CN104867161B - A kind of method for processing video frequency and device - Google Patents
A kind of method for processing video frequency and device Download PDFInfo
- Publication number
- CN104867161B CN104867161B CN201510246765.8A CN201510246765A CN104867161B CN 104867161 B CN104867161 B CN 104867161B CN 201510246765 A CN201510246765 A CN 201510246765A CN 104867161 B CN104867161 B CN 104867161B
- Authority
- CN
- China
- Prior art keywords
- shot
- moving target
- target area
- gaussian profile
- segmentation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 67
- 238000012545 processing Methods 0.000 title claims abstract description 17
- 230000011218 segmentation Effects 0.000 claims abstract description 92
- 230000033001 locomotion Effects 0.000 claims abstract description 65
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 53
- 239000000284 extract Substances 0.000 claims abstract description 11
- 230000008569 process Effects 0.000 claims description 21
- 238000009826 distribution Methods 0.000 claims description 14
- 238000000605 extraction Methods 0.000 claims description 11
- 238000005520 cutting process Methods 0.000 claims description 6
- 230000015572 biosynthetic process Effects 0.000 claims description 5
- 230000005055 memory storage Effects 0.000 claims description 5
- 238000003786 synthesis reaction Methods 0.000 claims description 4
- 239000004744 fabric Substances 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 abstract description 7
- 238000001514 detection method Methods 0.000 description 18
- 230000008859 change Effects 0.000 description 17
- 230000006870 function Effects 0.000 description 16
- 241000209140 Triticum Species 0.000 description 10
- 235000021307 Triticum Nutrition 0.000 description 10
- 230000009977 dual effect Effects 0.000 description 9
- 230000000694 effects Effects 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 230000035772 mutation Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 206010068052 Mosaicism Diseases 0.000 description 2
- 230000036624 brainpower Effects 0.000 description 2
- 239000003086 colorant Substances 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 210000003765 sex chromosome Anatomy 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000002045 lasting effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000009017 pursuit movement Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/73—Querying
- G06F16/738—Presentation of query results
- G06F16/739—Presentation of query results in form of a video summary, e.g. the video summary being a video sequence, a composite still image or having synthesized frames
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/215—Motion-based segmentation
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of method for processing video frequency and device, belongs to video signal processing field, and methods described includes:Target area is marked in video pictures according to preparatory condition;According to the first preset algorithm by the Video segmentation of the target area into multiple shot segmentations;The key frame of each shot segmentation is chosen according to preset rules;The moving target of each shot segmentation is obtained by the second preset algorithm, to the tracing of the movement of acquired each moving target, extracts each two field picture of the movement locus of each moving target;Each two field picture of the movement locus of each moving target of the target area is superimposed to corresponding to the movement locus of the moving target on the key frame of shot segmentation, generates video frequency abstract.It is an object of the invention to provide a kind of method for processing video frequency and device, to effectively improve the monitoring efficiency of existing Video Supervision Technique.
Description
Technical field
The present invention relates to video signal processing field, in particular to a kind of method for processing video frequency and device.
Background technology
Video monitoring turns into highly important technological means in modern society's safety-security area, but the information of video monitor
The poorly efficient sex chromosome mosaicism that superfluous She's property and information of interest are searched also is hampering effective application of video brainpower watch and control technology.
For example, tens of or even hundreds of hours may be up to by being related to the monitor video of a button, and wherein to solving a case really
Often only there is the of short duration time in monitored picture in useful important scenes, if using manually the mode of original video is checked
These written in water important informations are easily omitted, cause efficiency low.
The content of the invention
It is an object of the invention to provide a kind of method for processing video frequency and device, to effectively improve existing video monitoring skill
The monitoring efficiency of art.
In a first aspect, a kind of method for processing video frequency provided in an embodiment of the present invention, methods described include:
Target area is marked in video pictures according to preparatory condition;
According to the first preset algorithm by the Video segmentation of the target area into multiple shot segmentations;
The key frame of each shot segmentation is chosen according to preset rules;
The moving target of each shot segmentation is obtained by the second preset algorithm, to acquired each moving target
Tracing of the movement, extract each two field picture of the movement locus of each moving target;
Each two field picture of the movement locus of each moving target of the target area is superimposed to the moving target
Movement locus corresponding to shot segmentation key frame on, generate video frequency abstract.
With reference in a first aspect, the embodiment of the present invention additionally provides the first possible embodiment of first aspect, wherein, institute
That states obtains the moving target of each shot segmentation by the second preset algorithm, to the fortune of acquired each moving target
Dynamic track following, each two field picture of the movement locus of each moving target is extracted, including:
The fortune of each shot segmentation of the target area is found out by the Potential Prediction algorithm based on gauss hybrid models
Moving-target, to the tracing of the movement of each moving target found, extract each moving target found
Movement locus each two field picture.
With reference to the first possible embodiment of first aspect, the embodiment of the present invention additionally provides second of first aspect
Possible embodiment, wherein, it is described that the target area is found out by the Potential Prediction algorithm based on gauss hybrid models
The moving target of each shot segmentation, to the tracing of the movement of each moving target found, extract what is found
Each two field picture of the movement locus of each moving target, including:
Pixel corresponding to a coordinate position of each shot segmentation for obtaining the target area is calculated time t's
Observation is Xt, the pixel is { X in a series of observations at different moments1, X2..., Xt, Gaussian Profile is expressed as:
Obtain the Gaussian Profile of all pixels point corresponding to the coordinate position of the target area;
Gaussian Profile set is established, the institute of each shot segmentation of target area described in the Gaussian Profile set memory storage
There is the Gaussian Profile of pixel;
All pixels for not meeting the Gaussian Profile in the Gaussian Profile set are preserved according to default comparison rules
For a two field picture of the movement locus, wherein, the default comparison rules include:Judge the target area at current time
A pixel whether meetIf, it is believed that the pixel meets in the Gaussian Profile set
Gaussian Profile, if not, it is believed that the pixel does not meet the Gaussian Profile in the Gaussian Profile set.
With reference in a first aspect, the embodiment of the present invention additionally provides the third possible embodiment of first aspect, wherein, institute
That states chooses the key frame of each shot segmentation according to preset rules, including:
Key frame using the last frame on the time shaft of each shot segmentation as the shot segmentation, a pass
The corresponding shot segmentation of key frame.
With reference to the third possible embodiment of first aspect, the embodiment of the present invention additionally provides the 4th kind of first aspect
Possible embodiment, wherein, the Video segmentation of the target area is wrapped into multiple shot segmentations according to the first preset algorithm
Include:
Judge whether is similarity between two adjacent on a timeline frames by color histogram nomography based on RGB
Less than default first threshold;
If so, sentenced by the way that the color histogram nomography judgement based on HSV is described by the color histogram nomography based on RGB
Whether the similarity that disconnected similarity is less than two frames of the first threshold is less than default Second Threshold;
If so, judge that two frames belong to two different camera lenses;
Shot boundary is set between being judged to belonging to two frames of two different camera lenses;
According to set a plurality of shot boundary by the Video segmentation of the target area into multiple shot segmentations.
Second aspect, the embodiments of the invention provide a kind of video process apparatus, described device includes:
Indexing unit, for marking target area in video pictures according to preparatory condition;
Cutting unit, for according to the first preset algorithm by the Video segmentation of the target area into multiple shot segmentations;
Extraction unit, for choosing the key frame of each shot segmentation according to preset rules;
Statistic unit, for obtaining the moving target of each shot segmentation by the second preset algorithm, to acquired
Each moving target tracing of the movement, extract each two field picture of the movement locus of each moving target;
First synthesis unit, for each two field picture of the movement locus of each moving target of the target area to be folded
Add to corresponding to the movement locus of the moving target on the key frame of shot segmentation, generate video frequency abstract.
With reference to second aspect, the embodiment of the present invention additionally provides the first possible embodiment of second aspect, wherein, institute
Stating statistic unit also includes:
Subelement is followed the trail of, for finding out the every of the target area by the Potential Prediction algorithm based on gauss hybrid models
The moving target of individual shot segmentation, to the tracing of the movement of each moving target found, extraction is found every
Each two field picture of the movement locus of the individual moving target.
With reference to the first possible embodiment of second aspect, the 3rd of the second aspect that the embodiment of the present invention additionally provides the
Kind possible embodiment, wherein, the tracking subelement includes:
First computation subunit, for calculating a coordinate position pair of each shot segmentation for obtaining the target area
The pixel answered is X in time t observationt, the pixel is { X in a series of observations at different moments1, X2...,
Xt, Gaussian Profile is expressed as:
Second computation subunit, for obtaining the Gauss of all pixels point corresponding to the coordinate position of the target area
Distribution;
3rd computation subunit, for establishing Gaussian Profile set, target area described in the Gaussian Profile set memory storage
The Gaussian Profile of all pixels point of each shot segmentation in domain;
4th computation subunit, for according to default comparison rules by all height not met in the Gaussian Profile set
The pixel of this distribution saves as a two field picture of the movement locus, wherein, the default comparison rules include:Judge current
Whether one pixel of the target area at moment meetsIf, it is believed that the pixel meets institute
The Gaussian Profile in Gaussian Profile set is stated, if not, it is believed that the pixel does not meet the Gauss in the Gaussian Profile set
Distribution.
With reference to second aspect, the embodiment of the present invention additionally provides the third possible embodiment of second aspect, wherein, institute
Extraction unit is stated specifically for the pass using the last frame on the time shaft of each shot segmentation as the shot segmentation
Key frame, the corresponding shot segmentation of a key frame.
With reference to the third possible embodiment of second aspect, the embodiment of the present invention additionally provides the 4th kind of second aspect
Possible embodiment, wherein, the cutting unit includes:
First comparison subunit, for judging on a timeline adjacent two by color histogram nomography based on RGB
Whether the similarity between individual frame is less than default first threshold;
Second comparison subunit, if judging adjacent on a timeline two by color histogram nomography based on RGB
Similarity between frame is less than default first threshold, by color histogram nomography based on HSV judge it is described by based on
Whether the similarity that the similarity that RGB color histogram nomography judges is less than two frames of the first threshold is less than default the
Two threshold values;
Subelement is judged, if for passing through the color histogram based on RGB described in the color histogram nomography judgement based on HSV
The similarity that the similarity that nomography judges is less than two frames of the first threshold is less than default Second Threshold, judges two frame category
In two different camera lenses;
Subelement is set, for setting shot boundary between being judged to belonging to two frames of two different camera lenses;
Split subelement, for according to set a plurality of shot boundary by the Video segmentation of the target area into multiple
Shot segmentation.
The embodiment of the present invention in existing monitor video by delimiting target area, by regarding for set target area
Frequency division is cut into multiple shot segmentations, and the generation of an event of the target area is represented in each shot segmentation.
The moving target in the target area is extracted, obtains the movement locus of moving target, by the moving target
Each two field picture of movement locus all preserves, therefore just acquires the motion rail of all moving targets in current shot segmentation
Mark, then the movement locus of the moving target is added on the key frame of current shot segmentation, synthetic video summary.Therefore,
Compared with also having the video information in substantial amounts of non-interesting region in the monitor video of prior art, can more accurately it find interested
Video, and decrease the size of video.
Other features and advantages of the present invention will illustrate in subsequent specification, also, partly become from specification
It is clear that or by implementing understanding of the embodiment of the present invention.The purpose of the present invention and other advantages can be by saying what is write
Specifically noted structure is realized and obtained in bright book, claims and accompanying drawing.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to institute in embodiment
The accompanying drawing needed to use is briefly described, it should be apparent that, drawings in the following description are only some implementations of the present invention
Example, for those of ordinary skill in the art, on the premise of not paying creative work, can also be obtained according to these accompanying drawings
Obtain other accompanying drawings.By the way that shown in accompanying drawing, above and other purpose of the invention, feature and advantage will become apparent from.In whole
Identical reference instruction identical part in accompanying drawing.Deliberately accompanying drawing, emphasis are not drawn by actual size equal proportion scaling
It is the purport for showing the present invention.
Fig. 1 shows a kind of method flow diagram of the embodiment of method for processing video frequency provided in an embodiment of the present invention;
Fig. 2 shows a kind of module frame chart of the embodiment of video process apparatus provided in an embodiment of the present invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Whole description, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other under the premise of creative work is not made
Embodiment, belong to the scope of protection of the invention.
Video monitoring turns into highly important technological means in modern society's safety-security area, but the information of video monitor
The poorly efficient sex chromosome mosaicism that superfluous She's property and information of interest are searched also is hampering effective application of video brainpower watch and control technology.
For example, tens of or even hundreds of hours may be up to by being related to the monitor video of a button, and wherein to solving a case really
Often only there is the of short duration time in monitored picture in useful important scenes, if using manually the mode of original video is checked
These written in water important informations are easily omitted, cause efficiency low.
To solve above-mentioned difficulties, the embodiments of the invention provide a kind of method for processing video frequency, as shown in figure 1, methods described
Including:
S11:Mark target area;
Target area is marked in video pictures according to preparatory condition, such as the video on one section of monitoring square, and user
Interested is only video monitoring situation somewhere, such as puts the place of bicycle, then user is then according to oneself
Demand target area is chosen in monitor video, because each pixel corresponds to a coordinate in video, therefore can pass through
The mode selection target region of respective pixel coordinate is chosen in video.
S12:According to the first preset algorithm by the Video segmentation of target area into multiple shot segmentations;
It is crucial that correctly finding out position and the moment of lens mutation, i.e. mirror in the shot segmentation to target area
Head abrupt climatic change, can carry out Abrupt shot change detection in the following way in the embodiment of the present invention:
Shot Detection based on RGB color
The basic theories of RGB color space is exactly all to be by all colours by linear group of three kinds of primary colours of red, green, blue
Synthesis.The color space of this definition is used as this three-dimensional equivalent to using red, green, blue as in a three dimensions of base
The span of each base in space can only be (0,255] between.And in this color space, the color in reality is direct
Distance corresponding to difference and color space mountain can not turn into a kind of linear relationship.
Abrupt shot is detected according to RGB color, its archetype established:In color space, by it
It is divided into N number of subinterval, wherein H (fm, i) and as the sum of all pixels mesh for belonging to i-th of color interval in m frames, then in m frames and n
The distance between frame s (fm, fn) represent as follows:
The distance between two frames are calculated, is used as and compares so as to whether be camera lens side in the threshold value by setting in advance
Boundary.It is exactly that the color histogram for calculating two frames goes to judge the similarity between two frames so as to go to judge in another processing method
Whether it is shot boundary.For two pin similarity calculating methods mainly as shown in equation below:
In above-mentioned formula, wherein ai, biRatio value is normalized after the histogram obtained for two field pictures, wherein corresponding
Ai, biCalculating is respectively:
Wherein Ai, BiRespectively frame of video A, B two images obtain histogram and normalized result.
By calculating the similarity of two field pictures histogram, then made comparisons according to given threshold value, if similarity
Then can determine that it is not shot boundary more than threshold value, otherwise be shot boundary between can be determined that this two frame.
Histogram in rgb space be it is simple retouch numerically proportion of the plain color in entire image, but
It is not go to obtain the spatial positional information on color in structure.If tying in face of two width ratio on color data is identical
Entirely different on structure, the difference that two frames are calculated by rgb space should be zero, but the gap of the frame of this in reality two can not possibly be
Zero.In order to solve this problem, the detection method of another abrupt shot is just introduced.
(2) detection based on hsv color space abrupt shot
Colour information exactly is gone to carry out resolving into this three attributes such as saturation degree, tone, brightness by hsv color model
Description.In these three attributes, saturation degree and brightness represent to protect the ratio of color and black in color.Last attribute then represents
Basic pure color.This color model is to be based on a kind of perception and more intuitively color model space.This color
Space realizes that the numeral to color describes by describing the purity of color category, the size of brightness and color, tool experiment
Analysis, this description method are more suitable for the mode that the mankind observe color.
But when specifically analyzing video frame images, typically result in be r ' g ', the b ' of RGB color really
Value is cut, and the value of three attributes in HSV space can not be directly obtained.So to go to carry out shot boundary detector in HSV space,
Then need to be converted from rgb space to HSV space, its step of converting enters and has formula as follows:
R ', g ', b ' numerical value are defined, and calculation formula is:
In formula is appealed, wherein max=max (R, G, B), min=min (R, G, B).
Calculate hsv values
Before this, it is necessary first to define h ', and its calculating process is as follows:
If r=max&g=min, h '=5+b '
If r=max&g ≠ min, h '=1-g '
If g=max&b=min, h '=1+r '
If g=max&b ≠ min, h '=3-b '
H '=3+g ' if b=max&r=min '
In the other cases:H '=5+b '.
After h ' is obtained, can with three attributes obtained in specific HSV space occurrence calculation formula it is as follows:
Two frames are being calculated specifically after HSV space occurrence, are then handling can with the method told on RGB all
Equally, can go to carry out the detection to shot boundary by calculating the similarity of distance or histogram of two frames between this.
In embodiments of the present invention, it is to use both detection methods simultaneously to the border detection based on color space, so
Its result is done into a union afterwards, so as to draw last differentiation result.
The similarity between two adjacent on a timeline frames is judged by the color histogram nomography based on RGB is
It is no to be less than default first threshold;
If so, sentenced by the way that the color histogram nomography judgement based on HSV is described by the color histogram nomography based on RGB
Whether the similarity that disconnected similarity is less than two frames of the first threshold is less than default Second Threshold;
If so, judge that two frames belong to two different camera lenses.Be judged to belonging to two different camera lenses two frames it
Between shot boundary is set;
According to set a plurality of shot boundary by the Video segmentation of the target area into multiple shot segmentations.
Still an alternative is that based on sliding window extraction gradual shot, it is specific as follows:
Small in the situation in face of lens mutation, the effect obtained by the shot boundary detector based on color space is quite not
It is wrong.But facing to the camera lens of gradual change, the change of camera lens is shown in continuous several frames.Gradual shot so occurs at one
In change, it is a change procedure of a lasting multiframe.In schedule life, the process of gradual change is broadly divided into two kinds of shapes
Formula:One kind is dissolving, refers to that in former frame be to occur the same time slowly obscuring gradually strengthen with a later frame, so
It will occur in the image of this two frames frame certain overlapping;Another form is to be fade-in fade-out to refer to that the appearance of rear frame is necessary
After previous frame is wholly absent.For the camera lens of this gradual change, acquired imitate is detected to video shot boundary using optical flow method
Fruit or more satisfactory.The principle of the main basis of its optical flow method is exactly that camera lens will not produce when the switching of gradual change occurs
Light stream.
From optical flow method, for frame of video it is of overall importance for, a kind of rational detection side can not be found
Method, extraordinary effect can be obtained so as to be detected to whole frame of video.In order to solve this problem, many researchers
The local demand for meeting shot boundary detector is just removed by sliding window mechanism.For one section of smooth video, it is obtained
The theoretically distant requirement for reaching 24f/s and just having been able to meet usually to watch video of the renewal speed of key frame.Based on cunning
In dynamic window shot boundary detection algorithms, its main window size is that the influence to Detection results is very great.At this
When individual algorithm starts to be suggested use, the size fixed setting of its window is 15, and this just needs to ensure in each window
On can only once occur camera lens transformation.For such case in the different video database of the more magnanimity in face, this is certainly can not
Meet requirement of the people to shot boundary accuracy rate.
After the difference between comparing two frames, a threshold value is required for compared with it, so as to determine between two frames
Whether it is shot boundary.In the mode of the selection of threshold value, be the most simply exactly take choose a global threshold by it therewith
Two frame differences afterwards are compared.But the change of camera lens is mutation and gradual change in video.If if both are set
If a threshold value, then the effect of acquirement is less preferable certainly.In order to preferably combine sliding window in camera lens
Application in segmentation, the researcher of Video segmentation eliminate the algorithm of a superposition threshold value, and the threshold value of this algorithm, which is chosen, is
Correlation before and after video goes to choose corresponding local threshold.Compared with global threshold, the effect in face of Abrupt shot change detection exists
Strengthen to a certain extent, but the terminal of this algorithm is at the ending of sliding window, so this is to whole
Video carries out shot segmentation and does not have good adaptability.
In order to solve the of overall importance and locality of threshold value selection, in the selection of threshold value herein, the threshold of selection is taken
Value method is the dual threshold method of a local auto-adaptive.And cardinal principle is exactly to utilize wheat head criterion in dual threashold value-based algorithm
On, the dual threashold value-based algorithm of a local auto-adaptive can be so obtained so as to obtain corresponding more satisfactory Research on threshold selection.
Before this algorithm is introduced, first it is to be understood that timely dual threshold method and wheat head criterion the algorithm of itself, only exist
Solving the two algorithms can understand the dual threashold value-based algorithm of local auto-adaptive afterwards.
In dual threashold value-based algorithm, its main thought just just sets two threshold values TH and TL before being cut to video, this
The size of two threshold values is exactly one high and one low.Then go to detect abrupt shot using high threshold, and Low threshold goes to face
Gradual shot goes to detect.
Mainly solved the problems, such as in wheat head criterion be by N number of wheat head of different sizes it is random be placed on N number of different sequence
Number position.Numeric order according to institute's label go select a maximum wheat head, it is therein requirement be exactly when selection only
It can go to choose according to the order of this sequence number, it is impossible to recalled.If this problem is subjected to a solution, then
Need to go solve this problem using the maximum fuzzy value in applied optics long.The algorithm mainly provided in wheat head criterion is just
It is the position that the maximum wheat head is estimated using Principle of Statistics, its main way is exactly divided into M sections by N number of.And M-1 sections are sought
The maximum wheat head in subsegment is found out as a statistical analysis, then according to residing for statistical result estimates the whole maximum wheat head
Position.
Selection dynamic sliding window mechanism biggest advantage is exactly to have part well when cutting to video
Adaptability, it is exactly to be carried out according to dual-threshold voltage in the selection for some window threshold value.In the embodiment of the present invention, window W
The average value and standard deviation of difference between interior frame and frame are as follows:
Wherein L is the length of window, can be obtained by many experiments, is imitated when being cut acquired by dual threshold TH=5 μ, TL=3 μ
Fruit is best.But in order to further meet more video split requirements, this carries out adjustment somewhat, meter to TH, TL
It is as follows to calculate formula:
TH=5 μ+λ σ
TL=3 μ+λ σ
As shown in above formula, in order to meet this condition of TH > TL, while also meet the condition of wheat head criterion, wherein
Value requirement to λ is the random value between -1.2~1.8.Thus obtain the dual threshold of local auto-adaptive.
Therefore, it is as follows based on double sliding window mouth shot boundary detection algorithms:
After dual threshold TH, TL with local adaptation, the border detection algorithm can be carried out below detailed
Introduction:
The first step:Set the L=1 in initial value window W.
Second step:By L=1+L, and corresponding Di, TH and TL are calculated, be then compared the value of this three:
If Di< TL, then go back to second step;
If Di> TH, then go to the 3rd step;
Other, then go to the 4th step;
3rd step:Now think that window W undergos mutation, then window frame second from the bottom is marked, as key frame,
And window is slided into frame last, as the first frame after sliding window, and L=1 is set, go back to second step.
4th step:Now think that gradual change occurs for window W, in a start frame of the last frame as gradual shot of window
Q, by L=1, and window W only includes beginning frame Q, and window is carried out, from increasing once, then to do following operation:
Flag=0 is set, and calculates the color histogram difference D in window between framei1, and start frame Q and last
The color histogram difference D of framei2;
If Di1< TL, flag++
Otherwise flag=0, window W are moved, and are come back to (1);
Judge flag > 3, if so, the gradual change is abolished, window is moved into last frame, if not, and if Di1< TL
If with Di2> TL, last frame is marked.
Again sliding window W, and L=1 is set, return to second step.
Pass through above-mentioned algorithm, it is possible to which completion uses segmentation of the double sliding window mouth to camera lens.
S13:Obtain the key frame of each shot segmentation;
After shot segmentation is obtained, then one key frame of extraction to each camera lens is needed.Choose this camera lens key
Frame is preferably able to give expression to the content mainly included in this camera lens, is to like saying that the intermediate frame of camera lens is made in some algorithms
For key frame, in inventive embodiments, selection is the last frame for using camera lens as the key frame of video.
S14:Obtain the movement locus of the moving target of target area;
The moving target of each shot segmentation is obtained by the second preset algorithm, to acquired each moving target
Tracing of the movement, extract each two field picture of the movement locus of each moving target.
It is existing it is relatively good to moving object detection effect be exactly to use background subtraction algorithm, this algorithm ought
Preceding picture subtracts constructed video scene background and can be obtained by corresponding moving target.In the calculation of structure video background scene
In method, wherein Kosé mixed model and LBP texture models algorithm can obtain highly desirable algorithm.In the embodiment of the present invention
In, the algorithm of the motion detection used is namely based on gauss hybrid models algorithm, does detailed introduction to this algorithm below.
In gauss hybrid models, its implied terms be exactly each pixel of frame of video color value on a timeline
It is to meet Gaussian Profile.And due to the fluctuation of local scene so as to causing the color value in a pixel to be that can meet more points
Cloth, establish under this condition, then just complete the background constructing to video using this model.Its key step is as follows:
(1) model definition
Assuming that the observation of some pixel (x, y) time t in frame of video I is designated as Xt, for set point in difference
A series of observation { X at moment1, X2..., Xt, a statistics random process independent with other points is can be regarded as, then may be used
Represented with K Gaussian Profile:
Wherein, i=1,2 ..., K.K value determines by factors such as calculating performances, by substantial amounts of it is demonstrated experimentally that K
Value is most rational for 3 to 7.Then the probability distribution of t point (x, y) is:
Wherein, ωI, tFor the weights of i-th of Gaussian Profile of t, it reflects the ratio that the Gaussian Profile occurs, andη(Xt, μI, t∑I, t) it is that i-th of average of t is μI, t, covariance be ∑I, tProbability density function:
(2) model modification
If current pixel value meets with a certain Gaussian Profile averageThen think that the match is successful, passing through
Cross a large amount of it is demonstrated experimentally that the effect of D values 2.5 is best.If multiple matchings, then best one is selected.The match is successful
When, adjust the weighted value of each distribution:
ωI, t=(1- α) ωI, t-1+αMI, t
Wherein, α is learning rate, and for its value between (0,1), α is bigger, right value update it is faster, it is otherwise on the contrary;For matching
Distribution K, MI, tFor 1, remaining unmatched distribution of weights value increase for being distributed as 0, can so causing matching, do not reduce not
With distribution weighted value.
For the model to match with current pixel, its parameter will do following renewal:
μt=(1- ρ) μt-1+ρXt
Wherein, ρ is another learning rate, and its value is ρ=α η (Xt|μk, σk), it is Gaussian probability-density function.And for not having
There is the distribution of matching, its parameter keeps constant.
(3) foreground detection
According to ωI, t/σI, tValue is ranked up as descending to K Gaussian Profile, the more preceding Gaussian Profile of sequence, more suitable
Close description background.M (1≤M≤K) individual Gaussian Profile is perceived as the description to background before general selection meets:
Wherein, T is background model proportion threshold value, if T settings are smaller, GMM will deteriorate to single Gaussian distribution model;If
T values are larger, then can be that dynamic background establish the mixed models of multiple Gaussian Profiles to simulate.By the way that experimental results demonstrate T's
Empirical value desirable 0.6, comparatively effect now is best.
After extracting moving target, followed by the acquisition of the movement locus of moving target.
Motion target tracking, exactly go to find and required tracking target area position the most similar in image sequence
Put.Herein, the target of required tracking is exactly the moving target come out in upper one section by motion detection, and target following
It is exactly to orient by this moving target in image sequence with regard to simple one.It is to solve two problems in motion target tracking:
First, the model effective expression of moving target is come out;Second, which kind of target signature is selected to go to be matched.Nowadays mesh is being moved
In the algorithm for marking tracking, the appropriate method of phase knowledge and magnanimity and searching algorithm are can be divided mainly into, one is to utilize frame in both algorithms
Match somebody with somebody and the method for Background matching look for tracking target, but it is involved the problem of be all that the superfluous She's information of processing is relatively more, institute
It is exactly the average drifting method optimized to receiving rope scope with target tracking algorism selected in embodiments of the present invention
(meanshift algorithms), the essential idea of its algorithm are exactly the characteristic value by calculating pixel in target area and candidate region
Complete, to target and the model of candidate region, then to go to calculate the acquaintance of object module and candidate region using phase knowledge and magnanimity function
Degree.The maximum candidate family of selection phase knowledge and magnanimity goes to establish corresponding Meanshift vectors, and is repeating a process, due to
Mean shift algorithm inherently has this feature of convergence, and by iteration for several times, the algorithm can find the true of moving target
Position, so as to reach pursuit movement target.
Determine that target is feeling the locus and region of emerging region appearance by motion detection, it is assumed that in this region
There is n pixel, and with { zi, i=1...n represents the position of each pixel.Target area is subjected to color space progress
It is evenly dividing, it is possible to obtain the q of histogram, then object module formed between m equivalent numerical valuenThe probability density of formation can
To be represented by formula 23, wherein u=1 ..., m.
In formula is appealed, itsThe seat after this area pixel is normalized with target's center is represented, wherein (x0,
y0) belong to the centre coordinate of target.K is the kernel function represented, and what is selected herein is Epanechikov functions, b (zi) table
Show and belong to ziThe value of the histogram of color interval, u is the color index as histogram, and δ [b (zi)-u] main effect is just
It is to judge ziWhether u this unit is belonged to, and its intermediate value is only 1 or 0, and last parameter C is as normalization coefficient.
It is to need the center f according to t-1 when object matching is carried out to t frames0To receive in rope window
The heart, so as to obtain the centre coordinate f of candidate target.After target's center's coordinates are obtained, to the Nogata of its whole region
Figure.
In this target area, its pixel coordinate { zi, i=1...n is represented, the probability of its preselected area pixel is close
Spending function is
Wherein h is the size of the size of the yardstick as kernel function, i.e. window, determines the power of each pixel of candidate region
Redistribution.
Similarity come represent candidate region and track target degree of closeness, herein, have chosen Bhattacharyya
For coefficient function as description candidate region and the similarity degree of target, it defines such as equation 2 below 5.
By the center f of t-1 frames0As the center of the initial position of the receipts rope window of t frames, then in t frames
Look for causing the candidate region that acquaintance angle value is maximum around in frame.
Meanshift iterative process
In the iterative process of average drifting, while it is also the receipts rope process for finding target, will in order to which p (p, q) is maximum
Bhattacharyya coefficient functions carry out Taylor expansion, can obtain equation below 26.
In above formula, only the value of Section 2 can change as f changes, in the process of selection phase knowledge and magnanimity maximization
It is exactly from candidate region to the close process in real estate, it is mainly out Meanshift iterative equation and goes what is completed, its
The formula of iterative equation is as follows.
In formula is appealed, g (x)=- K ' (x), during continuous iteration, constantly to searching face exactly since fk
The region that color change is compared, then moved towards that position, it is known that both is specifically les than set threshold value, then repeatedly
In generation, terminates, and its position located now is exactly in the position of the target of present frame
S15:Synthetic video is made a summary;
Each two field picture of the movement locus of each moving target of the target area is superimposed to the moving target
Movement locus corresponding to shot segmentation key frame on, generate video frequency abstract.
The embodiment of the present invention in existing monitor video by delimiting target area, by regarding for set target area
Frequency division is cut into multiple shot segmentations, and the generation of an event of the target area is represented in each shot segmentation.
The moving target in the target area is extracted, obtains the movement locus of moving target, by the moving target
Each two field picture of movement locus all preserves, therefore just acquires the motion rail of all moving targets in current shot segmentation
Mark, then the movement locus of the moving target is added on the key frame of current shot segmentation, synthetic video summary.Therefore,
Compared with also having the video information in substantial amounts of non-interesting region in the monitor video of prior art, can more accurately it find interested
Video, and decrease the size of video.
In addition, a kind of video process apparatus that the embodiment of the present invention also provides, such as Fig. 2, described device include:
Indexing unit 201, for marking target area in video pictures according to preparatory condition;
Cutting unit 202, for according to the first preset algorithm by the Video segmentation of the target area into multiple segmentation mirrors
Head;
Extraction unit 203, for choosing the key frame of each shot segmentation according to preset rules;
Statistic unit 204, for obtaining the moving target of each shot segmentation by the second preset algorithm, to being obtained
The tracing of the movement of each moving target taken, extract each two field picture of the movement locus of each moving target;
First synthesis unit 205, for by each frame figure of the movement locus of each moving target of the target area
As corresponding to being superimposed to the movement locus of the moving target on the key frame of shot segmentation, video frequency abstract is generated.
The statistic unit 204 also includes:
Subelement is followed the trail of, for finding out the every of the target area by the Potential Prediction algorithm based on gauss hybrid models
The moving target of individual shot segmentation, to the tracing of the movement of each moving target found, extraction is found every
Each two field picture of the movement locus of the individual moving target.
The tracking subelement includes:
First computation subunit, for calculating a coordinate position pair of each shot segmentation for obtaining the target area
The pixel answered is X in time t observationt, the pixel is { X in a series of observations at different moments1, X2...,
Xt, Gaussian Profile is expressed as:
Second computation subunit, for obtaining the Gauss of all pixels point corresponding to the coordinate position of the target area
Distribution;
3rd computation subunit, for establishing Gaussian Profile set, target area described in the Gaussian Profile set memory storage
The Gaussian Profile of all pixels point of each shot segmentation in domain;
4th computation subunit, for according to default comparison rules by all height not met in the Gaussian Profile set
The pixel of this distribution saves as a two field picture of the movement locus, wherein, the default comparison rules include:Judge current
Whether one pixel of the target area at moment meetsIf, it is believed that the pixel meets institute
The Gaussian Profile in Gaussian Profile set is stated, if not, it is believed that the pixel does not meet the Gauss in the Gaussian Profile set
Distribution.
The extraction unit is specifically used for using the last frame on the time shaft of each shot segmentation as described point
Cut the key frame of camera lens, the corresponding shot segmentation of a key frame.
The cutting unit 202 includes:
First comparison subunit, for judging on a timeline adjacent two by color histogram nomography based on RGB
Whether the similarity between individual frame is less than default first threshold;
Second comparison subunit, if judging adjacent on a timeline two by color histogram nomography based on RGB
Similarity between frame is less than default first threshold, by color histogram nomography based on HSV judge it is described by based on
Whether the similarity that the similarity that RGB color histogram nomography judges is less than two frames of the first threshold is less than default the
Two threshold values;
Subelement is judged, if for passing through the color histogram based on RGB described in the color histogram nomography judgement based on HSV
The similarity that the similarity that nomography judges is less than two frames of the first threshold is less than default Second Threshold, judges two frame category
In two different camera lenses;
Subelement is set, for setting shot boundary between being judged to belonging to two frames of two different camera lenses;
Split subelement, for according to set a plurality of shot boundary by the Video segmentation of the target area into multiple
Shot segmentation.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit, the corresponding process in preceding method embodiment is may be referred to, will not be repeated here.
In addition, the flow chart and block diagram in accompanying drawing show system, method and the meter of multiple embodiments according to the present invention
Architectural framework in the cards, function and the operation of calculation machine program product.At this point, each square frame in flow chart or block diagram
Can represent a part for a module, program segment or code, the part of the module, program segment or code include one or
Multiple executable instructions for being used to realize defined logic function.It should also be noted that some as replace realization in, square frame
Middle marked function can also be with different from the order marked in accompanying drawing generation.For example, two continuous square frames are actually
It can perform substantially in parallel, they can also be performed in the opposite order sometimes, and this is depending on involved function.Also to note
Meaning, the combination of each square frame and block diagram in block diagram and/or flow chart and/or the square frame in flow chart can be with holding
Function as defined in row or the special hardware based system of action are realized, or can use specialized hardware and computer instruction
Combination realize.
A kind of computer program product of information interacting method of progress that the embodiment of the present invention is provided, including store journey
The computer-readable recording medium of sequence code, the instruction that described program code includes can be used for performing institute in previous methods embodiment
The method stated, specific implementation can be found in embodiment of the method, will not be repeated here.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit, the corresponding process in preceding method embodiment is may be referred to, will not be repeated here.
In several embodiments provided herein, it should be understood that disclosed systems, devices and methods, can be with
Realize by another way.Device embodiment described above is only schematical, for example, the division of the unit,
Only a kind of division of logic function, can there is other dividing mode when actually realizing, in another example, multiple units or component can
To combine or be desirably integrated into another system, or some features can be ignored, or not perform.It is another, it is shown or beg for
The mutual coupling of opinion or direct-coupling or communication connection can be by some communication interfaces, device or unit it is indirect
Coupling or communication connection, can be electrical, mechanical or other forms.
The unit illustrated as separating component can be or may not be physically separate, show as unit
The part shown can be or may not be physical location, you can with positioned at a place, or can also be distributed to multiple
On NE.Some or all of unit therein can be selected to realize the mesh of this embodiment scheme according to the actual needs
's.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, can also
That unit is individually physically present, can also two or more units it is integrated in a unit.
If the function is realized in the form of SFU software functional unit and is used as independent production marketing or in use, can be with
It is stored in a computer read/write memory medium.Based on such understanding, technical scheme is substantially in other words
The part to be contributed to prior art or the part of the technical scheme can be embodied in the form of software product, the meter
Calculation machine software product is stored in a storage medium, including some instructions are causing a computer equipment (can be
People's computer, server, or network equipment etc.) perform all or part of step of each embodiment methods described of the present invention.
And foregoing storage medium includes:USB flash disk, mobile hard disk, read-only storage (ROM, Read-Only Memory), arbitrary access are deposited
Reservoir (RAM, Random Access Memory), magnetic disc or CD etc. are various can be with the medium of store program codes.
It should be noted that herein, such as first and second or the like relational terms are used merely to a reality
Body or operation make a distinction with another entity or operation, and not necessarily require or imply and deposited between these entities or operation
In any this actual relation or order.Moreover, term " comprising ", "comprising" or its any other variant are intended to
Nonexcludability includes, so that process, method, article or equipment including a series of elements not only will including those
Element, but also the other element including being not expressly set out, or it is this process, method, article or equipment also to include
Intrinsic key element.In the absence of more restrictions, the key element limited by sentence " including one ... ", it is not excluded that
Other identical element in the process including the key element, method, article or equipment also be present.
The foregoing is only a specific embodiment of the invention, but protection scope of the present invention is not limited thereto, any
Those familiar with the art the invention discloses technical scope in, change or replacement can be readily occurred in, should all be contained
Cover within protection scope of the present invention.Therefore, protection scope of the present invention described should be defined by scope of the claims.
Claims (8)
1. a kind of method for processing video frequency, it is characterised in that methods described includes:
Target area is marked in video pictures according to preparatory condition;
Judge whether the similarity between two adjacent on a timeline frames is less than by the color histogram nomography based on RGB
Default first threshold;
If so, by passing through the color histogram nomography judgement based on RGB described in the color histogram nomography judgement based on HSV
Whether the similarity that similarity is less than two frames of the first threshold is less than default Second Threshold;
If so, judge that two frames belong to two different camera lenses;
Shot boundary is set between being judged to belonging to two frames of two different camera lenses;
According to set a plurality of shot boundary by the Video segmentation of the target area into multiple shot segmentations;
The key frame of each shot segmentation is chosen according to preset rules;
The moving target of each shot segmentation is obtained by the second preset algorithm, to the fortune of acquired each moving target
Dynamic track following, extract each two field picture of the movement locus of each moving target;
Each two field picture of the movement locus of each moving target of the target area is superimposed to the fortune of the moving target
Corresponding to dynamic rail mark on the key frame of shot segmentation, video frequency abstract is generated.
2. method for processing video frequency according to claim 1, it is characterised in that described to be obtained often by the second preset algorithm
The moving target of the individual shot segmentation, to the tracing of the movement of acquired each moving target, extract each motion mesh
Each two field picture of target movement locus, including:
The motion mesh of each shot segmentation of the target area is found out by the Potential Prediction algorithm based on gauss hybrid models
Mark, the fortune of each moving target found to the tracing of the movement of each moving target found, extraction
Each two field picture of dynamic rail mark.
3. method for processing video frequency according to claim 2, it is characterised in that described by based on gauss hybrid models
Potential Prediction algorithm finds out the moving target of each shot segmentation of the target area, to each motion mesh found
Target tracing of the movement, each two field picture of the movement locus of each moving target found is extracted, including:
Pixel corresponding to a coordinate position of each shot segmentation for obtaining the target area is calculated in time t observation
It is worth for Xt, the pixel is { X in a series of observations at different moments1,X2,...,Xt, Gaussian Profile is expressed as:
Obtain the Gaussian Profile of all pixels point corresponding to the coordinate position of the target area;
Gaussian Profile set is established, all pictures of each shot segmentation of target area described in the Gaussian Profile set memory storage
The Gaussian Profile of vegetarian refreshments;
All pixels for not meeting the Gaussian Profile in the Gaussian Profile set are saved as into institute according to default comparison rules
A two field picture of movement locus is stated, wherein, the default comparison rules include:Judge the one of the target area at current time
Whether individual pixel meetsIf, it is believed that the pixel meets the Gauss in the Gaussian Profile set
Distribution, if not, it is believed that the pixel does not meet the Gaussian Profile in the Gaussian Profile set.
4. method for processing video frequency according to claim 1, it is characterised in that described to choose each point according to preset rules
The key frame of camera lens is cut, including:
Key frame using the last frame on the time shaft of each shot segmentation as the shot segmentation, a key frame
A corresponding shot segmentation.
5. a kind of video process apparatus, it is characterised in that described device includes:
Indexing unit, for marking target area in video pictures according to preparatory condition;
Cutting unit, for being judged by color histogram nomography based on RGB between two adjacent on a timeline frames
Whether similarity is less than default first threshold;
If so, by passing through the color histogram nomography judgement based on RGB described in the color histogram nomography judgement based on HSV
Whether the similarity that similarity is less than two frames of the first threshold is less than default Second Threshold;
If so, judge that two frames belong to two different camera lenses;
Shot boundary is set between being judged to belonging to two frames of two different camera lenses;
According to set a plurality of shot boundary by the Video segmentation of the target area into multiple shot segmentations;
Extraction unit, for choosing the key frame of each shot segmentation according to preset rules;
Statistic unit, for obtaining the moving target of each shot segmentation by the second preset algorithm, to acquired every
The tracing of the movement of individual moving target, extract each two field picture of the movement locus of each moving target;
First synthesis unit, for each two field picture of the movement locus of each moving target of the target area to be superimposed to
Corresponding to the movement locus of the moving target on the key frame of shot segmentation, video frequency abstract is generated.
6. video process apparatus according to claim 5, it is characterised in that the statistic unit also includes:
Subelement is followed the trail of, for finding out each point of the target area by the Potential Prediction algorithm based on gauss hybrid models
The moving target of camera lens is cut, to the tracing of the movement of each moving target found, extracts each institute found
State each two field picture of the movement locus of moving target.
7. video process apparatus according to claim 6, it is characterised in that the tracking subelement includes:
First computation subunit, for calculating corresponding to a coordinate position of each shot segmentation for obtaining the target area
Pixel is X in time t observationt, the pixel is { X in a series of observations at different moments1,X2,...,Xt,
Gaussian Profile is expressed as:
Second computation subunit, for obtaining the Gauss minute of all pixels point corresponding to the coordinate position of the target area
Cloth;
3rd computation subunit, for establishing Gaussian Profile set, target area described in the Gaussian Profile set memory storage
The Gaussian Profile of all pixels point of each shot segmentation;
4th computation subunit, for dividing all Gausses not met in the Gaussian Profile set according to default comparison rules
The pixel of cloth saves as a two field picture of the movement locus, wherein, the default comparison rules include:Judge current time
A pixel of the target area whether meetIf, it is believed that the pixel meets the height
Gaussian Profile in this distributed collection, if not, it is believed that the pixel does not meet the Gaussian Profile in the Gaussian Profile set.
8. video process apparatus according to claim 5, it is characterised in that the extraction unit is specifically used for each institute
State key frame of the last frame as the shot segmentation on the time shaft of shot segmentation, the corresponding segmentation of a key frame
Camera lens.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510246765.8A CN104867161B (en) | 2015-05-14 | 2015-05-14 | A kind of method for processing video frequency and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510246765.8A CN104867161B (en) | 2015-05-14 | 2015-05-14 | A kind of method for processing video frequency and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104867161A CN104867161A (en) | 2015-08-26 |
CN104867161B true CN104867161B (en) | 2018-03-23 |
Family
ID=53912976
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510246765.8A Expired - Fee Related CN104867161B (en) | 2015-05-14 | 2015-05-14 | A kind of method for processing video frequency and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104867161B (en) |
Families Citing this family (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105869334A (en) * | 2016-05-28 | 2016-08-17 | 张维秀 | Fire alarm method and system |
CN106096496A (en) * | 2016-05-28 | 2016-11-09 | 张维秀 | A kind of fire monitoring method and system |
CN106210444B (en) * | 2016-07-04 | 2018-10-30 | 石家庄铁道大学 | Motion state self adaptation key frame extracting method |
CN106210878A (en) * | 2016-07-25 | 2016-12-07 | 北京金山安全软件有限公司 | Picture extraction method and terminal |
CN106611043B (en) * | 2016-11-16 | 2020-07-03 | 深圳市梦网视讯有限公司 | Video searching method and system |
CN106920250B (en) * | 2017-02-14 | 2019-08-13 | 华中科技大学 | Robot target identification and localization method and system based on RGB-D video |
CN107896313A (en) * | 2017-11-01 | 2018-04-10 | 无锡港湾网络科技有限公司 | A kind of underground pipe gallery wireless video monitoring system |
CN108834066A (en) * | 2018-06-27 | 2018-11-16 | 三星电子(中国)研发中心 | Method and apparatus for generating information |
CN110826365B (en) * | 2018-08-09 | 2023-06-23 | 阿里巴巴集团控股有限公司 | Video fingerprint generation method and device |
CN112889090A (en) * | 2018-08-17 | 2021-06-01 | 道特里斯艾欧公司 | System and method for performing modeling and control of a physical dynamic system using artificial intelligence |
CN109902730B (en) * | 2019-02-21 | 2021-06-01 | 国网山东省电力公司临沂供电公司 | Power transmission line broken strand detection method based on deep learning |
CN109948611B (en) * | 2019-03-14 | 2022-07-08 | 腾讯科技(深圳)有限公司 | Information area determination method, information display method and device |
CN110035320B (en) * | 2019-03-19 | 2021-06-22 | 星河视效科技(北京)有限公司 | Video advertisement loading rendering method and device |
CN110084765B (en) * | 2019-05-05 | 2021-08-06 | Oppo广东移动通信有限公司 | Image processing method, image processing device and terminal equipment |
CN110502663A (en) * | 2019-08-26 | 2019-11-26 | 西南交通大学 | A kind of rds data processing method and processing device based on artificial intelligence |
CN112712543A (en) * | 2019-10-25 | 2021-04-27 | 北京京东振世信息技术有限公司 | Article transport trajectory detection method, article transport trajectory detection device, and storage medium |
CN111008978B (en) * | 2019-12-06 | 2022-10-14 | 电子科技大学 | Video scene segmentation method based on deep learning |
CN110933520B (en) * | 2019-12-10 | 2020-10-16 | 中国科学院软件研究所 | Monitoring video display method based on spiral abstract and storage medium |
CN113392267B (en) * | 2020-03-12 | 2024-01-16 | 平湖莱顿光学仪器制造有限公司 | Method and device for generating two-dimensional microscopic video information of target object |
CN111461962A (en) * | 2020-03-27 | 2020-07-28 | 咪咕文化科技有限公司 | Image processing method, electronic equipment and computer readable storage medium |
CN111385644A (en) * | 2020-03-27 | 2020-07-07 | 咪咕文化科技有限公司 | Video processing method, electronic equipment and computer readable storage medium |
CN112347303B (en) * | 2020-11-27 | 2024-06-14 | 上海科江电子信息技术有限公司 | Media audio-visual information stream monitoring and supervision data sample and labeling method thereof |
CN113112519B (en) * | 2021-04-23 | 2023-04-18 | 电子科技大学 | Key frame screening method based on interested target distribution |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101789011A (en) * | 2009-01-26 | 2010-07-28 | 三菱电机株式会社 | The detection of similar video segments |
CN103227963A (en) * | 2013-03-20 | 2013-07-31 | 西交利物浦大学 | Static surveillance video abstraction method based on video moving target detection and tracing |
CN104063883A (en) * | 2014-07-07 | 2014-09-24 | 杭州银江智慧医疗集团有限公司 | Surveillance video abstract generating method based on combination of object and key frames |
CN104284158A (en) * | 2014-10-23 | 2015-01-14 | 南京信必达智能技术有限公司 | Event-oriented intelligent camera monitoring method |
CN104486585A (en) * | 2014-12-18 | 2015-04-01 | 深圳先进技术研究院 | Method and system for managing urban mass surveillance video based on GIS |
-
2015
- 2015-05-14 CN CN201510246765.8A patent/CN104867161B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101789011A (en) * | 2009-01-26 | 2010-07-28 | 三菱电机株式会社 | The detection of similar video segments |
CN103227963A (en) * | 2013-03-20 | 2013-07-31 | 西交利物浦大学 | Static surveillance video abstraction method based on video moving target detection and tracing |
CN104063883A (en) * | 2014-07-07 | 2014-09-24 | 杭州银江智慧医疗集团有限公司 | Surveillance video abstract generating method based on combination of object and key frames |
CN104284158A (en) * | 2014-10-23 | 2015-01-14 | 南京信必达智能技术有限公司 | Event-oriented intelligent camera monitoring method |
CN104486585A (en) * | 2014-12-18 | 2015-04-01 | 深圳先进技术研究院 | Method and system for managing urban mass surveillance video based on GIS |
Non-Patent Citations (4)
Title |
---|
Dynamic selection and effective compression of key frames for video abstraction;Xu-Dong Zhang 等;《Pattern Recognition》;20030630;第1523-1532页 * |
基于MPEG视频的镜头检测***;唐跃林;《中国优秀博硕士学位论文全文数据库 (硕士) 信息科技辑》;20050315;正文第4页1.2.3节第1段、第25页第2)节第1段、第26页第1段 * |
基于高斯混合模型的运动目标检测算法研究;马若飞;《中国优秀硕士学位论文全文数据库 信息科技辑》;20101015;正文第20页第1段 * |
视频结构挖掘的一种改进算法;杨西平, 宁亚宁;《医学信息学》;20071130;第1901-1904页 * |
Also Published As
Publication number | Publication date |
---|---|
CN104867161A (en) | 2015-08-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104867161B (en) | A kind of method for processing video frequency and device | |
Wang et al. | Integrating color and shape-texture features for adaptive real-time object tracking | |
Salman et al. | Real-time fish detection in complex backgrounds using probabilistic background modelling | |
US20230289979A1 (en) | A method for video moving object detection based on relative statistical characteristics of image pixels | |
Xiao et al. | Multiple view semantic segmentation for street view images | |
CN105141903B (en) | A kind of method for carrying out target retrieval in video based on colouring information | |
Butler et al. | Real-time adaptive foreground/background segmentation | |
CN106446015A (en) | Video content access prediction and recommendation method based on user behavior preference | |
CN105528794A (en) | Moving object detection method based on Gaussian mixture model and superpixel segmentation | |
CN101655983B (en) | Device and method for exacting dominant color | |
CN103996046A (en) | Personnel recognition method based on multi-visual-feature fusion | |
CN109344842A (en) | A kind of pedestrian's recognition methods again based on semantic region expression | |
CN110427795A (en) | A kind of property analysis method based on head photo, system and computer equipment | |
CN105139421A (en) | Video key frame extracting method of electric power system based on amount of mutual information | |
Zhang et al. | Detecting and removing visual distractors for video aesthetic enhancement | |
Yang et al. | A method of pedestrians counting based on deep learning | |
CN106096496A (en) | A kind of fire monitoring method and system | |
CN109215047B (en) | Moving target detection method and device based on deep sea video | |
Pavlov et al. | Application for video analysis based on machine learning and computer vision algorithms | |
CN109427068A (en) | Deep learning based on super-pixel automatic marking is divided into segmentation method | |
CN112487926A (en) | Scenic spot feeding behavior identification method based on space-time diagram convolutional network | |
Khryashchev et al. | Audience analysis system on the basis of face detection, tracking and classification techniques | |
Wan et al. | Illumination robust video foreground prediction based on color recovering | |
Anantharajah et al. | Quality based frame selection for face clustering in news video | |
Bian et al. | Efficient hierarchical temporal segmentation method for facial expression sequences |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
EXSB | Decision made by sipo to initiate substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180323 Termination date: 20180514 |