected is each confirmed as by server acquisition
The time of video frame.
Step 302, the video frame of Time Continuous is determined as a matching sub-piece.
In a kind of implementation, video frame continuous in time is a matching sub-piece, in a matching segment, includes
Several matching sub-pieces.
Step 303, judge whether the time difference between adjacent matching sub-piece is less than or equal to preset time of fusion
Poor threshold value.
In a kind of implementation, a matching sub-piece Si, this matching segment start frame be denoted asThis matching segment
End frame be denoted asIf two adjacent matching segment Si, Si+1Frame pitch in time is not more than T3, i.e.,
Wherein T3 is the setting of frame level error tolerance, and the representative value of T3 is K.
Step 304, the neighbor sub-piece that the time difference is less than or equal to preset time of fusion difference threshold value is fused to
One matching segment.
In a kind of implementation, the neighbor sub-piece of preset time of fusion difference threshold value is less than or equal to the time difference
Mixing operation is carried out, i.e. update SiEnd frameAnd S is deleted in matching list of pattern clipsi+1。
Wherein SiFor i-th of matching segment,The last frame of segment is matched for i-th,It is a for (i+1)
Match the last frame of segment, Si+1For (i+1) a matching segment.
Step 305, selected from fused matching segment one as target video segment.
In a kind of implementation, if in fused matching segment only including a segment, using this segment as
Target video segment.If in fused matching segment include two or more matching segments, select one as
Object matching segment.
From the foregoing, it can be seen that it is each true that server will meet image consistency using method provided in an embodiment of the present invention
The video frame for being set to target video segment in the video to be detected is merged, and is made from fused matching Piece Selection one
For target video segment, the target video segment obtained in this way is more accurate, more convenient to use.
Specifically, described select one as target video segment from fused matching segment, fusion can choose
Afterwards first in matching segment or the last one matching segment are determined as target video segment.
Specifically, referring to fig. 4, Fig. 4 is the selection mesh in a kind of detection method of video frame provided in an embodiment of the present invention
The method flow diagram for marking video clip, includes the following steps：
Step 401, it calculates in each matching segment, the head frame or piece tail frame that are confirmed as in the video to be detected
Shared ratio value.
In a kind of implementation, it is confirmed as fused matching segment S in the video to be detectediIn meet the figure
As percentage of the frame number in this segment totalframes of coherence request, percentage is calculated by formula (1), is denoted as
Stability percentage.
Wherein, SsiThe stability percentage of segment, f are matched for i-thmFor the frame number of the video frame of target video segment,The end frame of segment is matched for i-th,The start frame of segment is matched for i-th.
Step 402, the ratio value is maximum, and the matching segment for being greater than preset ratio threshold value is determined as current matching piece
Section.
In a kind of implementation, by the maximum matching segment of the stability, and stability of this matching segment need to be greater than
Preset ratio threshold value selects current matching segment by formula (2), (3).
ssi_best=max (ssi|i∈[1,…,I]) (2)
ssi_best> sst (3)
Wherein, ssi_bestFor the stability of the maximum matching segment of the stability, ssiThe steady of segment is matched for i-th
Qualitative, I is the serial number of the last one matching segment, sstFor preset ratio threshold value.
Step 403, if current matching segment includes the head frame detected in video in video to be detected, judge current
Whether next matching segment with segment meets preset first splicing condition.
In a kind of implementation, by server judge current matching segment next matching segment whether meet it is preset
First splicing condition.
Step 404, if meeting preset first splicing condition, by current matching segment with it is next match segment into
Spliced matching segment is determined as current matching segment by row splicing, returns to the next of the judgement current matching segment
Whether matching segment meets the step of preset first splicing condition.
Step 405, if being unsatisfactory for preset first splicing condition, current matching segment is determined as target video piece
Section.
Step 406, if current matching segment includes the piece tail frame in video to be detected in video to be detected, judgement is current
Whether the upper matching segment of matching segment meets preset second splicing condition.
Step 407, if meeting preset second splicing condition, by current matching segment with upper one match segment into
Spliced matching segment is determined as current matching segment by row splicing, returns to upper one of the judgement current matching segment
Whether matching segment meets the step of preset second splicing condition.
Step 408, if being unsatisfactory for preset second splicing condition, current matching segment is determined as target video piece
Section.
From the foregoing, it can be seen that using method provided in an embodiment of the present invention, the object matching segment determined by server is more
Accurately, enable head and/or trailer content be accurately identified out, skip head and/or run-out after convenient automatically.
Specifically, the preset first splicing condition is：If current matching segment and next time difference for matching segment
Less than or equal to default first splicing threshold value and next matching segment is confirmed as the head frame in the video to be detected
Or ratio value >=max (α ss shared by piece tail framei_best,sst), it is determined that current clip and next matching segment
Spliced；
In a kind of implementation, default first splicing condition is：
And ssnext_i≥max(α·ssi_best,sst) (5)
Wherein,For next matching segment first frame, ResultendFor current matching segment last frame, T4
For default first splicing threshold value, the representative value of T4 is twice of T3, i.e. 2K, ssnext_iFor the stability of next segment, α is
Goal-selling threshold value, ssi_bestFor the stability of the maximum matching segment of the stability, sstFor preset ratio threshold value.
The preset second splicing condition is：If the time difference that current matching segment matches segment with upper one is less than or waits
Splice threshold value in default second and the upper matching segment is confirmed as head frame or run-out in the video to be detected
Ratio value >=max (α ss shared by framei_best,sst), it is determined that segment is matched with described upper one to output segment and is spelled
It connects；
In a kind of implementation, default second splicing condition is：
And sslast_i≥max(α·ssi_best,sst) (7)
Wherein, ResultstaFor the current matching segment first frame,For it is described it is upper one matching segment last
Frame, T5For default second splicing threshold value, the representative value of T5 is twice of T3, i.e. 2K, sslast_iFor the stabilization of a upper segment
Property, α is goal-selling threshold value, ssi_bestFor the stability of the maximum matching segment of the stability, sstFor preset ratio threshold
Value.
From the foregoing, it can be seen that server can be according to preset threshold to stability using method provided in an embodiment of the present invention
Highest matching segment is spliced with thereon/following segment, acquisition target video segment, in this way, obtained target video
Segment is more accurate.
Specifically, if calculated maximum ratio value is not more than preset ratio threshold value, and current in video to be detected
With the head frame that segment includes in video to be detected, then first matching segment is determined as target video segment；
If calculated maximum ratio value is not more than preset ratio threshold value, and current matching segment packet in video to be detected
The piece tail frame in video to be detected is included, then the last one matching segment is determined as target video segment.
In a kind of implementation, target video segment is determined as to the head segment or run-out segment of video to be detected.Example
Such as, comprising the head frame being confirmed as in video to be detected in target video segment, and target video segment is by video to be detected
The 1st frame formed to the 100th frame, then 100 frames before this video to be detected are determined as to the head segment of this video to be detected.
From the foregoing, it can be seen that obtained target video segment is more accurate using method provided in an embodiment of the present invention.Because
If calculated maximum scale value is not more than preset threshold, illustrate the similarity of video clip to be detected Yu Sample video segment
It is very low, so the accuracy of target fragment can be improved using the above method.
From the foregoing, it can be seen that server can be based on image consistency to be detected using method provided in an embodiment of the present invention
Target video segment in video is detected, and accuracy in detection is high, and speed is fast, easy to use.
Due to identical technical concept, corresponding to embodiment of the method shown in Fig. 1, the embodiment of the invention also provides a kind of views
The detection device of frequency frame, as shown in figure 5, the device includes：
First obtains module 501, for obtaining video frame to be detected；
Second obtains module 502, for obtaining the Hash feature of the video frame to be detected；
Matching module 503, for by the Hash feature of the video frame to be detected with it is pre-stored each in database
Hash feature samples carry out characteristic matching respectively；The Hash feature samples stored in the database are a kind of pre-stored wind
The head of lattice and/or each Hash feature samples of run-out；
Matching result determining module 504, if the Hash feature of the video frame to be detected with it is pre-stored in database
A Hash feature samples characteristic matching in Hash feature samples is consistent, and the video frame to be detected is determined as view to be detected
Head frame or piece tail frame in frequency.
In embodiments of the present invention, the Hash feature samples, including：The perceived hash characteristics of each Sample video frame
With average Hash feature；
Described second obtains module, is specifically used for：Calculate the video frame to be detected perceived hash characteristics and average Kazakhstan
Uncommon feature；
The matching module, including：First distance computing unit and second distance computing unit；
The first distance computing unit, for will in the perceived hash characteristics of the video frame to be detected and database it is pre-
A kind of head of the style first stored and/or each perceived hash characteristics sample of run-out carry out first distance calculating respectively；
The second distance computing unit, for will in the average Hash feature of the video frame to be detected and database it is pre-
A kind of head of the style first stored and/or each average Hash feature samples of run-out carry out second distance calculating respectively；
The matching result determining module, is specifically used for：If in the Hash feature of the video frame to be detected and database
The first distance calculated result of Hash feature samples feature in pre-stored Hash feature samples is less than perceptual hash threshold value,
And second distance calculated result is less than average Hash threshold value, it is determined that in the Hash feature and database of the video frame to be detected
Pre-stored Hash feature samples characteristic matching is consistent, so that the video frame successful match to be detected；
Alternatively, if in the Hash feature of the video frame to be detected and database in pre-stored Hash feature samples
Each Hash feature samples feature matches inconsistent, then it fails to match for the video frame to be detected.
It in embodiments of the present invention, further include selecting module, for being determined as the video frame to be detected for described
After the step of head frame or piece tail frame in video to be detected,
It selects the video frame not detected as video frame to be detected, returns to the acquisition video frame to be detected
The step of Hash feature.
In embodiments of the present invention, described first module is obtained, be specifically used for：All video frames of video to be detected are made
Not detect video frame, is never detected in video frame by the sequence of broadcasting and obtain a frame as video frame to be detected.
In embodiments of the present invention, described first module is obtained, including：First detection unit and second detection unit；
The first detection unit, including：First range determines that subelement and the first video frame to be detected determine subelement；
First range determines subelement, is used for when carrying out head detection to video to be detected, will be described to be detected
The video frame in the first preset duration originated in video is determined as the first detection range；
First video frame to be detected determines subelement, for extracting key frame out of described first detection range, really
It is set in the video to be detected and does not detect video frame for detecting the first of head, does not detect view from first by extraction sequence
A frame is obtained in frequency frame as the first video frame to be detected；
The matching module is used for when carrying out head detection to video to be detected, by the Kazakhstan of the first video frame to be detected
Pre-stored each head Hash feature samples carry out characteristic matching respectively in uncommon feature and database；
The selecting module, for when carrying out head detection to video to be detected, by the sequential selection one the of broadcasting
One does not detect video frame as the first video frame to be detected；
The second detection unit, including：Second range determines that subelement and the second video frame to be detected determine subelement；
Second range determines subelement, is used for when carrying out run-out detection to video to be detected, will be described to be detected
The video frame in the second preset duration before terminating in video is determined as the second detection range；
Second video frame to be detected determines subelement, for extracting key frame out of described second detection range, really
It is set in the video to be detected and does not detect video frame for detecting the second of run-out, does not detect view from second by extraction sequence
A frame is obtained in frequency frame as the second video frame to be detected；
The matching module is used for when carrying out head detection to video to be detected, by the Kazakhstan of the second video frame to be detected
Pre-stored each run-out Hash feature samples carry out characteristic matching respectively in uncommon feature and database；
The selecting module, for when carrying out run-out detection to video to be detected, by the sequential selection one the of broadcasting
Two do not detect video frame as the second video frame to be detected.
In embodiments of the present invention, the described first video frame to be detected determines subelement, is specifically used for：Between being preset by first
Every extracting multiple first out of described first detection range and do not detect video frame；
Second video frame to be detected determines subelement, is specifically used for：By the second preset interval, from second detection
Multiple second is extracted in range does not detect video frame.
In embodiments of the present invention, described device further includes：
Described device further includes：
First judges matching result module, for not detecting in the selecting module by sequential selection one first of broadcasting
Before video frame is as the first video frame to be detected,
Judge matching result that upper one has carried out the matched first video frame to be detected whether with current matching
The matching result of one video frame to be detected is identical；
If it is not the same, then by upper one carried out the first of the matched first video frame to be detected and current matching to
The video frame not detected between detection video frame is determined as first and does not detect video frame；It executes described by the sequential selection one played
A first does not detect the step of video frame is as the first video frame to be detected；
Alternatively, execution is described not to detect video frame as first by sequential selection one first of broadcasting if identical
The step of video frame to be detected；
Described device further includes：
Second judges matching result module, for not detecting in the selecting module by sequential selection one second of broadcasting
Before video frame is as the second video frame to be detected,
Judge matching result that upper one has carried out the matched second video frame to be detected whether with current matching
The matching result of one video frame to be detected is identical；
If it is not the same, then by upper one carried out the second of the matched second video frame to be detected and current matching to
The video frame not detected between detection video frame is determined as second and does not detect video frame；It executes described by the sequential selection one played
A second does not detect the step of video frame is as the second video frame to be detected；
Alternatively, execution is described not to detect video frame as second by sequential selection one second of broadcasting if identical
The step of video frame to be detected.
In embodiments of the present invention, described device further includes Fusion Module；
The Fusion Module, for melting to the head frame or piece tail frame that are confirmed as in the video to be detected
It closes, obtains target video segment.
In embodiments of the present invention, the Fusion Module, including：Temporal information acquiring unit, matching sub-piece determine single
Member, time of fusion difference judging unit, matching segment composition unit and target video Piece Selection unit；
Temporal information acquiring unit, for obtaining the head frame or run-out that are each confirmed as in the video to be detected
The temporal information of frame；
Sub-piece determination unit is matched, for the video frame of Time Continuous to be determined as a matching sub-piece；
Time of fusion difference judging unit, for judging whether the time difference between adjacent matching sub-piece is less than or equal to
Preset time of fusion difference threshold value；
Segment composition unit is matched, for the time difference to be less than or equal to the neighbor of preset time of fusion difference threshold value
Sub-piece permeates a matching segment；
Target video Piece Selection unit, for selected from fused matching segment one as target video piece
Section.
In embodiments of the present invention, the target video Piece Selection unit, is specifically used for：By first or the last one
Matching segment is determined as target video segment.
In embodiments of the present invention, the target video Piece Selection unit, including：It is ratio value computation subunit, current
Matching segment determines that subelement, the first splicing condition judgment sub-unit, the first splicing subelement, first object video clip determine
Subelement, the second splicing condition judgment sub-unit, the second splicing subelement and the second target video segment determine subelement；
Ratio value computation subunit, for calculating in each matching segment, the piece being confirmed as in the video to be detected
Ratio value shared by head frame or piece tail frame；
Current matching segment determines subelement, for the ratio value is maximum, and is greater than the matching of preset ratio threshold value
Segment is determined as current matching segment；
First splicing condition judgment sub-unit, if including in video to be detected for current matching segment in video to be detected
Head frame, then judge whether next matching segment of current matching segment meets preset first splicing condition；
First splicing subelement, if for meeting preset first splicing condition, by current matching segment with it is next
A matching segment is spliced, and spliced matching segment is determined as current matching segment, returns to the judgement current matching
Whether next matching segment of segment meets the step of preset first splicing condition；
First object video clip determines subelement, if for being unsatisfactory for preset first splicing condition, it will be current
Matching segment is determined as target video segment；
Second splicing condition judgment sub-unit, if including in video to be detected for current matching segment in video to be detected
Piece tail frame, then judge whether the upper matching segment of current matching segment meets preset second splicing condition；
Second splicing subelement, if for meeting preset second splicing condition, by current matching segment and upper one
A matching segment is spliced, and spliced matching segment is determined as current matching segment, returns to the judgement current matching
Whether the upper matching segment of segment meets the step of preset second splicing condition；
Second target video segment determines subelement, if for being unsatisfactory for preset second splicing condition, it will be current
Matching segment is determined as target video segment.
In embodiments of the present invention, the preset first splicing condition is：If current matching segment and next matching
The time difference of segment is less than or equal to default first splicing threshold value and next matching segment is confirmed as the view to be detected
Ratio value >=max (α ss shared by head frame and piece tail frame in frequencyi_best,sst), it is determined that current clip and it is described under
One matching segment is spliced；
The preset second splicing condition is：If the time difference that current matching segment matches segment with upper one be less than or
Equal to default second splicing threshold value and the upper matching segment is confirmed as head frame or piece in the video to be detected
Ratio value >=max (α ss shared by tail framei_best,sst), it is determined that segment is matched with described upper one to output segment and is spelled
It connects；
Wherein α is goal-selling threshold value, ssi_bestTo be confirmed as the head frame or run-out in the video to be detected
Shared by the head frame being confirmed as in the video to be detected or piece tail frame of the maximum matching segment of ratio value shared by frame
Ratio value, sstFor preset ratio threshold value.
In embodiments of the present invention, the determining current matching segment subelement, is also used to：
If calculated maximum ratio value is not more than preset ratio threshold value, and current matching segment packet in video to be detected
The head frame in video to be detected is included, then first matching segment is determined as target video segment；
If calculated maximum ratio value is not more than preset ratio threshold value, and current matching segment packet in video to be detected
The piece tail frame in video to be detected is included, then the last one matching segment is determined as target video segment.
The embodiment of the invention also provides a kind of electronic equipment, as shown in fig. 6, include processor 601, communication interface 602,
Memory 603 and communication bus 604, wherein processor 601, communication interface 602, memory 603 are complete by communication bus 604
At mutual communication,
Memory 603, for storing computer program；
Processor 601 when for executing the program stored on memory 603, realizes following steps：
Obtain video frame to be detected；
Obtain the Hash feature of the video frame to be detected；
Each Hash feature samples pre-stored in the Hash feature of the video frame to be detected and database are distinguished
Carry out characteristic matching；The Hash feature samples stored in the database are the pre-stored a kind of head and/or piece of style
Each Hash feature samples of tail；
If one in the Hash feature of the video frame to be detected and database in pre-stored Hash feature samples
Hash feature samples characteristic matching is consistent, head frame or the run-out video frame to be detected being determined as in video to be detected
Frame.
There are the above embodiments as it can be seen that since the embodiment of the present invention is Sino-Kazakhstan by the Hash feature of video to be detected and database
Uncommon feature samples are carried out apart from calculating, therefore be can use image consistency and judged whether frame to be detected is in head or run-out
Hold, and by extracting key frame, shortens the detection used time.In turn, the target of video to be detected is detected using the embodiment of the present invention
When video clip, target video segment can be rapidly and accurately obtained.
The communication bus that above-mentioned electronic equipment is mentioned can be Peripheral Component Interconnect standard (Peripheral Component
Interconnect, PCI) bus or expanding the industrial standard structure (Extended Industry Standard
Architecture, EISA) bus etc..The communication bus can be divided into address bus, data/address bus, control bus etc..For just
It is only indicated with a thick line in expression, figure, it is not intended that an only bus or a type of bus.
Communication interface is for the communication between above-mentioned electronic equipment and other equipment.
Memory may include random access memory (Random Access Memory, RAM), also may include non-easy
The property lost memory (Non-Volatile Memory, NVM), for example, at least a magnetic disk storage.Optionally, memory may be used also
To be storage device that at least one is located remotely from aforementioned processor.
Above-mentioned processor can be general processor, including central processing unit (Central Processing Unit,
CPU), network processing unit (Network Processor, NP) etc.；It can also be digital signal processor (Digital Signal
Processing, DSP), it is specific integrated circuit (Application Specific Integrated Circuit, ASIC), existing
It is field programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete
Door or transistor logic, discrete hardware components.
In another embodiment provided by the invention, a kind of computer readable storage medium is additionally provided, which can
It reads to be stored with computer program in storage medium, the computer program realizes any of the above-described video frame when being executed by processor
The step of method.
In another embodiment provided by the invention, a kind of computer program product comprising instruction is additionally provided, when it
When running on computers, so that the method that computer executes any video frame in above-described embodiment.
In the above-described embodiments, can come wholly or partly by software, hardware, firmware or any combination thereof real
It is existing.When implemented in software, it can entirely or partly realize in the form of a computer program product.The computer program
Product includes one or more computer instructions.When loading on computers and executing the computer program instructions, all or
It partly generates according to process or function described in the embodiment of the present invention.The computer can be general purpose computer, dedicated meter
Calculation machine, computer network or other programmable devices.The computer instruction can store in computer readable storage medium
In, or from a computer readable storage medium to the transmission of another computer readable storage medium, for example, the computer
Instruction can pass through wired (such as coaxial cable, optical fiber, number from a web-site, computer, server or data center
User's line (DSL)) or wireless (such as infrared, wireless, microwave etc.) mode to another web-site, computer, server or
Data center is transmitted.The computer readable storage medium can be any usable medium that computer can access or
It is comprising data storage devices such as one or more usable mediums integrated server, data centers.The usable medium can be with
It is magnetic medium, (for example, floppy disk, hard disk, tape), optical medium (for example, DVD) or semiconductor medium (such as solid state hard disk
Solid State Disk (SSD)) etc..
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including the element.
Each embodiment in this specification is all made of relevant mode and describes, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for device,
For the embodiments such as electronic equipment, since it is substantially similar to the method embodiment, so being described relatively simple, related place ginseng
See the part explanation of embodiment of the method.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the scope of the present invention.It is all
Any modification, equivalent replacement, improvement and so within the spirit and principles in the present invention, are all contained in protection scope of the present invention
It is interior.
Claims (27)
1. a kind of detection method of video frame, which is characterized in that including：
Obtain video frame to be detected；
Obtain the Hash feature of the video frame to be detected；
The Hash feature of the video frame to be detected and each Hash feature samples pre-stored in database are carried out respectively
Characteristic matching；The Hash feature samples stored in the database are the head and/or run-out of a kind of pre-stored style
Each Hash feature samples；
If a Hash in the Hash feature of the video frame to be detected and database in pre-stored Hash feature samples
Feature samples characteristic matching is consistent, the head frame or piece the tail frame video frame to be detected being determined as in video to be detected.
2. according to the method described in claim 1, it is characterized in that：The Hash feature samples, including：Each Sample video
The perceived hash characteristics of frame and average Hash feature；
The step of Hash feature for obtaining the video frame to be detected, including：
Calculate the perceived hash characteristics and average Hash feature of the video frame to be detected；
The Hash feature of the video frame to be detected and each Hash feature samples pre-stored in database are carried out respectively
The step of characteristic matching, including：
By a kind of head and/or piece of style pre-stored in the perceived hash characteristics of the video frame to be detected and database
Each perceived hash characteristics sample of tail carries out first distance calculating respectively；
By the pre-stored a kind of head and/or piece of style in the average Hash feature of the video frame to be detected and database
Each average Hash feature samples of tail carry out second distance calculating respectively；
If the Hash feature in the Hash feature of the video frame to be detected and database in pre-stored Hash feature samples
The first distance calculated result of sample characteristics is less than perceptual hash threshold value, and second distance calculated result is less than average Hash threshold
Value, it is determined that pre-stored Hash feature samples characteristic matching one in the Hash feature and database of the video frame to be detected
It causes, so that the video frame successful match to be detected；
Alternatively, if each of the Hash feature of the video frame to be detected and Hash feature samples pre-stored in database
Hash feature samples feature matches inconsistent, then it fails to match for the video frame to be detected.
3. the method according to claim 1, wherein described be determined as view to be detected for the video frame to be detected
After the step of head frame or piece tail frame in frequency, further include：
It selects the video frame not detected as video frame to be detected, returns to the Hash for obtaining the video frame to be detected
The step of feature.
4. according to the method described in claim 3, it is characterized in that：
The step of acquisition video frame to be detected, including：Using all video frames of video to be detected as not detecting video frame,
It is never detected in video frame by the sequence of broadcasting and obtains a frame as video frame to be detected.
5. according to the method described in claim 3, it is characterized in that：
If carrying out head detection to video to be detected, the step of acquisition video frame to be detected, including：
Video frame in the first preset duration originated in the video to be detected is determined as the first detection range；
Key frame is extracted out of described first detection range, is determined as being used to detect the first of head in the video to be detected
Video frame is not detected；It is not detected in video frame by extraction sequence from first and obtains a frame as the first video frame to be detected；
Pre-stored each Hash feature samples are distinguished in the Hash feature by the video frame to be detected and database
Carry out characteristic matching the step of be：By pre-stored each head in the Hash feature of the first video frame to be detected and database
Hash feature samples carry out characteristic matching respectively；
The step of video frame for selecting one not detect is as video frame to be detected for：By sequential selection one of broadcasting
One does not detect video frame as the first video frame to be detected；
If carrying out run-out detection to video to be detected, the step of acquisition video frame to be detected, including：
Video frame in the second preset duration before terminating in the video to be detected is determined as the second detection range；
Key frame is extracted out of described second detection range, is determined as being used to detect the second of run-out in the video to be detected
Video frame is not detected；It is not detected in video frame by extraction sequence from second and obtains a frame as the second video frame to be detected；
Pre-stored each Hash feature samples are distinguished in the Hash feature by the video frame to be detected and database
Carry out characteristic matching the step of be：By pre-stored each run-out in the Hash feature of the second video frame to be detected and database
Hash feature samples carry out characteristic matching respectively；
The step of video frame for selecting one not detect is as video frame to be detected for：By sequential selection one of broadcasting
Two do not detect video frame as the second video frame to be detected.
6. according to the method described in claim 5, it is characterized in that：
It is described to extract key frame out of described first detection range, it is determined as being used to detect head in the video to be detected
First the step of not detecting video frame, including：
By the first preset interval, multiple first are extracted out of described first detection range and does not detect video frame；
It is described to extract key frame out of described second detection range, it is determined as being used to detect head in the video to be detected
Second the step of not detecting video frame, including：
By the second preset interval, multiple second are extracted out of described second detection range and does not detect video frame.
7. according to the method described in claim 5, it is characterized in that：
Before not detecting video frame as the first video frame to be detected by sequential selection one first of broadcasting, this method is also wrapped
It includes：
Judge matching result that upper one has carried out the matched first video frame to be detected whether with the first of current matching to
The matching result for detecting video frame is identical；
If it is not the same, then having carried out the first to be detected of the matched first video frame to be detected and current matching for upper one
The video frame not detected between video frame is determined as first and does not detect video frame；It executes described by playing sequence selection one first
The step of video frame is as the first video frame to be detected is not detected；
Alternatively, if identical, execute that described by sequential selection one first of broadcasting not detect video frame to be checked as first
The step of surveying video frame；
Before not detecting video frame as the second video frame to be detected by sequential selection one second of broadcasting, this method is also wrapped
It includes：
Judge matching result that upper one has carried out the matched second video frame to be detected whether with the second of current matching to
The matching result for detecting video frame is identical；
If it is not the same, then having carried out the second to be detected of the matched second video frame to be detected and current matching for upper one
The video frame not detected between video frame is determined as second and does not detect video frame；It executes described by the sequential selection played one the
Two do not detect the step of video frame is as the second video frame to be detected；
Alternatively, if identical, execute that described by sequential selection one second of broadcasting not detect video frame to be checked as second
The step of surveying video frame.
8. described in any item methods according to claim 1~7, it is characterised in that：
The method also includes：The head frame or piece tail frame that are confirmed as in the video to be detected are merged, obtained
Target video segment.
9. according to the method described in claim 8, it is characterized in that：
The described pair of head frame being confirmed as in the video to be detected or piece tail frame merge, and obtain target video segment
The step of, including：
Obtain the temporal information of the head frame or piece tail frame that are each confirmed as in the video to be detected；
The video frame of Time Continuous is determined as a matching sub-piece；
Judge whether the time difference between adjacent matching sub-piece is less than or equal to preset time of fusion difference threshold value；
The neighbor sub-piece that time difference is less than or equal to preset time of fusion difference threshold value is permeated a matching segment；
Selected from fused matching segment one as target video segment.
10. according to the method described in claim 9, it is characterized in that：
It is described selected from fused matching segment one as target video segment the step of, including：
By first or the last one matching segment is determined as target video segment.
11. according to the method described in claim 9, it is characterized in that：
It is described selected from fused matching segment one as target video segment the step of, including：
It calculates in each matching segment, ratio shared by the head frame or piece tail frame being confirmed as in the video to be detected
Value；
The ratio value is maximum, and the matching segment for being greater than preset ratio threshold value is determined as current matching segment；
If current matching segment includes the head frame in video to be detected in video to be detected, judge under current matching segment
Whether one matching segment meets preset first splicing condition；
If meeting preset first splicing condition, current matching segment is spliced with next segment that matches, will be spelled
Matching segment after connecing is determined as current matching segment, whether returns to the next matching segment for judging current matching segment
The step of meeting preset first splicing condition；
If being unsatisfactory for preset first splicing condition, current matching segment is determined as target video segment；
If current matching segment includes the piece tail frame in video to be detected in video to be detected, the upper of current matching segment is judged
Whether one matching segment meets preset second splicing condition；
If meeting preset second splicing condition, current matching segment is matched into segment with upper one and is spliced, will spelled
Matching segment after connecing is determined as current matching segment, whether returns to the upper matching segment for judging current matching segment
The step of meeting preset second splicing condition；
If being unsatisfactory for preset second splicing condition, current matching segment is determined as target video segment.
12. according to the method for claim 11, it is characterised in that：
The preset first splicing condition is：If current matching segment is less than or equal to pre- with next time difference for matching segment
If first splicing threshold value and next matching segment be confirmed as head frame or piece tail frame institute in the video to be detected
Ratio value >=max (the α ss accounted fori_best, sst), it is determined that current clip is spliced with next segment that matches；
The preset second splicing condition is：If the time difference that current matching segment matches segment with upper one is less than or equal to pre-
If second splicing threshold value and it is described it is upper one match segment be confirmed as head frame or piece tail frame institute in the video to be detected
Ratio value >=max (the α ss accounted fori_best, sst), it is determined that segment is matched with described upper one to output segment and is spliced；
Wherein α is goal-selling threshold value, ssi_bestFor the head frame being confirmed as in the video to be detected or piece tail frame institute
Ratio shared by the head frame being confirmed as in the video to be detected or piece tail frame of the maximum matching segment of the ratio value accounted for
Example value, sstFor preset ratio threshold value.
13. according to the method for claim 11, which is characterized in that the method also includes：
If calculated maximum ratio value be not more than preset ratio threshold value, and in video to be detected current matching segment include to
The head frame in video is detected, then first matching segment is determined as target video segment；
If calculated maximum ratio value be not more than preset ratio threshold value, and in video to be detected current matching segment include to
The piece tail frame in video is detected, then the last one matching segment is determined as target video segment.
14. a kind of detection device of video frame, which is characterized in that including：
First obtains module, for obtaining video frame to be detected；
Second obtains module, for obtaining the Hash feature of the video frame to be detected；
Matching module, for by pre-stored each Hash feature in the Hash feature of the video frame to be detected and database
Sample carries out characteristic matching respectively；The Hash feature samples stored in the database are a kind of pre-stored head of style
And/or each Hash feature samples of run-out；
Matching result determining module, if for pre-stored Hash in the Hash feature and database of the video frame to be detected
A Hash feature samples characteristic matching in feature samples is consistent, and the video frame to be detected is determined as in video to be detected
Head frame or piece tail frame.
15. device according to claim 14, it is characterised in that：The Hash feature samples, including：Each sample view
The perceived hash characteristics of frequency frame and average Hash feature；
Described second obtains module, is specifically used for：Perceived hash characteristics and the average Hash for calculating the video frame to be detected are special
Sign；
The matching module, including：First distance computing unit and second distance computing unit；
The first distance computing unit, for will be deposited in advance in the perceived hash characteristics of the video frame to be detected and database
A kind of head of style and/or each perceived hash characteristics sample of run-out of storage carry out first distance calculating respectively；
The second distance computing unit, for depositing the average Hash feature of the video frame to be detected in advance with database
A kind of head of style of storage and/or each average Hash feature samples of run-out carry out second distance calculating respectively；
The matching result determining module, is specifically used for：If in the Hash feature of the video frame to be detected and database in advance
The first distance calculated result of Hash feature samples feature in the Hash feature samples of storage is less than perceptual hash threshold value, and the
Two are less than average Hash threshold value apart from calculated result, it is determined that in the Hash feature and database of the video frame to be detected in advance
The Hash feature samples characteristic matching of storage is consistent, so that the video frame successful match to be detected；
Alternatively, if each of the Hash feature of the video frame to be detected and Hash feature samples pre-stored in database
Hash feature samples feature matches inconsistent, then it fails to match for the video frame to be detected.
16. device according to claim 14, which is characterized in that further include selecting module, for it is described will be described to be checked
It surveys after the step of video frame is determined as the head frame in video to be detected or piece tail frame,
It selects the video frame not detected as video frame to be detected, returns to the Hash for obtaining the video frame to be detected
The step of feature.
17. device according to claim 16, which is characterized in that described first obtains module, is specifically used for：It will be to be detected
All video frames of video are never detected in video frame by the sequence of broadcasting as video frame is not detected and obtain a frame as to be checked
Survey video frame.
18. device according to claim 16, which is characterized in that described first obtains module, including：First detection unit
And second detection unit；
The first detection unit, including：First range determines that subelement and the first video frame to be detected determine subelement；
First range determines subelement, is used for when carrying out head detection to video to be detected, by the video to be detected
Video frame in first preset duration of middle starting is determined as the first detection range；
First video frame to be detected determines subelement, for extracting key frame out of described first detection range, is determined as
Video frame is not detected for detecting the first of head in the video to be detected, does not detect video frame from first by extraction sequence
One frame of middle acquisition is as the first video frame to be detected；
The matching module is used for when carrying out head detection to video to be detected, and the Hash of the first video frame to be detected is special
Sign carries out characteristic matching with pre-stored each head Hash feature samples in database respectively；
The selecting module, for video to be detected carry out head detection when, by broadcasting sequential selection one first not
Video frame is detected as the first video frame to be detected；
The second detection unit, including：Second range determines that subelement and the second video frame to be detected determine subelement；
Second range determines subelement, is used for when carrying out run-out detection to video to be detected, by the video to be detected
The video frame in the second preset duration before middle end is determined as the second detection range；
Second video frame to be detected determines subelement, for extracting key frame out of described second detection range, is determined as
Video frame is not detected for detecting the second of run-out in the video to be detected, does not detect video frame from second by extraction sequence
One frame of middle acquisition is as the second video frame to be detected；
The matching module is used for when carrying out run-out detection to video to be detected, and the Hash of the second video frame to be detected is special
Sign carries out characteristic matching with pre-stored each run-out Hash feature samples in database respectively；
The selecting module, for video to be detected carry out run-out detection when, by broadcasting sequential selection one second not
Video frame is detected as the second video frame to be detected.
19. device according to claim 18, which is characterized in that
First video frame to be detected determines subelement, is specifically used for：By the first preset interval, from first detection range
Interior extraction multiple first does not detect video frame；
Second video frame to be detected determines subelement, is specifically used for：By the second preset interval, from second detection range
Interior extraction multiple second does not detect video frame.
20. device according to claim 18, it is characterised in that：
Described device further includes：
First judges matching result module, for not detecting video by sequential selection one first of broadcasting in the selecting module
Before frame is as the first video frame to be detected,
Judge matching result that upper one has carried out the matched first video frame to be detected whether with the first of current matching to
The matching result for detecting video frame is identical；
If it is not the same, then having carried out the first to be detected of the matched first video frame to be detected and current matching for upper one
The video frame not detected between video frame is determined as first and does not detect video frame；It executes described by the sequential selection played one the
One does not detect the step of video frame is as the first video frame to be detected；
Alternatively, if identical, execute that described by sequential selection one first of broadcasting not detect video frame to be checked as first
The step of surveying video frame；
Described device further includes：
Second judges matching result module, for not detecting video by sequential selection one second of broadcasting in the selecting module
Before frame is as the second video frame to be detected,
Judge matching result that upper one has carried out the matched second video frame to be detected whether with the first of current matching to
The matching result for detecting video frame is identical；
If it is not the same, then having carried out the second to be detected of the matched second video frame to be detected and current matching for upper one
The video frame not detected between video frame is determined as second and does not detect video frame；It executes described by the sequential selection played one the
Two do not detect the step of video frame is as the second video frame to be detected；
Alternatively, if identical, execute that described by sequential selection one second of broadcasting not detect video frame to be checked as second
The step of surveying video frame.
21. 4~20 described in any item devices according to claim 1, it is characterised in that：Described device further includes Fusion Module；
The Fusion Module is obtained for merging to the head frame or piece tail frame that are confirmed as in the video to be detected
Obtain target video segment.
22. device according to claim 21, it is characterised in that：
The Fusion Module, including：Temporal information acquiring unit, matching sub-piece determination unit, time of fusion difference judging unit,
Match segment composition unit and target video Piece Selection unit；
Temporal information acquiring unit, for obtain the head frame or piece tail frame that are each confirmed as in the video to be detected when
Between information；
Sub-piece determination unit is matched, for the video frame of Time Continuous to be determined as a matching sub-piece；
Time of fusion difference judging unit, for judging it is default whether the time difference between adjacent matching sub-piece is less than or equal to
Time of fusion difference threshold value；
Segment composition unit is matched, for the time difference to be less than or equal to the neighbor sub-pieces of preset time of fusion difference threshold value
Section permeates a matching segment；
Target video Piece Selection unit, for selected from fused matching segment one as target video segment.
23. device according to claim 22, it is characterised in that：
The target video Piece Selection unit, is specifically used for：By first or the last one matching segment is determined as target view
Frequency segment.
24. device according to claim 22, it is characterised in that：
The target video Piece Selection unit, including：Ratio value computation subunit, current matching segment determine subelement,
One splicing condition judgment sub-unit, the first splicing subelement, first object video clip determine that subelement, the second splicing condition are sentenced
Disconnected subelement, the second splicing subelement and the second target video segment determine subelement；
Ratio value computation subunit, for calculating in each matching segment, the head frame being confirmed as in the video to be detected
Or ratio value shared by piece tail frame；
Current matching segment determines subelement, for the ratio value is maximum, and is greater than the matching segment of preset ratio threshold value
It is determined as current matching segment；
First splicing condition judgment sub-unit, if including the piece in video to be detected for current matching segment in video to be detected
Head frame, then judge whether next matching segment of current matching segment meets preset first splicing condition；
First splicing subelement, if for meeting preset first splicing condition, by current matching segment and next
Spliced with segment, spliced matching segment is determined as current matching segment, returns to the judgement current matching segment
Next matching segment whether meet it is preset first splicing condition the step of；
First object video clip determines subelement, if for being unsatisfactory for preset first splicing condition, by current matching
Segment is determined as target video segment；
Second splicing condition judgment sub-unit, if including the piece in video to be detected for current matching segment in video to be detected
Tail frame, then judge whether the upper matching segment of current matching segment meets preset second splicing condition；
Second splicing subelement, if for meeting preset second splicing condition, by current matching segment and upper one
Spliced with segment, spliced matching segment is determined as current matching segment, returns to the judgement current matching segment
Upper matching segment whether meet it is preset second splicing condition the step of；
Second target video segment determines subelement, if for being unsatisfactory for preset second splicing condition, by current matching
Segment is determined as target video segment.
25. device according to claim 24, it is characterised in that：
The preset first splicing condition is：If current matching segment is less than or equal to pre- with next time difference for matching segment
If first splicing threshold value and next matching segment be confirmed as head frame or piece tail frame institute in the video to be detected
Ratio value >=max (the α ss accounted fori_best, sst), it is determined that current clip is spliced with next segment that matches；
The preset second splicing condition is：If the time difference that current matching segment matches segment with upper one is less than or equal to pre-
If second splicing threshold value and it is described it is upper one match segment be confirmed as head frame or piece tail frame institute in the video to be detected
Ratio value >=max (the α ss accounted fori_best, sst), it is determined that segment is matched with described upper one to current clip and is spliced；
Wherein α is goal-selling threshold value, ssi_bestFor the head frame being confirmed as in the video to be detected or piece tail frame institute
Ratio shared by the head frame being confirmed as in the video to be detected or piece tail frame of the maximum matching segment of the ratio value accounted for
Example value, sstFor preset ratio threshold value.
26. device according to claim 24, which is characterized in that the determining current matching segment subelement is also used to：
If calculated maximum ratio value be not more than preset ratio threshold value, and in video to be detected current matching segment include to
The head frame in video is detected, then first matching segment is determined as target video segment；
If calculated maximum ratio value be not more than preset ratio threshold value, and in video to be detected current matching segment include to
The piece tail frame in video is detected, then the last one matching segment is determined as target video segment.
27. a kind of electronic equipment, which is characterized in that including processor, communication interface, memory and communication bus, wherein processing
Device, communication interface, memory complete mutual communication by communication bus；
Memory, for storing computer program；
Processor when for executing the program stored on memory, realizes any method and step of claim 1-13.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810638981.0A CN108924586B (en) | 2018-06-20 | 2018-06-20 | Video frame detection method and device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810638981.0A CN108924586B (en) | 2018-06-20 | 2018-06-20 | Video frame detection method and device and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108924586A true CN108924586A (en) | 2018-11-30 |
| CN108924586B CN108924586B (en) | 2021-01-08 |
Family
ID=64420706
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810638981.0A Active CN108924586B (en) | 2018-06-20 | 2018-06-20 | Video frame detection method and device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108924586B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110290419A (en) * | 2019-06-25 | 2019-09-27 | 北京奇艺世纪科技有限公司 | Video broadcasting method, device and electronic equipment |
| CN110324657A (en) * | 2019-05-29 | 2019-10-11 | 北京奇艺世纪科技有限公司 | Model generation, method for processing video frequency, device, electronic equipment and storage medium |
| CN110505495A (en) * | 2019-08-23 | 2019-11-26 | 北京达佳互联信息技术有限公司 | Multimedia resource takes out frame method, device, server and storage medium |
| CN110855904A (en) * | 2019-11-26 | 2020-02-28 | Oppo广东移动通信有限公司 | Video processing method, electronic device and storage medium |
| CN111027419A (en) * | 2019-11-22 | 2020-04-17 | 腾讯科技（深圳）有限公司 | Method, device, equipment and medium for detecting video irrelevant content |
| CN111291610A (en) * | 2019-12-12 | 2020-06-16 | 深信服科技股份有限公司 | Video detection method, device, equipment and computer readable storage medium |
| CN111356015A (en) * | 2020-02-25 | 2020-06-30 | 北京奇艺世纪科技有限公司 | Duplicate video detection method and device, computer equipment and storage medium |
| CN111479130A (en) * | 2020-04-02 | 2020-07-31 | 腾讯科技（深圳）有限公司 | Video positioning method and device, electronic equipment and storage medium |
| CN112040287A (en) * | 2020-08-31 | 2020-12-04 | 聚好看科技股份有限公司 | Display device and video playing method |
| CN112291589A (en) * | 2020-10-29 | 2021-01-29 | 腾讯科技（深圳）有限公司 | Video file structure detection method and device |
| CN113382283A (en) * | 2020-03-09 | 2021-09-10 | 上海哔哩哔哩科技有限公司 | Video title identification method and system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102323948A (en) * | 2011-09-07 | 2012-01-18 | 上海大学 | Automatic detection method for title sequence and tail leader of TV play video |
| US20130114745A1 (en) * | 2006-05-23 | 2013-05-09 | Lg Electronics Inc. | Digital television transmitting system and receiving system and method of processing broadcast data |
| CN103152632A (en) * | 2013-03-05 | 2013-06-12 | 天脉聚源(北京)传媒科技有限公司 | Method and device for locating multimedia program |
| CN105554514A (en) * | 2015-12-09 | 2016-05-04 | 福建天晴数码有限公司 | Method and system for processing opening songs of videos |
| CN106101573A (en) * | 2016-06-24 | 2016-11-09 | 中译语通科技（北京）有限公司 | The grappling of a kind of video labeling and matching process |
| CN107135401A (en) * | 2017-03-31 | 2017-09-05 | 北京奇艺世纪科技有限公司 | Key frame extraction method and system |
| CN107133266A (en) * | 2017-03-31 | 2017-09-05 | 北京奇艺世纪科技有限公司 | The detection method and device and database update method and device of video lens classification |
-
2018
- 2018-06-20 CN CN201810638981.0A patent/CN108924586B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130114745A1 (en) * | 2006-05-23 | 2013-05-09 | Lg Electronics Inc. | Digital television transmitting system and receiving system and method of processing broadcast data |
| CN102323948A (en) * | 2011-09-07 | 2012-01-18 | 上海大学 | Automatic detection method for title sequence and tail leader of TV play video |
| CN103152632A (en) * | 2013-03-05 | 2013-06-12 | 天脉聚源(北京)传媒科技有限公司 | Method and device for locating multimedia program |
| CN105554514A (en) * | 2015-12-09 | 2016-05-04 | 福建天晴数码有限公司 | Method and system for processing opening songs of videos |
| CN106101573A (en) * | 2016-06-24 | 2016-11-09 | 中译语通科技（北京）有限公司 | The grappling of a kind of video labeling and matching process |
| CN107135401A (en) * | 2017-03-31 | 2017-09-05 | 北京奇艺世纪科技有限公司 | Key frame extraction method and system |
| CN107133266A (en) * | 2017-03-31 | 2017-09-05 | 北京奇艺世纪科技有限公司 | The detection method and device and database update method and device of video lens classification |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110324657A (en) * | 2019-05-29 | 2019-10-11 | 北京奇艺世纪科技有限公司 | Model generation, method for processing video frequency, device, electronic equipment and storage medium |
| CN110290419B (en) * | 2019-06-25 | 2021-11-26 | 北京奇艺世纪科技有限公司 | Video playing method and device and electronic equipment |
| CN110290419A (en) * | 2019-06-25 | 2019-09-27 | 北京奇艺世纪科技有限公司 | Video broadcasting method, device and electronic equipment |
| CN110505495A (en) * | 2019-08-23 | 2019-11-26 | 北京达佳互联信息技术有限公司 | Multimedia resource takes out frame method, device, server and storage medium |
| CN110505495B (en) * | 2019-08-23 | 2021-12-07 | 北京达佳互联信息技术有限公司 | Multimedia resource frame extraction method, device, server and storage medium |
| CN111027419A (en) * | 2019-11-22 | 2020-04-17 | 腾讯科技（深圳）有限公司 | Method, device, equipment and medium for detecting video irrelevant content |
| CN111027419B (en) * | 2019-11-22 | 2023-10-20 | 腾讯科技（深圳）有限公司 | Method, device, equipment and medium for detecting video irrelevant content |
| CN110855904B (en) * | 2019-11-26 | 2021-10-01 | Oppo广东移动通信有限公司 | Video processing method, electronic device and storage medium |
| CN110855904A (en) * | 2019-11-26 | 2020-02-28 | Oppo广东移动通信有限公司 | Video processing method, electronic device and storage medium |
| CN111291610A (en) * | 2019-12-12 | 2020-06-16 | 深信服科技股份有限公司 | Video detection method, device, equipment and computer readable storage medium |
| CN111291610B (en) * | 2019-12-12 | 2024-05-28 | 深信服科技股份有限公司 | Video detection method, device, equipment and computer readable storage medium |
| CN111356015A (en) * | 2020-02-25 | 2020-06-30 | 北京奇艺世纪科技有限公司 | Duplicate video detection method and device, computer equipment and storage medium |
| CN111356015B (en) * | 2020-02-25 | 2022-05-10 | 北京奇艺世纪科技有限公司 | Duplicate video detection method and device, computer equipment and storage medium |
| CN113382283A (en) * | 2020-03-09 | 2021-09-10 | 上海哔哩哔哩科技有限公司 | Video title identification method and system |
| CN113382283B (en) * | 2020-03-09 | 2023-07-04 | 上海哔哩哔哩科技有限公司 | Video title identification method and system |
| CN111479130B (en) * | 2020-04-02 | 2023-09-26 | 腾讯科技（深圳）有限公司 | Video positioning method and device, electronic equipment and storage medium |
| CN111479130A (en) * | 2020-04-02 | 2020-07-31 | 腾讯科技（深圳）有限公司 | Video positioning method and device, electronic equipment and storage medium |
| CN112040287A (en) * | 2020-08-31 | 2020-12-04 | 聚好看科技股份有限公司 | Display device and video playing method |
| CN112291589A (en) * | 2020-10-29 | 2021-01-29 | 腾讯科技（深圳）有限公司 | Video file structure detection method and device |
| CN112291589B (en) * | 2020-10-29 | 2023-09-22 | 腾讯科技（深圳）有限公司 | Method and device for detecting structure of video file |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108924586B (en) | 2021-01-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108924586A (en) | A kind of detection method of video frame, device and electronic equipment | |
| CN112565825B (en) | Video data processing method, device, equipment and medium | |
| CN109803180B (en) | Video preview generation method and device, computer equipment and storage medium | |
| US8326087B2 (en) | Synchronizing image sequences | |
| KR20190116199A (en) | Video data processing method, device and readable storage medium | |
| CN109034159A (en) | image information extracting method and device | |
| CN109121022B (en) | Method and apparatus for marking video segments | |
| CN113691836B (en) | Video template generation method, video generation method and device and electronic equipment | |
| CN113542777B (en) | Live video editing method and device and computer equipment | |
| CN110309060B (en) | Detection method and device for updating identification algorithm, storage medium and computer equipment | |
| CN108519991A (en) | A kind of method and apparatus of main broadcaster's account recommendation | |
| CN109743589B (en) | Article generation method and device | |
| CN109409364A (en) | Image labeling method and device | |
| CN109408367A (en) | A kind of method and terminal of the control element identifying interactive interface | |
| CN108062341A (en) | The automatic marking method and device of data | |
| CN108197030A (en) | Software interface based on deep learning tests cloud platform device and test method automatically | |
| CN113407773A (en) | Short video intelligent recommendation method and system, electronic device and storage medium | |
| CN113852832A (en) | Video processing method, device, equipment and storage medium | |
| CN111340015B (en) | Positioning method and device | |
| CN109409321A (en) | A kind of determination method and device of camera motion mode | |
| CN113989476A (en) | Object identification method and electronic equipment | |
| CN113515998A (en) | Video data processing method and device and readable storage medium | |
| CN113515997A (en) | Video data processing method and device and readable storage medium | |
| US20140286624A1 (en) | Method and apparatus for personalized media editing | |
| CN113934888A (en) | Video tag processing method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | 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 |