CN108921876A - Method for processing video frequency, device and system and storage medium - Google Patents
Method for processing video frequency, device and system and storage medium Download PDFInfo
- Publication number
- CN108921876A CN108921876A CN201810753529.9A CN201810753529A CN108921876A CN 108921876 A CN108921876 A CN 108921876A CN 201810753529 A CN201810753529 A CN 201810753529A CN 108921876 A CN108921876 A CN 108921876A
- Authority
- CN
- China
- Prior art keywords
- target image
- image block
- pursuit path
- target
- video
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30241—Trajectory
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the present invention provides a kind of method for processing video frequency, device and system and storage medium.Method for processing video frequency includes：Obtain video flowing；Target detection and tracking are carried out to video flowing, to determine at least one pursuit path；For each of at least one pursuit path, the target image block comprising target corresponding to the pursuit path is extracted, respectively from at least partly video frame of the pursuit path to obtain at least one target image block；For each of at least one pursuit path, at least partly target image block at least one target image block is clustered, to obtain at least one cluster centre；For each of at least one pursuit path, target image block corresponding at least one cluster centre is exported.The information that target leaves under pursuit path can be pushed to back-end server redundantly as much as possible by the above method and device, so as to effectively improve the speed and effect that capture machine pushes away figure.
Description
Technical field
The present invention relates to field of image processing, relates more specifically to a kind of method for processing video frequency, device and system and deposit
Storage media.
Background technique
Capture machine is that one kind comes out target detection from video flowing, then takes out by image block where target, then
It is transmitted to the intelligent camera front end of back-end server.Currently, a typical case of capture machine is face snap, i.e., from video flowing
In detect face, and the image block (facial image) comprising these faces is pushed to back-end server.Common capture machine
Pushing away drawing method, there are mainly two types of：1. all being pushed for the facial image under the same pursuit path；2. each tracking rail
Mark only pushes a clearest facial image.
Both the above mode all existing defects.Firstly, all push will cause data redundancy, to increase back-end server
Processing pressure；Secondly, only pushing clearest image can not be characterized in what the face under current pursuit path left comprehensively
Information.Therefore, existing capture machine, which is difficult to, is not pushed to back-end server redundantly for all face informations as much as possible.
Summary of the invention
The present invention is proposed in view of the above problem.The present invention provides a kind of method for processing video frequency, device and system
And storage medium.
According to an aspect of the present invention, a kind of method for processing video frequency is provided.Method for processing video frequency includes：Obtain video flowing；
Target detection and tracking are carried out to video flowing, to determine at least one pursuit path；For every at least one pursuit path
One, extract the target image comprising target corresponding to the pursuit path respectively from at least partly video frame of the pursuit path
Block, to obtain at least one target image block；For each of at least one pursuit path, at least one target image
At least partly target image block in block is clustered, to obtain at least one cluster centre；For at least one pursuit path
Each of, export target image block corresponding at least one cluster centre.
Illustratively, at least partly target image block at least one target image block is clustered, to obtain extremely
A cluster centre includes less：Extract the feature vector of each target image block at least partly target image block；And it is right
The feature vector of at least partly target image block is clustered, to obtain at least one cluster centre, wherein at least one cluster
Each of center is the feature vector of at least partly one of target image block.
Illustratively, method further includes：For each of at least one pursuit path, at least to the pursuit path
One target image block carries out quality evaluation respectively, to obtain the quality score of at least one target image block；And according to extremely
The quality score of a few target image block, selects at least partly target image block from least one target image block.
Illustratively, it according to the quality score of at least one target image block, is selected from least one target image block
At least partly target image block includes：Quality score is selected to be higher than the target figure of predetermined threshold from least one target image block
As block, or the target image block of the highest predetermined number of quality score is selected from least one target image block, to obtain
At least partly target image block.
Illustratively, it according to the quality score of at least one target image block, is selected from least one target image block
At least partly target image block includes：If there are the targets that quality score is higher than predetermined threshold at least one target image block
Image block then selects quality score to be higher than the target image block of predetermined threshold from least one target image block, to obtain extremely
Small part target image block；If there is no the target images that quality score is higher than predetermined threshold at least one target image block
Block then selects the target image block of the highest predetermined number of quality score from least one target image block, to obtain at least
Partial target image block.
Illustratively, quality score is obtained based on the sharpness computation of target image block.
Illustratively, at least partly target image block at least one target image block is clustered, to obtain extremely
The step of lacking a cluster centre and exporting target image block corresponding at least one cluster centre is in the pursuit path
Length executes in the case where being greater than length threshold, and method further includes：For each of at least one pursuit path, if should
The length of pursuit path is less than or equal to length threshold, then exports at least one target image block of the pursuit path, Huo Zhecong
At least one target image block of the pursuit path selects one or more target image blocks, and exports selected target image
Block.
Illustratively, one or more target image block packets are selected from least one target image block of the pursuit path
It includes：One or more target image blocks are randomly choosed from least one target image block of the pursuit path.
Illustratively, one or more target image block packets are selected from least one target image block of the pursuit path
It includes：Quality evaluation is carried out respectively at least one target image block of the pursuit path, to obtain at least one target image block
Quality score；And the quality score according at least one target image block, one is selected from least one target image block
A or multiple target image blocks.
Illustratively, it is extracted respectively in at least partly video frame from the pursuit path comprising corresponding to the pursuit path
The target image block of target, before obtaining at least one target image block, method further includes：To all views of the pursuit path
Frequency frame carries out quality evaluation respectively, to obtain the quality score of all video frames of the pursuit path；And according to the tracking rail
The quality score of all video frames of mark selects at least partly video frame from all video frames of the pursuit path.
According to a further aspect of the invention, a kind of video process apparatus is provided, including：Module is obtained, for obtaining video
Stream；Detection and tracking module, for carrying out target detection and tracking to video flowing, to determine at least one pursuit path；It extracts
Module, for being mentioned respectively from at least partly video frame of the pursuit path for each of at least one pursuit path
The target image block comprising target corresponding to the pursuit path is taken, to obtain at least one target image block；Cluster module is used for
For each of at least one pursuit path, at least partly target image block at least one target image block is carried out
Cluster, to obtain at least one cluster centre；Output module, for for each of at least one pursuit path, output
Target image block corresponding at least one cluster centre.
According to a further aspect of the invention, a kind of processing system for video, including processor and memory are provided, wherein institute
It states and is stored with computer program instructions in memory, for executing when the computer program instructions are run by the processor
State method for processing video frequency.
According to a further aspect of the invention, a kind of storage medium is provided, stores program instruction on said storage,
Described program instruction is at runtime for executing above-mentioned method for processing video frequency.
Method for processing video frequency, device and system and storage medium according to an embodiment of the present invention, can service to the back-end
Device pushes several most representative target image blocks, what this mode can as much as possible leave target under pursuit path
Information is not pushed to back-end server redundantly, so as to effectively improve the speed and effect that capture machine pushes away figure.
Detailed description of the invention
The embodiment of the present invention is described in more detail in conjunction with the accompanying drawings, the above and other purposes of the present invention,
Feature and advantage will be apparent.Attached drawing is used to provide to further understand the embodiment of the present invention, and constitutes explanation
A part of book, is used to explain the present invention together with the embodiment of the present invention, is not construed as limiting the invention.In the accompanying drawings,
Identical reference label typically represents same parts or step.
Fig. 1 shows showing for the exemplary electronic device for realizing method for processing video frequency according to an embodiment of the present invention and device
Meaning property block diagram；
Fig. 2 shows the schematic flow charts of method for processing video frequency according to an embodiment of the invention；
Fig. 3 shows the schematic diagram of the video process flow of capture machine according to an embodiment of the invention；
Fig. 4 shows the schematic block diagram of video process apparatus according to an embodiment of the invention；And
Fig. 5 shows the schematic block diagram of processing system for video according to an embodiment of the invention.
Specific embodiment
In order to enable the object, technical solutions and advantages of the present invention become apparent, root is described in detail below with reference to accompanying drawings
According to example embodiments of the present invention.Obviously, described embodiment is only a part of the embodiments of the present invention, rather than this hair
Bright whole embodiments, it should be appreciated that the present invention is not limited by example embodiment described herein.
To solve the above-mentioned problems, the embodiment of the present invention provides a kind of new method for processing video frequency and device, this method and
Device can several most representative target image blocks of rear end server push, to describe a target as completely as possible
The information left under a pursuit path, this method and device can effectively improve the speed and effect that capture machine pushes away figure.
Method for processing video frequency and device according to an embodiment of the present invention can be applied to the various fields for being related to target and capturing, such as electronics
The face snap in the fields such as commercial affairs, banking, security monitoring, vehicle or the license plate candid photograph in traffic monitoring field, etc..
Firstly, describing the example for realizing method for processing video frequency according to an embodiment of the present invention and device referring to Fig.1
Electronic equipment 100.
As shown in Figure 1, electronic equipment 100 includes one or more processors 102, one or more storage devices 104.It can
Selection of land, electronic equipment 100 can also include input unit 106, output device 108 and image collecting device 110, these groups
Part passes through the interconnection of bindiny mechanism's (not shown) of bus system 112 and/or other forms.It should be noted that electronics shown in FIG. 1 is set
Standby 100 component and structure be it is illustrative, and not restrictive, as needed, the electronic equipment also can have it
His component and structure.
The processor 102 can use microprocessor, digital signal processor (DSP), field programmable gate array
(FPGA), at least one of programmable logic array (PLA) example, in hardware realizes that the processor 102 can be center
Processing unit (CPU), image processor (GPU), dedicated integrated circuit (ASIC) have data-handling capacity and/or refer to
The combination of one or more of processing unit of other forms of executive capability is enabled, and can control the electronic equipment
Other components in 100 are to execute desired function.
The storage device 104 may include one or more computer program products, and the computer program product can
To include various forms of computer readable storage mediums, such as volatile memory and/or nonvolatile memory.It is described easy
The property lost memory for example may include random access memory (RAM) and/or cache memory (cache) etc..It is described non-
Volatile memory for example may include read-only memory (ROM), hard disk, flash memory etc..In the computer readable storage medium
On can store one or more computer program instructions, processor 102 can run described program instruction, to realize hereafter institute
The client functionality (realized by processor) in the embodiment of the present invention stated and/or other desired functions.In the meter
Can also store various application programs and various data in calculation machine readable storage medium storing program for executing, for example, the application program use and/or
The various data etc. generated.
The input unit 106 can be the device that user is used to input instruction, and may include keyboard, mouse, wheat
One or more of gram wind and touch screen etc..
The output device 108 can export various information (such as image and/or sound) to external (such as user), and
It and may include one or more of display, loudspeaker etc..Optionally, the input unit 106 and the output device
108 can integrate together, be realized using same interactive device (such as touch screen).
Described image acquisition device 110 can acquire image (including still image and video frame), such as the face of user
Image etc., and acquired image is stored in the storage device 104 for the use of other components.Image collecting device
110 can be individual camera, the imaging sensor in camera or capture machine in mobile terminal.It should be appreciated that image is adopted
Acquisition means 110 are only examples, and electronic equipment 100 can not include image collecting device 110.In such a case, it is possible to utilize
Other devices with Image Acquisition ability acquire image, and the image of acquisition is sent to electronic equipment 100.
Illustratively, the exemplary electronic device for realizing method for processing video frequency according to an embodiment of the present invention and device can
To be realized in the equipment of personal computer or remote server etc..
In the following, method for processing video frequency according to an embodiment of the present invention will be described with reference to Fig. 2.Fig. 2 shows according to the present invention one
The schematic flow chart of the method for processing video frequency 200 of a embodiment.As shown in Fig. 2, method for processing video frequency 200 includes the following steps
S210, S220, S230, S240 and S250.
In step S210, video flowing is obtained.
The video flowing that step S210 is obtained can be image collecting device (such as imaging sensor of capture machine) and collect
Original video stream, be also possible to preprocessed video stream obtained after pre-processing to original video stream.The pretreatment
It may include for the processing such as scaling, denoising of video frame.
The length for the video flowing that step S210 is obtained can be arbitrary.Step S210 obtain video flowing may include to
A few video frame.Optionally, in step S210, video flowing can be obtained in real time, i.e., obtain each current video frame in real time.It can
Selection of land can obtain entire video flowing in step S210 simultaneously.
In step S220, target detection and tracking are carried out to video flowing, to determine at least one pursuit path.
Target as described herein can be any object, including but not limited to：Text, specific pattern, people or human body one
Partially (face), animal, vehicle, building etc..Hereinafter, by mainly by taking target is face as an example it is each herein to describe
Embodiment, but this is not limitation of the present invention.
Fig. 3 shows the schematic diagram of the video process flow of capture machine according to an embodiment of the invention.As shown in figure 3,
The imaging sensor of capture machine can acquire video flowing in real time, and each video frame in video flowing is passed to capture machine in real time
Main control module.After main control module receives current video frame, current video frame can be pre-processed, after being pre-processed
Video frame.Pretreated video frame is sent into detector module and carries out target detection (such as Face datection) by main control module.
After detector module receives current video frame, it can use algorithm of target detection and detect institute in current video frame
There is target, output is used to indicate the position of the bounding box (bounding box, may be simply referred to as bbox) of each target position
Information.Illustratively, bounding box can be rectangle frame.Illustratively, the location information of bounding box can be with four numerical value come table
Show.For example, the location information of bounding box can be shown with following numerical tabular：The upper left corner abscissa x of the bounding box, the vertical seat in the upper left corner
Mark the height h of y, the width w of bounding box, bounding box.In another example the location information of bounding box can use four tops of the bounding box
The coordinate representation of point.Detector module can export the location information (bbox information) of institute's bounding box of current video frame extremely
Tracker module.
The bbox information input tracker module that can be will test out, tracker module, can be with by target tracking algorism
Obtain at least one pursuit path relevant to the target in video flowing.For example, a large amount of band tracking marks can be obtained by tracking
Know the bounding box of symbol (track ID).
Specifically, it can determine which bounding box in two adjacent video frames belongs to same tracking rail by track algorithm
Mark, the bounding box of same target can distribute identical track ID.That is, each track ID can represent one with
Track track, such as the track ID of face A can be 1, the track ID of face B can be 2, and so on.Therefore, for every
A target can obtain a pursuit path.According to the track ID of each bounding box can know the bounding box belong to which with
Track track also can know which target the bounding box belongs to.
Tracker module can complete the link for belonging to the bounding box of same target between different video frame, until the target
It disappears in the visible range of capture machine.Each pursuit path can be understood as a target in the visible range of capture machine from
There is the motion profile left to during disappearing.Assuming that the same target (such as face A) first leaves candid photograph scene, then pass through
Appear in again after a period of time capture scene in, if the extinction time of the target is shorter, can obtain one with
Track track can obtain two sseparated pursuit paths of the target if the extinction time of the target is longer.
Illustratively, if a target occurs in several video frames before video flowing, but at subsequent one or more
It disappears in a video frame, then tracker module can speculate target rear according to motion information of the target in preceding several video frames
Position in continuous one or more video frames, that is to say, that even if target temporary extinction in the video frame, tracker module
It can continue to track it, until the number of video frame that the target continuously disappears is more than preset quantity threshold (such as 5
Frame) until.If the number for the video frame that target continuously disappears is more than preset quantity threshold, can stop to the target
Tracking, and bounding box corresponding to the video frame of target appearance is chained up, obtain the pursuit path of the target.If in mesh
Before the number for marking the video frame continuously to disappear is more than preset quantity threshold, target occurs again, then target can occur,
It disappears and occurs bounding box corresponding to the video frame in this whole process again and be chained up, obtain the tracking of the target
Track, wherein bounding box corresponding to the video frame that target disappears can be the bounding box that tracker module deduces.Above-mentioned mesh
Mark speculates that the mode of its position can reduce target missing inspection bring tracking mistake when disappearing.
It, can be using algorithm of target detection and target following any existing or be likely to occur in the future in step S220
Algorithm carries out target detection and tracking.
In step S230, for each of at least one pursuit path, from at least partly video of the pursuit path
The target image block comprising target corresponding to the pursuit path is extracted in frame, respectively to obtain at least one target image block.
In one example, for each pursuit path, can distinguish directly from all video frames of the pursuit path
The target image block comprising target corresponding to the pursuit path is extracted, to obtain at least one target image block.
In another example, for each pursuit path, the high video frame of quality can be selected from the pursuit path,
Target image block is extracted from the high video frame of quality again.Optionally, for each of at least one pursuit path, from
The target image block comprising target corresponding to the pursuit path is extracted in at least partly video frame of the pursuit path, respectively to obtain
Before obtaining at least one target image block (step S230), method for processing video frequency 200 can also include：For at least one tracking
Each of track carries out quality evaluation to all video frames of the pursuit path respectively, to obtain the institute of the pursuit path
There is the quality score of video frame；And the quality score of all video frames according to the pursuit path, from the institute of the pursuit path
Have and selects at least partly video frame in video frame.
For each pursuit path, quality evaluation can be carried out respectively to all video frames of the pursuit path first, with
The quality score of all video frames is obtained, and selects quality score to be higher than preset matter from all video frames of the pursuit path
The video frame of threshold value (video frame quality threshold) is measured, to obtain at least partly video frame of the pursuit path.It then, can be from institute
The target image block comprising target corresponding to the target trajectory is extracted in the video frame of selection, respectively to obtain at least one target
Image block.
Illustratively, the quality score of video frame can be obtained based on the sharpness computation of video frame.In this case,
Only consider that the clarity of video frame can therefrom extract target image block if some video frame is clear enough, if it is not
It is enough clear, then it can filter this out, no longer execution subsequent processing.
However, above-described embodiment is not limitation of the present invention, the quality of video frame can be by the way of any appropriate
It calculates and measures.For example, the quality evaluation of video frame may include to following one or more assessment：The image of video frame is clear
Clear degree, the clarity of target in video frame, the coverage extent of target in video frame, target in video frame angle, view
The size of target etc. in frequency frame.For example, can the image definition to video frame, the fuzzy journey of the face in video frame
Degree, the angle of face, face the parameters such as coverage extent carry out certain operations, such as be weighted and averaged, data obtained can be with
It is considered as the quality score of video frame.Furthermore, it is possible to which video frame input video frame quality evaluation network is commented using video frame quality
Valence network screens the higher video frame of mass.Video frame quality evaluation network can be real using any appropriate neural network
It is existing, such as convolutional neural networks etc..
In step S240, for each of at least one pursuit path, at least one target image block extremely
Small part target image block is clustered, to obtain at least one cluster centre.
In one example, for each pursuit path, can directly to all target image blocks of the pursuit path into
Row cluster, to obtain at least one cluster centre of the pursuit path.
In another example, for each pursuit path, the high mesh of quality can be selected first from the pursuit path
Logo image block, the target image block high to selected quality cluster, to obtain at least one cluster of the pursuit path
Center.Fig. 3 shows a kind of implementation of the present embodiment.As shown in figure 3, can be by all targets under the same pursuit path
Image block inputs a quality assessment module, which can screen and export the target image that quality is met the requirements
Block, i.e., filtered target image block.
Illustratively, for each of at least one pursuit path, at least one target image block at least
Partial target image block is clustered, and may include to obtain at least one cluster centre (i.e. step S240)：For at least one
Each of a pursuit path extracts the feature vector of each target image block at least partly target image block；And
For each of at least one pursuit path, the feature vector of at least partly target image block is clustered, to obtain
At least one cluster centre, wherein each of at least one cluster centre is the spy of at least partly one of target image block
Levy vector.
As shown in figure 3, filtered target image block input feature vector extraction module can be carried out feature extraction.
It, can be using any existing or feature extraction algorithm in the cards in future to each in characteristic extracting module
At least partly target image block of pursuit path carries out feature extraction, obtains the feature vector of each target image block.For example, can be with
Each target image block is inputted into trained feature extraction network, target image block is mapped to one by feature extraction network
Low-dimensional feature space, and obtain the low-dimensional feature vector exported by feature extraction network.It, can be with for each pursuit path
Obtain the corresponding set of eigenvectors of the pursuit path.Feature extraction network can use any appropriate neural fusion, example
Such as convolutional neural networks.
Extract feature about target image block using feature extraction network, can in a manner of a kind of tight Minato strong earth's surface
Levy the difference under the same pursuit path between different target image block.
Illustratively, clustering to set of eigenvectors can be realized using k-medoids clustering algorithm.For example, can be with
The corresponding set of eigenvectors of each pursuit path is sent into k-medoids cluster module.The module is based on k-medoids cluster and calculates
Method realizes the purpose concentrated from feature vector and pick out most representative feature vector.It is appreciated that cluster module output
Each cluster centre is a feature vector, and is feature vector (the i.e. at least partly target image inputted in cluster module
One of the feature vector of block).
K-medoids clustering algorithm is similar to k-means clustering algorithm.K-means clustering algorithm is empty in low-dimensional Euclidean
Between in find k cluster centre (or saying cluster centre point), it is desirable that the center to all data in corresponding cluster (cluster)
The Euclidean distance of point is most short, and cluster centre can be any one data point in low-dimensional Euclidean space.K-medoids cluster is calculated
The purpose of method is equally to find k cluster centre, but the value of each cluster centre is limited to from the data point of current cluster
It finds, rather than the arbitrary number strong point in space.Therefore, using k-medoids clustering algorithm, it is ensured that each of its return
The data of cluster centre can correctly correspond to a target image block.
By cluster, may be implemented to pick out information redundance from the target image block under the same pursuit path minimum
K target image block, the image block collection as composed by this k target image block can represent about the mesh under the pursuit path
The most information of target.
Based on the set of eigenvectors formed after feature extraction, completed using clustering algorithm to most representative containing information content
Target image block select, this working method can be completed to be determined not by human eye originally in a kind of method of quantization
The image block for being easy quantization selects work.
In step S250, for each of at least one pursuit path, export corresponding at least one cluster centre
Target image block.
The corresponding target image block of at least one cluster centre can be pushed to rear end by the modes such as wired or wireless
Server, so that background user such as checks or carry out image retrieval at the operation.
Method for processing video frequency according to an embodiment of the present invention at least has following advantage：
1, k cluster centre is obtained based on clustering algorithm, only releases the corresponding k image block of these cluster centres, these
Image block can reflect information content approximately uniform with all image blocks under pursuit path, therefore, under push pursuit path
The figure mode that pushes away of all image blocks is compared, and the use of information of initial data can be improved in the figure mode provided in an embodiment of the present invention that pushes away
Rate；
2, with only push clearest image block push away figure mode compared with, it is provided in an embodiment of the present invention push away figure mode can
To push under pursuit path most several (possible more than one) image blocks of information content, so as to provide richer letter
Breath is conducive to abundant user in the way of data and Potential feasibility；
3, the function of capture machine can be enhanced in method provided in an embodiment of the present invention, reduces back-end server garbled data
Pressure saves the expense of server end arrangement, so as to bring direct economic interests.
Illustratively, method for processing video frequency according to an embodiment of the present invention can be in setting with memory and processor
It is realized in standby, device or system.
Method for processing video frequency according to an embodiment of the present invention can be deployed at personal terminal, such as smart phone, plate
Computer, personal computer etc..
Alternatively, method for processing video frequency according to an embodiment of the present invention can also be deployed in server end and client with being distributed
At end.For example, can obtain video flowing (such as facial image in Image Acquisition end acquisition user) in client, client will
The image of acquisition sends server end (or cloud) to, carries out video processing by server end (or cloud).
According to embodiments of the present invention, method for processing video frequency 200 can also include：For every at least one pursuit path
One, quality evaluation is carried out respectively at least one target image block of the pursuit path, to obtain at least one target image
The quality score of block；And for each of at least one pursuit path, according to the quality of at least one target image block
Scoring, selects at least partly target image block from least one target image block.
Illustratively, the quality score of target image block can be obtained based on the sharpness computation of target image block.At this
In the case of kind, only consider that the clarity of target image block can participate in subsequent if some target image block is clear enough
The operation such as cluster, if it is not clear enough, can filter this out, no longer execution subsequent processing.Due to target image block
Clarity is affected to the efficiency and accuracy of subsequent processing, therefore the factor can be considered with emphasis.
However, above-described embodiment is not limitation of the present invention, the quality of target image block can use any appropriate
Mode is calculated and is measured.For example, the quality evaluation of target image block may include to following one or more assessment：Target figure
As the clarity of the target in block, the coverage extent of target, the angle of target, the size of target etc..For example, can be to target
The fog-level of face in image block, the angle of face, the parameters such as coverage extent of face carry out certain operations, such as weight
Average, data obtained can be considered as the quality score of target image block.Furthermore, it is possible to by target image block input picture block
Quality evaluation network screens the higher target image block of mass using image block quality evaluation network.Image block quality evaluation
Network can use any appropriate neural fusion, such as convolutional neural networks etc..
For each pursuit path, quality score can be selected high from least one target image block of the pursuit path
In the video frame of preset quality threshold (image block quality threshold), to obtain at least partly target image of the pursuit path
Block.
The high image block of screening quality in advance can mitigate the calculation amount of the operations such as subsequent cluster, improve clustering precision, into
And the quality for being pushed to the image block of back-end server can be improved, mitigate the processing pressure and storage pressure of server end.
It is real that the high embodiment of image block of screening quality and the embodiment of the high video frame of above-mentioned screening quality can select one
It is existing, that is, to can choose the screening for carrying out image before or after extracting image block, guarantee the figure of the operations such as subsequent cluster
Image quality amount.Certainly, the embodiment of the high video frame of the embodiment and screening quality of the high image block of screening quality can also be simultaneously
It realizes.
According to embodiments of the present invention, according to the quality score of at least one target image block, from least one target image
At least partly target image block is selected to include in block：Quality score is selected to be higher than predetermined threshold from least one target image block
Target image block, or from least one target image block select the highest predetermined number of quality score target image
Block, to obtain at least partly target image block.
Predetermined threshold and predetermined number can be set as needed, and be limited herein not to this.For example, it is assumed that a certain
One pursuit path of face has 100 facial image blocks, wherein the quality of only 5 facial image blocks is higher than predetermined threshold
(i.e. above-mentioned image block quality threshold), then can choose this 5 facial images.In another example a pursuit path of a certain face
With 100 facial image blocks, and predetermined number is 20, then can select quality score from this 100 facial images
Whether the preceding 20 facial image blocks of ranking, the quality score without regard to this 20 facial image blocks are higher than predetermined threshold
Value.
According to embodiments of the present invention, according to the quality score of at least one target image block, from least one target image
At least partly target image block is selected to may include in block：If there are quality scores to be higher than in advance at least one target image block
Determine the target image block of threshold value, then quality score is selected to be higher than the target image of predetermined threshold from least one target image block
Block, to obtain at least partly target image block；If there is no quality scores to be higher than predetermined threshold at least one target image block
The target image block of value then selects the target image of the highest predetermined number of quality score from least one target image block
Block, to obtain at least partly target image block.
In the present embodiment, it can preferentially select quality score can if not having higher than the target image block of predetermined threshold
To consider the target image block of the selection highest predetermined number of quality score.It can guarantee to participate in the behaviour such as subsequent cluster as far as possible in this way
The quality of the target image block of work can satisfy requirement, while can be to avoid entire pursuit path because without the enough high targets of quality
Image block and can not participate in it is subsequent cluster etc. operation.
According to embodiments of the present invention, at least partly target image block at least one target image block is clustered,
To obtain at least one cluster centre (i.e. step S240) and export target image block corresponding at least one cluster centre
The step of (i.e. step S250), executes in the case where the length of the pursuit path is greater than length threshold, method for processing video frequency 200
Can also include：For each of at least one pursuit path, if the length of the pursuit path is less than or equal to length
Threshold value then exports at least one target image block of the pursuit path, or at least one target image from the pursuit path
Block selects one or more target image blocks, and exports selected target image block.
The length of pursuit path refers to the time span of the pursuit path, and optionally, the length of pursuit path can be used should
The number for the video frame that pursuit path includes indicates.
It will be understood by those skilled in the art that between usually there is the smaller time between adjacent video frames in video flowing
Every variation of the target in two adjacent video frames may very little.If pursuit path is shorter, it includes video frame it is less,
Variation of the target possible in this way in entire pursuit path is all not too large.The video frame for including due to shorter pursuit path
It is few, therefore even if the target complete image block under the pursuit path is all pushed out, it will not be for back-end server increase too
Big processing pressure, therefore can choose whole push.In addition, target is in entirely tracking rail also due to pursuit path is shorter
Variation in mark is little, the information that less target image block can also leave than more fully characterizing target, therefore can also be with
Part (such as one) target image block under the pursuit path is selected to be pushed.
If pursuit path long enough, since data volume increases, while variation of the target under entire pursuit path
May be larger, the information content left may also be more, can choose the target image block to the pursuit path in this case
It is clustered, k target image block for selecting most information content is pushed.
The different sides for pushing away figure mode, target image block will be directly selected pushing are selected according to the length of pursuit path
Formula is combined in such a way that Clustering and selection target image block is pushed, and can obtain calculation amount, information content, back-end services
A kind of equilibrium of a variety of aspects such as device processing pressure pushes away figure efficiency and quality so as to further increase capture machine.
Aforesaid operations can carry out in main control module, main control module may determine that some pursuit path length whether
It, can be by main control module by target complete image block or selected section target figure no more than length threshold more than length threshold
As block is pushed to back-end server, more than length threshold, target image block can be sent to quality assessment module or feature
Extraction module performs corresponding processing.
Above-mentioned length threshold can be set as needed, and can be arbitrary value, is limited herein not to this.
According to embodiments of the present invention, for each of at least one pursuit path, from least the one of the pursuit path
A target image block selects one or more target image blocks to include：For each of at least one pursuit path, from this
One or more target image blocks are randomly choosed at least one target image block of pursuit path.
It illustratively, can be random from least one target image block of the pursuit path for each pursuit path
Select the target image block of preset number.Preset number can be set as needed, and can be arbitrary, such as 3.
According to embodiments of the present invention, for each of at least one pursuit path, from least the one of the pursuit path
A target image block selects one or more target image blocks to include：For each of at least one pursuit path, to this
At least one target image block of pursuit path carries out quality evaluation respectively, is commented with obtaining the quality of at least one target image block
Point；And for each of at least one pursuit path, according to the quality score of at least one target image block, from least
One or more target image blocks are selected in one target image block.
The mode for carrying out quality evaluation in the present embodiment to target image block carries out target image block with above-described
The mode of quality evaluation is similar, and those skilled in the art can understand the present embodiment with reference to above description, and details are not described herein.
The example according to an embodiment of the present invention for carrying out face snap using capture machine and push facial image is described below
Property process.
A. video acquisition is carried out firstly, capture machine is installed on some crowded channel.Each view of video flowing
Frequency frame can be passed to main control module in real time.Optionally, main control module can pre-process current video frame.
B. then, detector module is sent by pretreated video frame, detector module can will be in video frame
All faces detected, and obtain the bbox information of face.Detector module exports the bbox information of all faces to tracker
Module.
C. tracker module is after receiving the bbox information of the face in the video frame, if this face is to go out for the first time
It is existing, then tracker algorithm is initialized with the bbox information；If this face has already appeared in a upper video frame, tracker
Module can get up bbox information of the face in current video frame and the bbox Info Link in preceding several video frames, structure
At the pursuit path of the face.Based on this, tracker module can be carried out when missing inspection occurs for detector module according to track trend
The filling of coordinate information.
D. the information of pursuit path is sent back main control module by tracker module, and main control module can carry out pursuit path
Summarize, manage.
E. when a pursuit path completes (i.e. a face this mistake from occurring to disappearing in the visible range of capture machine
Journey terminates) after, the face images block under the pursuit path can be sent into image block quality evaluation network by main control module,
Quality evaluation to facial image block is completed by the network, filters out undesirable facial image block.
F. then, filtered facial image block is further fed into feature extraction network, by the network by facial image
Block is mapped to low-dimensional Euclidean space, realizes another characteristic manner to facial image block, it is possible thereby to empty with less storage
Between represent the different information between different faces image block.
G. the set of eigenvectors obtained after mapping feeding k-medoids cluster module is clustered, is obtained in k cluster
The heart, i.e. k feature vector.
H. k feature vector of acquisition is sent back into main control module, main control module can be selected according to this k feature vector
Its corresponding facial image block simultaneously pushes it to server end.
For each pursuit path, above step e-h is executed, i.e. step e-h can recycle execution.
Optionally, main control module described herein, detector module, tracker module, quality assessment module, feature extraction
Module, cluster module can be realized in same hardware device (processor 102 as shown in Figure 1).Optionally, it is described herein
Main control module, detector module, tracker module, quality assessment module, characteristic extracting module, any one in cluster module
Item can be realized in independent hardware device.
According to a further aspect of the invention, a kind of video process apparatus is provided.Fig. 4 is shown according to an embodiment of the present invention
Video process apparatus 400 schematic block diagram.
As shown in figure 4, video process apparatus 400 according to an embodiment of the present invention includes obtaining module 410, detection and tracking
Module 420, extraction module 430, cluster module 440 and output module 450.The modules can execute respectively above in conjunction with
The each step/function for the method for processing video frequency that Fig. 2-3 is described.Below only to the master of each component of the video process apparatus 400
It wants function to be described, and omits the detail content having been described above.
Module 410 is obtained for obtaining video flowing.Obtaining module 410 can processing in electronic equipment as shown in Figure 1
The program instruction that stores in 102 Running storage device 103 of device is realized.
Detection and tracking module 420 is used to carry out target detection and tracking to the video flowing, with determine at least one with
Track track.Detection and tracking module 420 can be in 102 Running storage device 103 of processor in electronic equipment as shown in Figure 1
The program instruction of storage is realized.
Extraction module 430 is used for for each of at least one described pursuit path, at least from the pursuit path
The target image block comprising target corresponding to the pursuit path is extracted in partial video frame, respectively to obtain at least one target figure
As block.The journey that extraction module 430 can store in 102 Running storage device 103 of processor in electronic equipment as shown in Figure 1
Sequence instructs to realize.
Cluster module 440 is used for for each of at least one described pursuit path, at least one described target
At least partly target image block in image block is clustered, to obtain at least one cluster centre.Cluster module 440 can be by
The program instruction that stores in 102 Running storage device 103 of processor in electronic equipment shown in FIG. 1 is realized.
Output module 450 is used for for each of at least one described pursuit path, and output is described, and at least one is poly-
Target image block corresponding to class center.Output module 450 can the processor 102 in electronic equipment as shown in Figure 1 run
The program instruction that stores in storage device 103 is realized.
Those of ordinary skill in the art may be aware that list described in conjunction with the examples disclosed in the embodiments of the present disclosure
Member and algorithm steps can be realized with the combination of electronic hardware or computer software and electronic hardware.These functions are actually
It is implemented in hardware or software, the specific application and design constraint depending on technical solution.Professional technician
Each specific application can be used different methods to achieve the described function, but this realization is it is not considered that exceed
The scope of the present invention.
Fig. 5 shows the schematic block diagram of processing system for video 500 according to an embodiment of the invention.Video processing system
System 500 includes image collecting device 510, storage device 520 and processor 530.Processing system for video 500 can be candid photograph
Machine, such as face snap machine.
Described image acquisition device 510 is for acquiring video flowing.Image collecting device 510 is optional, video processing system
System 500 can not include image collecting device 510.In this case, it is alternatively possible to be adopted using other image collecting devices
Collect video flowing, and by the video stream of acquisition to processing system for video 500.
The storage of storage device 520 is for realizing the corresponding steps in method for processing video frequency according to an embodiment of the present invention
Computer program instructions.
The processor 530 is for running the computer program instructions stored in the storage device 520, to execute basis
The corresponding steps of the method for processing video frequency of the embodiment of the present invention.
In one embodiment, for executing following step when the computer program instructions are run by the processor 530
Suddenly：Obtain video flowing；Target detection and tracking are carried out to video flowing, to determine at least one pursuit path；For at least one
Each of pursuit path is extracted respectively from at least partly video frame of the pursuit path comprising corresponding to the pursuit path
The target image block of target, to obtain at least one target image block；For each of at least one pursuit path, to extremely
At least partly target image block in a few target image block is clustered, to obtain at least one cluster centre；For extremely
Each of few pursuit path, exports target image block corresponding at least one cluster centre.
In addition, according to embodiments of the present invention, additionally providing a kind of storage medium, storing program on said storage
Instruction, when described program instruction is run by computer or processor for executing the method for processing video frequency of the embodiment of the present invention
Corresponding steps, and for realizing the corresponding module in video process apparatus according to an embodiment of the present invention.The storage medium
It such as may include the storage card of smart phone, the storage unit of tablet computer, the hard disk of personal computer, read-only memory
(ROM), Erasable Programmable Read Only Memory EPROM (EPROM), portable compact disc read-only memory (CD-ROM), USB storage,
Or any combination of above-mentioned storage medium.
In one embodiment, described program instruction can make computer or place when being run by computer or processor
Reason device realizes each functional module of video process apparatus according to an embodiment of the present invention, and and/or can execute according to this
The method for processing video frequency of inventive embodiments.
In one embodiment, described program instruction is at runtime for executing following steps：Obtain video flowing；To video
Stream carries out target detection and tracking, to determine at least one pursuit path；For each of at least one pursuit path, from
The target image block comprising target corresponding to the pursuit path is extracted in at least partly video frame of the pursuit path, respectively to obtain
Obtain at least one target image block；For each of at least one pursuit path, at least one target image block
At least partly target image block is clustered, to obtain at least one cluster centre；For every at least one pursuit path
One, export target image block corresponding at least one cluster centre.
Each module in processing system for video according to an embodiment of the present invention can pass through reality according to an embodiment of the present invention
The processor computer program instructions that store in memory of operation of the electronic equipment of video processing are applied to realize, or can be with
The computer instruction stored in the computer readable storage medium of computer program product according to an embodiment of the present invention is counted
Calculation machine is realized when running.
Although describing example embodiment by reference to attached drawing here, it should be understood that above example embodiment are only exemplary
, and be not intended to limit the scope of the invention to this.Those of ordinary skill in the art can carry out various changes wherein
And modification, it is made without departing from the scope of the present invention and spiritual.All such changes and modifications are intended to be included in appended claims
Within required the scope of the present invention.
Those of ordinary skill in the art may be aware that list described in conjunction with the examples disclosed in the embodiments of the present disclosure
Member and algorithm steps can be realized with the combination of electronic hardware or computer software and electronic hardware.These functions are actually
It is implemented in hardware or software, the specific application and design constraint depending on technical solution.Professional technician
Each specific application can be used different methods to achieve the described function, but this realization is it is not considered that exceed
The scope of the present invention.
In several embodiments provided herein, it should be understood that disclosed device and method can pass through it
Its mode is realized.For example, apparatus embodiments described above are merely indicative, for example, the division of the unit, only
Only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components can be tied
Another equipment is closed or is desirably integrated into, or some features can be ignored or not executed.
In the instructions provided here, numerous specific details are set forth.It is to be appreciated, however, that implementation of the invention
Example can be practiced without these specific details.In some instances, well known method, structure is not been shown in detail
And technology, so as not to obscure the understanding of this specification.
Similarly, it should be understood that in order to simplify the present invention and help to understand one or more of the various inventive aspects,
To in the description of exemplary embodiment of the present invention, each feature of the invention be grouped together into sometimes single embodiment, figure,
Or in descriptions thereof.However, the method for the invention should not be construed to reflect following intention：It is i.e. claimed
The present invention claims features more more than feature expressly recited in each claim.More precisely, such as corresponding power
As sharp claim reflects, inventive point is that the spy of all features less than some disclosed single embodiment can be used
Sign is to solve corresponding technical problem.Therefore, it then follows thus claims of specific embodiment are expressly incorporated in this specific
Embodiment, wherein each, the claims themselves are regarded as separate embodiments of the invention.
It will be understood to those skilled in the art that any combination pair can be used other than mutually exclusive between feature
All features disclosed in this specification (including adjoint claim, abstract and attached drawing) and so disclosed any method
Or all process or units of equipment are combined.Unless expressly stated otherwise, this specification (is wanted including adjoint right
Ask, make a summary and attached drawing) disclosed in each feature can be replaced with an alternative feature that provides the same, equivalent, or similar purpose.
In addition, it will be appreciated by those of skill in the art that although some embodiments described herein include other embodiments
In included certain features rather than other feature, but the combination of the feature of different embodiments mean it is of the invention
Within the scope of and form different embodiments.For example, in detail in the claims, embodiment claimed it is one of any
Can in any combination mode come using.
Various component embodiments of the invention can be implemented in hardware, or to run on one or more processors
Software module realize, or be implemented in a combination thereof.It will be understood by those of skill in the art that can be used in practice
Microprocessor or digital signal processor (DSP) realize some moulds in video process apparatus according to an embodiment of the present invention
The some or all functions of block.The present invention is also implemented as a part or complete for executing method as described herein
The program of device (for example, computer program and computer program product) in portion.It is such to realize that program of the invention can store
On a computer-readable medium, it or may be in the form of one or more signals.Such signal can be from internet
Downloading obtains on website, is perhaps provided on the carrier signal or is provided in any other form.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and ability
Field technique personnel can be designed alternative embodiment without departing from the scope of the appended claims.In the claims,
Any reference symbol between parentheses should not be configured to limitations on claims.Word "comprising" does not exclude the presence of not
Element or step listed in the claims.Word "a" or "an" located in front of the element does not exclude the presence of multiple such
Element.The present invention can be by means of including the hardware of several different elements and being come by means of properly programmed computer real
It is existing.In the unit claims listing several devices, several in these devices can be through the same hardware branch
To embody.The use of word first, second, and third does not indicate any sequence.These words can be explained and be run after fame
Claim.
The above description is merely a specific embodiment or to the explanation of specific embodiment, protection of the invention
Range is not limited thereto, and anyone skilled in the art in the technical scope disclosed by the present invention, can be easily
Expect change or replacement, should be covered by the protection scope of the present invention.Protection scope of the present invention should be with claim
Subject to protection scope.
Claims (13)
1. a kind of method for processing video frequency, including：
Obtain video flowing；
Target detection and tracking are carried out to the video flowing, to determine at least one pursuit path；
For each of at least one described pursuit path,
Extract the target image comprising target corresponding to the pursuit path respectively from at least partly video frame of the pursuit path
Block, to obtain at least one target image block；
At least partly target image block at least one described target image block is clustered, to obtain at least one cluster
Center；
Export target image block corresponding at least one described cluster centre.
2. the method for claim 1, wherein at least partly target at least one described target image block
Image block is clustered, and includes to obtain at least one cluster centre：
Extract the feature vector of each target image block in at least partly target image block；And
The feature vector of at least partly target image block is clustered, to obtain at least one described cluster centre,
In, each of at least one described cluster centre is the feature vector of described at least partly one of target image block.
3. the method for claim 1, wherein the method also includes：
For each of at least one described pursuit path,
Quality evaluation is carried out respectively at least one target image block described in the pursuit path, to obtain at least one described mesh
The quality score of logo image block；And
According to the quality score of at least one target image block, selected from least one described target image block it is described to
Small part target image block.
4. method as claimed in claim 3, wherein the quality score of at least one target image block according to, from
The selection at least partly target image block includes at least one described target image block：
Quality score is selected to be higher than the target image block of predetermined threshold from least one described target image block, or from described
The target image block of the highest predetermined number of quality score is selected at least one target image block, it is described at least partly with acquisition
Target image block.
5. method as claimed in claim 3, wherein the quality score of at least one target image block according to, from
The selection at least partly target image block includes at least one described target image block：
If there are the target image blocks that quality score is higher than predetermined threshold at least one described target image block, from described
Quality score is selected to be higher than the target image block of predetermined threshold at least one target image block, to obtain at least partly mesh
Logo image block；
If there is no the target image blocks that quality score is higher than predetermined threshold at least one described target image block, from institute
The target image block that the highest predetermined number of quality score is selected at least one target image block is stated, to obtain at least portion
Partial objectives for image block.
6. such as the described in any item methods of claim 3 to 5, wherein clarity of the quality score based on target image block
It calculates and obtains.
7. such as method described in any one of claim 1 to 5, wherein it is described at least one described target image block extremely
Small part target image block is clustered, to obtain at least one cluster centre and export at least one described cluster centre institute
The step of corresponding target image block, executes in the case where the length of the pursuit path is greater than length threshold,
The method also includes：
For each of at least one described pursuit path, if the length of the pursuit path is less than or equal to the length
Threshold value, then export at least one described target image block of the pursuit path, or from described in the pursuit path at least one
Target image block selects one or more target image blocks, and exports selected target image block.
8. the method for claim 7, wherein described to be selected from least one target image block described in the pursuit path
One or more target image blocks include：
One or more of target image blocks are randomly choosed from least one target image block described in the pursuit path.
9. the method for claim 7, wherein described to be selected from least one target image block described in the pursuit path
One or more target image blocks include：
Quality evaluation is carried out respectively at least one target image block described in the pursuit path, to obtain at least one described mesh
The quality score of logo image block；And
According to the quality score of at least one target image block, described one is selected from least one described target image block
A or multiple target image blocks.
10. such as method described in any one of claim 1 to 5, wherein in at least partly video frame from the pursuit path
It is middle to extract the target image block comprising target corresponding to the pursuit path respectively, before obtaining at least one target image block,
The method also includes：
Quality evaluation is carried out respectively to all video frames of the pursuit path, to obtain the matter of all video frames of the pursuit path
Amount scoring；And
According to the quality score of all video frames of the pursuit path, selected from all video frames of the pursuit path it is described to
Small part video frame.
11. a kind of video process apparatus, including：
Module is obtained, for obtaining video flowing；
Detection and tracking module, for carrying out target detection and tracking to the video flowing, to determine at least one pursuit path；
Extraction module is used for for each of at least one described pursuit path, from at least partly view of the pursuit path
The target image block comprising target corresponding to the pursuit path is extracted in frequency frame, respectively to obtain at least one target image block；
Cluster module is used for for each of at least one described pursuit path, at least one described target image block
In at least partly target image block clustered, to obtain at least one cluster centre；
Output module, for exporting at least one described cluster centre for each of at least one described pursuit path
Corresponding target image block.
12. a kind of processing system for video, including processor and memory, wherein be stored with computer program in the memory
Instruction, it is as described in any one of claim 1 to 10 for executing when the computer program instructions are run by the processor
Method for processing video frequency.
13. a kind of storage medium stores program instruction on said storage, described program instruction is at runtime for holding
Row method for processing video frequency as described in any one of claim 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810753529.9A CN108921876A (en) | 2018-07-10 | 2018-07-10 | Method for processing video frequency, device and system and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810753529.9A CN108921876A (en) | 2018-07-10 | 2018-07-10 | Method for processing video frequency, device and system and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108921876A true CN108921876A (en) | 2018-11-30 |
Family
ID=64411324
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810753529.9A Pending CN108921876A (en) | 2018-07-10 | 2018-07-10 | Method for processing video frequency, device and system and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108921876A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109631307A (en) * | 2018-12-06 | 2019-04-16 | 朱朝峰 | Intelligent quick heating type water heater |
| CN109978918A (en) * | 2019-03-21 | 2019-07-05 | 腾讯科技（深圳）有限公司 | A kind of trajectory track method, apparatus and storage medium |
| CN110428445A (en) * | 2019-06-26 | 2019-11-08 | 西安电子科技大学 | A kind of piecemeal tracking and its device, equipment, storage medium |
| CN110502651A (en) * | 2019-08-15 | 2019-11-26 | 深圳市商汤科技有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN111143589A (en) * | 2019-12-06 | 2020-05-12 | Oppo广东移动通信有限公司 | Image processing method and device and storage medium |
| CN112437274A (en) * | 2020-11-17 | 2021-03-02 | 浙江大华技术股份有限公司 | Transmission method of snapshot picture and snapshot machine |
| CN112861565A (en) * | 2019-11-12 | 2021-05-28 | 上海高德威智能交通***有限公司 | Method and device for determining track similarity, computer equipment and storage medium |
| CN112866817A (en) * | 2021-01-06 | 2021-05-28 | 浙江大华技术股份有限公司 | Video playback method, device, electronic device and storage medium |
| JP2022523282A (en) * | 2019-12-24 | 2022-04-22 | 商▲湯▼国▲際▼私人有限公司 | Image selection methods, devices, and electronic devices |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110170744A1 (en) * | 2010-01-08 | 2011-07-14 | University Of Washington | Video-based vehicle detection and tracking using spatio-temporal maps |
| CN104239898A (en) * | 2014-09-05 | 2014-12-24 | 浙江捷尚视觉科技股份有限公司 | Method for carrying out fast vehicle comparison and vehicle type recognition at tollgate |
| CN104731964A (en) * | 2015-04-07 | 2015-06-24 | 上海海势信息科技有限公司 | Face abstracting method and video abstracting method based on face recognition and devices thereof |
| CN106327236A (en) * | 2015-07-10 | 2017-01-11 | 华为技术有限公司 | Method and device for determining user action track |
| CN106687990A (en) * | 2014-09-02 | 2017-05-17 | 柯达阿拉里斯股份有限公司 | A method for selecting frames from video sequences based on incremental improvement |
| CN107729809A (en) * | 2017-09-11 | 2018-02-23 | 汕头大学 | A kind of method, apparatus and its readable storage medium storing program for executing of adaptive generation video frequency abstract |
-
2018
- 2018-07-10 CN CN201810753529.9A patent/CN108921876A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110170744A1 (en) * | 2010-01-08 | 2011-07-14 | University Of Washington | Video-based vehicle detection and tracking using spatio-temporal maps |
| CN106687990A (en) * | 2014-09-02 | 2017-05-17 | 柯达阿拉里斯股份有限公司 | A method for selecting frames from video sequences based on incremental improvement |
| CN104239898A (en) * | 2014-09-05 | 2014-12-24 | 浙江捷尚视觉科技股份有限公司 | Method for carrying out fast vehicle comparison and vehicle type recognition at tollgate |
| CN104731964A (en) * | 2015-04-07 | 2015-06-24 | 上海海势信息科技有限公司 | Face abstracting method and video abstracting method based on face recognition and devices thereof |
| CN106327236A (en) * | 2015-07-10 | 2017-01-11 | 华为技术有限公司 | Method and device for determining user action track |
| CN107729809A (en) * | 2017-09-11 | 2018-02-23 | 汕头大学 | A kind of method, apparatus and its readable storage medium storing program for executing of adaptive generation video frequency abstract |
Non-Patent Citations (4)
| Title |
|---|
| 刘守达: "基于多目标跟踪及轨迹组合优化的视频摘要", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 孙恩情: "监控视频摘要提取与展示技术", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 杨露菁: "《军队"2110工程"三期建设教材 战场情报信息综合处理技术》", 30 June 2017, 国防工业出版社 * |
| 葛致磊等: "《导弹导引***原理》", 31 March 2016, 国防工业出版社 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109631307A (en) * | 2018-12-06 | 2019-04-16 | 朱朝峰 | Intelligent quick heating type water heater |
| CN109978918A (en) * | 2019-03-21 | 2019-07-05 | 腾讯科技（深圳）有限公司 | A kind of trajectory track method, apparatus and storage medium |
| CN110428445A (en) * | 2019-06-26 | 2019-11-08 | 西安电子科技大学 | A kind of piecemeal tracking and its device, equipment, storage medium |
| CN110428445B (en) * | 2019-06-26 | 2023-06-27 | 西安电子科技大学 | Block tracking method and device, equipment and storage medium thereof |
| CN110502651A (en) * | 2019-08-15 | 2019-11-26 | 深圳市商汤科技有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN110502651B (en) * | 2019-08-15 | 2022-08-02 | 深圳市商汤科技有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN112861565A (en) * | 2019-11-12 | 2021-05-28 | 上海高德威智能交通***有限公司 | Method and device for determining track similarity, computer equipment and storage medium |
| CN111143589A (en) * | 2019-12-06 | 2020-05-12 | Oppo广东移动通信有限公司 | Image processing method and device and storage medium |
| JP2022523282A (en) * | 2019-12-24 | 2022-04-22 | 商▲湯▼国▲際▼私人有限公司 | Image selection methods, devices, and electronic devices |
| CN112437274A (en) * | 2020-11-17 | 2021-03-02 | 浙江大华技术股份有限公司 | Transmission method of snapshot picture and snapshot machine |
| CN112437274B (en) * | 2020-11-17 | 2022-08-30 | 浙江大华技术股份有限公司 | Transmission method of snapshot picture and snapshot machine |
| CN112866817A (en) * | 2021-01-06 | 2021-05-28 | 浙江大华技术股份有限公司 | Video playback method, device, electronic device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108921876A (en) | Method for processing video frequency, device and system and storage medium | |
| CN109325933B (en) | Method and device for recognizing copied image | |
| CA3000127C (en) | System and method for appearance search | |
| US10672140B2 (en) | Video monitoring method and video monitoring system | |
| CN108875676A (en) | Biopsy method, apparatus and system | |
| CN104050449B (en) | A kind of face identification method and device | |
| CN105518709B (en) | The method, system and computer program product of face for identification | |
| CN108846365B (en) | Detection method and device for fighting behavior in video, storage medium and processor | |
| CA2884670C (en) | System and method for generating an activity summary of a person | |
| US8548258B2 (en) | Method system and associated modules and software components for providing image sensor based human machine interfacing | |
| CN108875540A (en) | Image processing method, device and system and storage medium | |
| CN108875535A (en) | image detecting method, device and system and storage medium | |
| CN110245561B (en) | Face recognition method and device | |
| CN106033601B (en) | The method and apparatus for detecting abnormal case | |
| CN108805047A (en) | A kind of biopsy method, device, electronic equipment and computer-readable medium | |
| EP2428036A1 (en) | Systems and methods for the autonomous production of videos from multi-sensored data | |
| CN108875542A (en) | A kind of face identification method, device, system and computer storage medium | |
| EP2783508A1 (en) | Geographic map based control | |
| CN109791615A (en) | For detecting and tracking the method, target object tracking equipment and computer program product of target object | |
| CN109711357A (en) | A kind of face identification method and device | |
| TW201710991A (en) | Analytics system and method | |
| JP2016181159A (en) | System, retrieval method and program | |
| KR20160035121A (en) | Method and Apparatus for Counting Entity by Using Location Information Extracted from Depth Image | |
| KR20160033800A (en) | Method for counting person and counting apparatus | |
| CN110309720A (en) | Video detecting method, device, electronic equipment and computer-readable medium |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181130 |