CN107657626A - The detection method and device of a kind of moving target - Google Patents
The detection method and device of a kind of moving target Download PDFInfo
- Publication number
- CN107657626A CN107657626A CN201610598717.XA CN201610598717A CN107657626A CN 107657626 A CN107657626 A CN 107657626A CN 201610598717 A CN201610598717 A CN 201610598717A CN 107657626 A CN107657626 A CN 107657626A
- Authority
- CN
- China
- Prior art keywords
- frame
- target
- image
- prospect
- frame set
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06T2207/30232—Surveillance
Landscapes
- Image Analysis (AREA)
Abstract
The present invention provides a kind of detection method and device of moving target, and this method includes：The first image is extracted from monitor video, described first image includes foreground image；Operation is carried out out to described first image, obtains the second image, the first object frame set of present frame is got from second image, includes multiple first object frames in the first object frame set；Closed operation is carried out to described first image, obtains the 3rd image, the second target frame set of present frame is got from the 3rd image, includes multiple second target frames in the second target frame set；The prospect frame set of present frame is obtained using the first object frame set and the second target frame set；Moving target is detected using the prospect frame set.By technical scheme, whether moving target can be accurately detected by mixing whether line, moving target are swarmed into forbidden zone or leave forbidden zone, improves accuracy, the accuracy safeguarded of moving object detection.
Description
Technical field
The present invention relates to the detection method and device of communication technical field, more particularly to a kind of moving target.
Background technology
In recent years, with computer, network and image procossing, the rapid development of transmission technology, video monitoring system
Universalness trend is more and more obvious, and video monitoring system progressively marches toward high Qinghua, intelligent, and video monitoring system can answer
For various fields, such as intelligent transportation, wisdom garden, safe city.
In video monitoring system, monitor video can be analyzed using computer vision technique, to reach automatic
Detection exception and the purpose alarmed in real time, so as to substitute artificial inspection of fixed place and time or artificial long-time viewing monitoring
The working method of video, inspection can be avoided missing, fatigue caused by the artificial monitor video of viewing for a long time be avoided, public
In terms of security protection with private site, there is huge application prospect.
At present, there is the technology analyzed in real time monitor video, set such as in monitor video and mix line, or sense
The region of interest, to substitute artificial observation point, by carrying out region intrusion detection, pedestrian's abnormality detection, local people's stream statistics, office
Portion's crowd density detection etc., when meeting certain condition, automatic output alarm.
Wherein, setting is mixed line and referred to：Straight line or curve are drawn in monitor video, to substitute the effect of gate inhibition, when
Moving target from any direction pass through when, then can export alarm automatically.Region interested is set to refer to：In monitor video
It is forbidden zone to draw a fixed region, or using the region of whole monitor video as forbidden zone, when have moving target swarm into the forbidden zone or
When person leaves the forbidden zone, then alarm can be exported automatically.
How moving target is detected whether by mixing line, how detecting whether moving target is swarmed into forbidden zone or leave taboo
Area, at present there is the defects of detection error is larger in unsuitable solution, existing scheme.
The content of the invention
The present invention provides a kind of detection method of moving target, and methods described includes：
The first image is extracted from monitor video, described first image includes foreground image；
Operation is carried out out to described first image, obtains the second image, and present frame is got from second image
First object frame set, include multiple first object frames in the first object frame set；
Closed operation is carried out to described first image, obtains the 3rd image, and present frame is got from the 3rd image
The second target frame set, include multiple second target frames in the second target frame set；
The prospect frame set of present frame is obtained using the first object frame set and the second target frame set；
Moving target is detected using the prospect frame set.
It is described that operation is carried out out to described first image, the process of the second image is obtained, is specifically included：
Etching operation is carried out to described first image by the first Erodent Algorithm, obtains the image after etching operation；
Expansive working is carried out to the image after the etching operation by the first expansion template, obtains the second image；
It is described that closed operation is carried out to described first image, the process of the 3rd image is obtained, is specifically included：
Expansive working is carried out to described first image by the second expansion template, obtains the image after expansive working；
Etching operation is carried out to the image after the expansive working by the second Erodent Algorithm, obtains the 3rd image.
The second expansion template specifically includes：
Second Erodent Algorithm specifically includes：
The mistake of the prospect frame set that present frame is obtained using the first object frame set and the second target frame set
Journey, specifically include：The prospect frame set of former frame is obtained, utilizes the first object frame set, the second target frame set
With the prospect frame set of the former frame, the prospect frame set of the present frame is obtained.
The prospect frame collection using the first object frame set, the second target frame set and the former frame
Close, obtain the process of the prospect frame set of the present frame, specifically include：For each in the second target frame set
Two target frames perform following processing, obtain the prospect frame set of present frame；
Count in the first object frame set, fall the pixel in all first object frames of the second target inframe
Number sum, obtains pixel quantity corresponding to the second target frame；Obtain area corresponding to the second target frame；Described in acquisition
Second target frame and the beeline of the prospect frame set of the former frame；
Whether judge the second target frame using pixel quantity, area, beeline corresponding to the second target frame
Meet selection condition；If meeting, the second target frame is added to the prospect frame set of present frame；If not meeting, prohibit
The second target frame is only added to the prospect frame set of present frame.
Whether judge the second target frame using pixel quantity, area, beeline corresponding to the second target frame
Meet selection condition, specifically include：
(if the pixel quantity-area described in default first numerical value *)-beeline described in default second value *, is more than
Equal to default third value, and the area is more than default 4th numerical value, it is determined that the second target frame meets the choosing
Condition is taken, otherwise, it determines the second target frame does not meet the selection condition；Wherein, the first numerical value, default second number are preset
Value, default third value, default 4th numerical value are all higher than 0.
Based on tracking box corresponding to the moving target, the mistake that moving target is detected using the prospect frame set
Journey, specifically include：The angle point for counting the tracking box falls the angle point number of each prospect frame in the prospect frame set；Such as
All prospect frames of fruit do not include the angle point of the tracking box, it is determined that the moving target is lost；It is if one or more
Individual prospect frame includes the angle point of the tracking box, then obtains comprising the most prospect frame of angle point number, and by before the acquisition
Scape frame is defined as the prospect frame to match with the moving target, and the prospect frame of the acquisition is updated into the moving target
Corresponding tracking box.
Based on tracking box corresponding to multiple moving targets, by the prospect frame of the acquisition be updated to corresponding to moving target with
The process of track frame, is specifically included：
If the coordinate of the prospect frame to match with the moving target, and the prospect frame to match with other moving targets
Coordinate is different, then the prospect frame of the acquisition is updated into tracking box corresponding to the moving target；
After the prospect frame for being defined as matching with the moving target by the prospect frame of the acquisition, methods described is also
Including：If the coordinate of the prospect frame to match with the moving target, and the seat of the prospect frame to match with other moving targets
Mark identical, then obtain the barycenter of all angle points of the moving target currently in corresponding tracking box, and using the barycenter in
The heart, update tracking box corresponding to the moving target.
After the detection moving target using the prospect frame set, methods described further comprises：
Using the position of the moving target detected, whether the moving target is detected by mixing line；And/or inspection
Survey whether the moving target is swarmed into forbidden zone or leave forbidden zone.
The present invention provides a kind of detection means of moving target, and described device specifically includes：
Extraction module, for extracting the first image from monitor video, described first image includes foreground image；
First obtains module, for carrying out out operation to described first image, obtains the second image, and from second figure
The first object frame set of present frame is got as in, includes multiple first object frames in the first object frame set；
Second obtains module, for carrying out closed operation to described first image, obtains the 3rd image, and from the 3rd figure
The second target frame set of present frame is got as in, includes multiple second target frames in the second target frame set；
3rd obtains module, for obtaining present frame using the first object frame set and the second target frame set
Prospect frame set；
Detection module, for utilizing prospect frame set detection moving target.
Described first obtains module, specifically for carrying out out operation to the first image, during obtaining the second image,
Etching operation is carried out to the first image by the first Erodent Algorithm, obtains the image after etching operation；Pass through the first expansion template
Expansive working is carried out to the image after the etching operation, obtains the second image；
Described second obtains module, specifically for carrying out closed operation to the first image, during obtaining the 3rd image,
Expansive working is carried out to the first image by the second expansion template, obtains the image after expansive working；Pass through the second Erodent Algorithm
Etching operation is carried out to the image after the expansive working, obtains the 3rd image.
The second expansion template specifically includes：
Second Erodent Algorithm specifically includes：
Described 3rd obtains module, specifically for being obtained using the first object frame set and the second target frame set
During the prospect frame set of present frame, the prospect frame set of former frame is obtained, and utilize the first object frame set, institute
The second target frame set and the prospect frame set of the former frame are stated, obtains the prospect frame set of the present frame.
Described 3rd obtains module, specifically for utilizing the first object frame set, the second target frame set
With the prospect frame set of the former frame, during the prospect frame set for obtaining the present frame, for second target
Each second target frame in frame set performs following processing, obtains the prospect frame set of present frame；Count the first object
In frame set, fall the number of pixels sum in all first object frames of the second target inframe, obtain second target
Pixel quantity corresponding to frame；Obtain area corresponding to the second target frame；Obtain the second target frame and the former frame
Prospect frame set beeline；
Whether judge the second target frame using pixel quantity, area, beeline corresponding to the second target frame
Meet selection condition；If meeting, the second target frame is added to the prospect frame set of present frame；If not meeting, prohibit
The second target frame is only added to the prospect frame set of present frame.
Described 3rd obtains module, specifically for utilizing pixel quantity corresponding to the second target frame, area, most short
During whether the second target frame meets selection condition described in Distance Judgment, if (the pixel quantity-default first numerical value *
The area)-beeline described in second value * is preset, more than or equal to default third value, and the area is more than default
4th numerical value, it is determined that the second target frame meets the selection condition, otherwise, it determines the second target frame does not meet institute
State selection condition；Wherein, the first numerical value, default second value, default third value, default 4th numerical value are preset and is all higher than 0.
The detection module, during detecting moving target in the utilization prospect frame set, based on institute
Tracking box corresponding to moving target is stated, the angle point for counting the tracking box falls each prospect frame in the prospect frame set
Angle point number；If all prospect frames do not include the angle point of the tracking box, it is determined that the moving target is lost；If
One or more prospect frame includes the angle point of the tracking box, then obtains comprising the most prospect frame of angle point number, by described in
The prospect frame of acquisition is defined as the prospect frame to match with the moving target, and the prospect frame of the acquisition is updated to described
Tracking box corresponding to moving target.
The detection module, specifically for the prospect frame of the acquisition is being updated into tracking box corresponding to moving target
During, based on tracking box corresponding to multiple moving targets, if the coordinate of the prospect frame to match with the moving target, and with
The coordinate for the prospect frame that other moving targets match is different, then the prospect frame of the acquisition is updated into the moving target pair
The tracking box answered；
The detection module, it is additionally operable to before being defined as matching with the moving target by the prospect frame of the acquisition
After scape frame, if the coordinate of the prospect frame to match with the moving target, and the prospect frame to match with other moving targets
Coordinate it is identical, then obtain the barycenter of all angle points of the moving target currently in corresponding tracking box, and with the barycenter
Centered on, update tracking box corresponding to the moving target.
The detection module, it is additionally operable to after using prospect frame set detection moving target, utilizes what is detected
The position of the moving target, whether the moving target is detected by mixing line；And/or detect whether the moving target is rushed
Enter to forbidden zone or leave forbidden zone.
Based on above-mentioned technical proposal, in the embodiment of the present invention, detection and maintenance mode to moving target are optimized,
Moving target can accurately be detected whether by mixing whether line, moving target are swarmed into forbidden zone or leave forbidden zone, improve and mix
The accuracy rate that the behaviors such as line, forbidden zone judge, accuracy, the accuracy safeguarded of moving object detection are improved, is significantly reduced multiple
Influence of the miscellaneous background to moving target recognition, reduce false drop rate.
Brief description of the drawings
, below will be to the present invention in order to clearly illustrate the embodiment of the present invention or technical scheme of the prior art
The required accompanying drawing used is briefly described in embodiment or description of the prior art, it should be apparent that, in describing below
Accompanying drawing is only some embodiments described in the present invention, for those of ordinary skill in the art, can also be according to these
Accompanying drawing obtains other accompanying drawings.
Fig. 1 is the flow chart of the detection method of the moving target in one embodiment of the present invention；
Fig. 2 is the hardware structure diagram of the headend equipment in one embodiment of the present invention；
Fig. 3 is the structure chart of the detection means of the moving target in one embodiment of the present invention.
Embodiment
Purpose in terminology used in the present invention merely for the sake of description specific embodiment, is not intended to limit the present invention.This hair
" one kind " of singulative used in bright and claims, " described " and "the" are also intended to including most forms, unless
Context clearly shows that other implications.It is also understood that term "and/or" used herein refers to comprising one or more
Associated list items purpose any or all may combine.
It will be appreciated that though various information, but this may be described using term first, second, third, etc. in the present invention
A little information should not necessarily be limited by these terms.These terms are only used for same type of information being distinguished from each other out.For example, do not departing from
In the case of the scope of the invention, the first information can also be referred to as the second information, and similarly, the second information can also be referred to as
One information.Depending on linguistic context, in addition, used word " if " can be construed to " ... when " or " when ... "
Or " in response to determining ".
For problems of the prior art, a kind of detection method of moving target is proposed in the embodiment of the present invention, should
Method can apply on headend equipment (such as web camera, analog video camera), shown in Figure 1, for the moving target
Detection method flow chart, this method may comprise steps of：
Step 101, the first image is extracted from monitor video, first image includes foreground image.
Step 102, operation is carried out out to first image, obtains the second image, and gets and works as from second image
The first object frame set of previous frame, the first object frame set is interior to include multiple first object frames.
Step 103, closed operation is carried out to first image, obtains the 3rd image, and got and work as from the 3rd image
Second target frame set of previous frame, the second target frame set is interior to include multiple second target frames.
Step 104, the prospect frame set of present frame is obtained using the first object frame set and the second target frame set.
Step 105, moving target is detected using the prospect frame set.
Wherein, above-mentioned execution sequence is an example of the embodiment of the present invention, suitable to this execution in the embodiment of the present invention
Sequence is not limited, can such as first carry out " closed operation is carried out to first image, obtains the 3rd image, and from the 3rd image
Get the second target frame set of present frame ", it is rear perform " operation is carried out out to first image, obtains the second image, and from
The first object frame set of present frame is got in second image ".In order to facilitate description, carried out by taking above-mentioned execution sequence as an example
Illustrate, execution sequence can also be adjusted in practical application.
For step 101, headend equipment can obtain the video image of successive frame, pin during acquisition monitoring video
To every frame video image, foreground image can be extracted from the video image, the foreground image is exactly the first of present frame
Image, you can to extract the first image of present frame from the video image.
In one example, foreground image can be extracted from video image using VIBE algorithms.VIBE algorithms are one
The foreground detection algorithm of kind Pixel-level, its basic thought are：A sample set is stored for each pixel, this sample set includes this
The pixel value of the individual past pixel value of pixel and its surrounding neighbours point, after this sample set is obtained, when running into one
During new pixel, just the collection point in this pixel value and sample set is contrasted, judge this new pixel value whether be
Background dot, and finally give foreground point.
, it is necessary to generate a random number during foreground image is extracted from video image using VIBE algorithms,
And the extraction of foreground image is completed using the random number.In view of the process of extraction foreground image every time, it is required to regenerate
One random number, the poor-performing of VIBE algorithms, real-time are poor.Based on this, in of the invention example, in order to improve
The performance of VIBE algorithms, meet the requirement of processing real-time, traditional VIBE algorithms can be optimized, the optimization method
For：Be pre-configured with a table of random numbers, multiple random numbers previously generated included in the table of random numbers, using VIBE algorithms from
During extracting foreground image in video image, random number can be directly inquired from the table of random numbers, and without current
Go to generate a random number again.Show through overtesting, the VIBE algorithms after optimization are compared with traditional VIBE algorithms, foreground detection effect
Fruit is suitable, and run time shorten to 1/3 or so of traditional VIBE algorithms, meets the requirement of real-time.
In one example, the foreground image extracted from the video image be exactly present frame the first image (in order to
Distinguish a convenient example), moreover, the foreground image is the foreground image of a binaryzation.
For step 102, operation is carried out out to the first image, obtains the process of the second image, included but is not limited to as follows
Mode：Etching operation is carried out to the first image by the first Erodent Algorithm, obtains the image after etching operation；Pass through the first expansion
Template carries out expansive working to the image after etching operation, obtains the second image.
In one example, the first Erodent Algorithm can include but is not limited to：First expansion template
It can include but is not limited to：Due to the first Erodent Algorithm using 3x3 and the first expansion template, therefore have
Play the role of to remove noise, eliminate wisp, in very thin junction separating objects.
For step 103, closed operation is carried out to the first image, obtains the process of the 3rd image, is included but is not limited to as follows
Mode：Expansive working is carried out to the first image by the second expansion template, obtains the image after expansive working；Pass through the second corrosion
Template carries out etching operation to the image after expansive working, obtains the 3rd image.
In one example, the second expansion template can include but is not limited to：Second is rotten
Erosion template can include but is not limited to：Wherein, the second expansion template and second Erodent Algorithm
Effect is：It can merge the target of division, the minuscule hole in filler body, and smooth object border.Due to moving target
The movement velocity of upper each structure is inconsistent, or the otherness of each structural color and background colour, can frequently result in a motion mesh
Put on and detect multiple prospect frames, such as a pedestrian target, usually will detect that two prospect frames, one, head prospect frame,
A prospect frame on trunk, the prospect frame of the two divisions should be now combined, form a complete object, and passed through
Closed operation is carried out to the first image using the second expansion template and the second Erodent Algorithm, it becomes possible to effective to merge division target.
Further, although carrying out closed operation to the first image by using the second expansion template and the second Erodent Algorithm,
It can effectively merge division target, but rock in leaf, under the interference environment such as water surface ripple, illumination variation, can usually give birth to
Into the prospect frame of mistake, such as in the case where leaf rocks environment, using the second expansion template and the second Erodent Algorithm to the first image
When carrying out closed operation, the leaf rocked may be merged into a big target, so as to cause flase drop.In order to effectively detect
Real moving target, operation can be carried out open to the first image by using the first expansion template and the first Erodent Algorithm, and
By closed operation and open operation and combine, so as to filter out the influence of environment above factor, greatly improve moving object detection
Accuracy.
In one example, for step 102, connected region algorithm can be used to carry out Objective extraction to the second image,
First object frame set is got, the first object frame set is designated as RectA, and includes multiple the in the first object frame set
One target frame, such as A1、A2、AmDeng, therefore, RectA={ A1,A2,…,Am}。
In one example, for step 103, connected region algorithm can be used to carry out Objective extraction to the 3rd image,
The second target frame set is got, the second target frame set is designated as RectB, and includes multiple the in the second target frame set
Two target frames, such as B1、B2、BnDeng, therefore, RectB={ B1,B2,…,Bn}。
Wherein, the set of first object frame and the second target frame set are target frame set corresponding to present frame.
Wherein, for each first object frame A in first object frame seti={ xi,yi,wi,hi(i=1,2 ...,
M), i.e., first object frame can be by top left co-ordinate (xi,yi) and first object frame wide height (wi,hi) uniquely determine, moreover,
First object frame AiThe foreground pixel number inside included is Pi(i=1,2 ..., m).For example, first object frame A1Before inside including
Scape number of pixels is P1.Similarly, for each second target frame B in the second target frame seti(i=1,2 ..., n), it is realized
Mode is similar with the implementation of first object frame, will not be repeated here.
Wherein, connected region algorithm is the technology commonly used in image procossing, for detecting the region of UNICOM in foreground image,
The effect of target area detection is served as in many tracing detection algorithms, this is no longer repeated in detail.
For step 104, the first object frame set and the prospect frame of the second target frame set acquisition present frame are utilized
The process of set, following manner can be included but is not limited to：The prospect frame set of former frame is obtained, and utilizes the first object frame
Set, the second target frame set and the prospect frame set of the former frame, obtain the prospect frame set of present frame.Further,
Using the prospect frame set of the set of first object frame, the second target frame set and the former frame, the prospect of present frame is obtained
The process of frame set, following manner can be included but is not limited to：For each second target frame in the second target frame set
Following processing is performed, obtains the prospect frame set of present frame；Count in the first object frame set, fall in the second target inframe
All first object frames number of pixels sum, obtain pixel quantity corresponding to the second target frame；Obtain second target
Area corresponding to frame；Obtain the beeline of the prospect frame set of the second target frame and former frame；Utilize the second target frame
Corresponding pixel quantity, area, beeline judge whether the second target frame meets selection condition；If meeting, by this
Two target frames are added to the prospect frame set of present frame；If not meeting, forbid the second target frame being added to present frame
Prospect frame set.
In one example, for " counting in the first object frame set, falling all first in the second target inframe
The number of pixels sum of target frame, obtain pixel quantity corresponding to the second target frame " process, for the second target frame set
Each second target frame B in RectBi(i=1,2 ..., n), count in first object frame set RectA, fall in the second target
Frame BiNumber of pixels (i.e. connected region number of pixels) sum of interior all first object frames, the number of pixels sum are exactly
Two target frame BiCorresponding pixel quantity SPi.If for example, first object frame A in first object frame set RectA2, A5,
A7Fall in the second target frame BiIt is interior, then SPi=P2+P5+P7, P2For first object frame A2The foreground pixel number inside included, P5For
First object frame A5The foreground pixel number inside included, P7For first object frame A7The foreground pixel number inside included.
In one example, for the process of " obtaining area corresponding to the second target frame ", because the second target frame can
With by top left co-ordinate (xi,yi) and the second target frame wide height (wi,hi) uniquely determine, therefore the second target frame can be known
Each second target frame B in set RectBiThe width of (i=1,2 ..., n) is high, and based on the second target frame B of the wide high calculatingi
Area, by the second target frame BiArea be designated as Areai(i=1,2 ..., n).
In one example, for the mistake of " beeline for obtaining the prospect frame set of the second target frame and former frame "
Journey, it is assumed that the prospect frame collection of former frame is combined into RectC={ C1,C2,…,Ck, then it can utilize preset algorithm (such as Euclidean distance
Algorithm) calculate the second target frame BiWith the distance of each prospect frame in the prospect frame set of former frame, such as the second target frame BiWith
Prospect frame C1Distance, the second target frame BiWith prospect frame C2Distance, by that analogy, and one is chosen from the distance calculated
Individual minimum range, the minimum range are exactly the second target frame BiWith the beeline of prospect frame set, s can be designated asi.Wherein, may be used
To obtain the prospect frame set of former frame using the first object frame set of former frame and the second target frame set, this process is no longer
Repeat.
In one example, second mesh is judged using pixel quantity, area, beeline corresponding to the second target frame
Whether mark frame meets the process of selection condition, can include but is not limited to following manner：If (the pixel quantity-default first
Area described in numerical value *)-beeline described in second value * is preset, more than or equal to default third value, and the area is big
In default 4th numerical value, it is determined that the second target frame meets selection condition, otherwise, it determines the second target frame does not meet selection
Condition；Wherein, default first numerical value, the default second value, the default third value, default 4th numerical value
It is all higher than 0.
In one example, if the second target frame BiMeet equation below simultaneously, illustrate the second target frame BiMeet selection bar
Part, it is added to prospect frame set.If the second target frame BiIt is unsatisfactory for any one in equation below or multiple, explanation
Second target frame BiSelection condition is not met, is not added to prospect frame set.
In formula 1, due to the second target frame BiBelong to the second target frame set RectB, therefore formula 1 meets.
In formula 2, SPiIt is exactly the second target frame BiCorresponding pixel quantity, Areai(i=1,2 ..., n) it is exactly the second mesh
Mark frame BiArea, siIt is exactly the second target frame BiWith the beeline of prospect frame set, λ, μ be respectively default first numerical value and
Default second value, ε are default third value.Wherein, λ, μ and ε can be respectively parameter coefficient and threshold condition, and can be all
More than 0, in actual applications, above-mentioned λ, μ and ε value, it can be configured, be will not be repeated here according to practical experience.It is public
Formula 2 is made up of two parts, first half (SPi-λAreai) proportion that valid pixel accounts for target has been reacted, its value is bigger, then illustrates
Valid pixel is more, and its value is smaller, then illustrates that valid pixel is fewer.Further, valid pixel is more, illustrates to pass through opening operation
Smaller with the difference of the first object frame set obtained after closed operation and the second target frame set, target is more accurate.siReact
Current goal most matches the distance of target with previous frame, and the smaller explanation of its value more matches, and the bigger explanation of its value more mismatches.Enter one
Step, when valid pixel is more in a target and is more matched with former frame target, then illustrate that the target is more accurate.To sum up institute
State, (SPi-λAreai)-μsiValue it is bigger, the target detected is also more accurate, the condition of formula 2, can effectively reduce
The interference of the environment such as leaf rocks, water surface ripple.
In formula 3, Areai(i=1,2 ..., n) it is exactly the second target frame BiArea, β is default 4th numerical value, in reality
In the application of border, β value can be configured according to practical experience, will not be repeated here.By setting rational β value,
The second small target frame of the filter area of formula 3 can be passed through.
In formula 4 and formula 5, wiFor the width of the second target frame, hiFor the height of the second target frame, wjFor the width of prospect frame, hj
For the height of prospect frame, the prospect frame is the minimum prospect frame with the second target frame distance in the prospect frame set of former frame.
By formula 4 and formula 5, the second target frame of present frame and the prospect frame of former frame can be contrasted, so as to filter out
The the second target frame acutely changed occurs for deformation.
After above-mentioned processing, for each second target frame BiIf the second target frame BiMeet selection condition, then will
It is added to prospect frame set；If the second target frame BiSelection condition is not met, then forbids being added to prospect frame set, this
Sample, it is possible to obtain a prospect frame set, and prospect frame set RectD={ D1,D2,…,Dt}.For example, when the second mesh
Mark frame B1、B3、B5、B7、B9Meet selection condition, then by the second target frame B1、B3、B5、B7、B9It is added to prospect frame set, when
Two target frame B2、B4、B6、B8Selection condition is not met, then is forbidden the second target frame B2、B4、B6、B8It is added to prospect frame set,
Therefore, the second target frame B is included in the set of prospect frame1、B3、B5、B7、B9, in these the second target frames i.e. prospect frame set
Prospect frame, subsequently illustrated by taking prospect frame as an example.
For step 105, the process of moving target is detected using the prospect frame set, such as lower section can be included but is not limited to
Formula：Based on tracking box corresponding to moving target, the angle point for counting the tracking box falls each prospect frame in prospect frame set
Angle point number.If all prospect frames do not include the angle point of the tracking box, it is determined that the moving target is lost.If one
Individual or multiple prospect frames include the angle point of the tracking box, then obtain comprising the most prospect frame of angle point number, and obtained described
The prospect frame taken is defined as the prospect frame to match with the moving target, and the prospect frame of the acquisition is updated into the fortune
Tracking box corresponding to moving-target.
Further, during the prospect frame of the acquisition is updated into tracking box corresponding to the moving target,
Based on multiple tracking box corresponding to multiple moving targets (each moving target corresponds to a unique tracking box), if with the fortune
The coordinate for the prospect frame that moving-target matches, and it is different from the coordinate for the prospect frame that other moving targets match, obtained described
The prospect frame taken is updated to tracking box corresponding to the moving target.
Further, after the prospect frame of the acquisition is updated into tracking box corresponding to the moving target, if with
The coordinate for the prospect frame that the moving target matches, and it is identical with the coordinate for the prospect frame that other moving targets match, then
The barycenter of all angle points of the moving target currently in corresponding tracking box can also be obtained, and centered on the barycenter,
Update tracking box corresponding to the moving target.
In one example, it can use and target following is carried out based on pyramidal LK (Lucas Kanade) algorithm.Pin
To each moving target, a tracking box can be all corresponded to, in an initial condition, tracking box can be matched somebody with somebody according to being actually needed
Put, in subsequent process, tracking box can be updated according to prospect frame.
For moving target 1, it is assumed that tracking box is tracking box 1, and the angle point of tracking box 1 is carried out more first by LK algorithms
Newly, specific update mode repeats no more.Assuming that the set of prospect frame includes the angle of prospect frame 1 and prospect frame 2, then statistical trace frame 1
Point fall prospect frame 1 angle point number and fall the angle point number in prospect frame 2.If prospect frame 1 and prospect frame 2 do not include with
The angle point of track frame 1, it is determined that moving target 1 is lost, and moving target 1 is added to losing in target.If prospect frame 1 and/or
Prospect frame 2 includes the angle point of tracking box 1, then obtains comprising the most prospect frame of angle point number, as prospect frame 1 includes tracking box 1
15 angle points, prospect frame 2 includes 10 angle points of tracking box 1, then the prospect frame got is the prospect frame 1, by prospect frame 1
It is defined as the prospect frame to match with moving target 1, and prospect frame 1 is updated to the tracking box of moving target 1.In subsequent process
In, the tracking box 1 of moving target 1 is exactly the prospect frame 1.
In one example, it is contemplated that a kind of abnormal conditions, i.e., the prospect to match with two or more moving targets
Frame is same, then illustrates that a prospect inframe includes two or more moving targets, in the case, it is impossible to before this
Scape frame is updated to the tracking box of the two or multiple moving targets.
In the above example, it has been determined that the prospect frame to match with moving target 1 is prospect frame 1, it is assumed that is also existed
Moving target 2, and the tracking box of moving target 2 is tracking box 2, prospect frame 1 is updated to moving target 1 tracking box it
Before, it can also be handled as follows：The angle point of statistical trace frame 2 fall prospect frame 1 angle point number and fall at the angle of prospect frame 2
Count out.If prospect frame 1 and/or prospect frame 2 include the angle point of tracking box 2, obtain comprising the most prospect of angle point number
Frame.Situation one, the prospect frame that gets of hypothesis are prospect frame 2, due to the coordinate of prospect frame 1 to match with moving target 1, and
It is different from the coordinate for the prospect frame 2 that moving target 2 matches, therefore, prospect frame 1 is updated to the tracking box of moving target 1, and
Prospect frame 2 is updated to the tracking box of moving target 2.Situation two, the prospect frame that gets of hypothesis are prospect frame 1, due to fortune
The coordinate for the prospect frame 1 that moving-target 1 matches, and it is different from the coordinate for the prospect frame 1 that moving target 2 matches, i.e., the two is
Same prospect frame, it is thus impossible to which prospect frame 1 to be updated to the tracking box of moving target 1, prospect frame 1 can not be updated to
The tracking box of moving target 2, but with all angles in tracking box 1 corresponding to moving target 1 (tracking box 1 before not updating)
Centered on the barycenter of point, to update the tracking box of moving target 1, and with tracking box 2 corresponding to moving target 2 (before not updating
Tracking box 2) in all angle points barycenter centered on, to update the tracking box of moving target 2.
, can be with during the tracking box for updating moving target centered on the barycenter of all angle points in tracking box
The barycenter O (x, y) of all angle points in the tracking box is calculated according to equation below, and centered on barycenter O (x, y), with this
Rectangle determined by the width of tracking box, height updates the tracking box of moving target.In the formula, (x1,y1),(x2,y2),…
(xm,ym) for the coordinate of all angle points in the tracking box, m is angle point quantity.
Based on above-mentioned implementation, tracking target is corrected by using prospect frame so that target is safeguarded more accurate
Really, it is stable, when moving target interlocks, remain to effectively safeguard the movement locus of target.
In one example, after step 105, the position of the moving target detected, detection motion mesh can also be utilized
Whether mark is by mixing line；And/or whether detection moving target is swarmed into forbidden zone or leaves forbidden zone.
Wherein, the line of mixing refers to：Straight line or curve are drawn in monitor video, to substitute the effect of gate inhibition, when
Moving target from any direction pass through when, then can export alarm automatically.The forbidden zone refers to：It is fixed one to be drawn in monitor video
Region is forbidden zone, or using the region of whole monitor video as forbidden zone, when having moving target to swarm into the forbidden zone or leave this
During forbidden zone, then alarm can be exported automatically.
Whether to detect moving target by exemplified by mixing line, first, based on user configuration, initial direction can be got
Vector sum rule line equation, it is assumed that direction vector is<μ1,ν1>, regular line equation is y=ax+b.
By the processing of above-mentioned steps 101- steps 105, the tracking box of the moving target of arbitrary frame can be got, and is selected
Take two tracking box for same moving target for being separated by P frames.Such as it is directed to moving target 1, it is assumed that P 6, then can choose
The tracking box of the tracking box of 2 frames and the 8th frame.Assuming that the center point coordinate of the tracking box of the 2nd frame is (x1,y1), the tracking of the 8th frame
The center point coordinate of frame is (x2,y2), then the direction of motion vector of moving target 1 is<μ2,ν2>, and<μ2,ν2>Can be by as follows
Formula calculates：
Furthermore, it is possible to utilize equation below calculated direction vector<μ1,ν1>And direction vector<μ2,ν2>Angle theta：
On this basis, if 90 ° of θ ＜, it can illustrate that the direction of motion of moving target meets that the condition of line is mixed in triggering.
If moreover, meet condition (ax1+b)(ax2+ b) ＜ 0, a and b are default values, and a and b are positive numbers, then can illustrate
Moving target touches line.When the direction of motion of moving target meets condition, and moving target touches line, then motion mesh can be detected
Mark can trigger alarm by mixing line.
Based on above-mentioned technical proposal, in the embodiment of the present invention, detection and maintenance mode to moving target are optimized,
Moving target can accurately be detected whether by mixing whether line, moving target are swarmed into forbidden zone or leave forbidden zone, improve and mix
The accuracy rate that the behaviors such as line, forbidden zone judge, improves accuracy, the accuracy safeguarded of moving object detection, and greatly subtracts
Influence of few complex background (such as leaf rocks, in the environment of the interference of water surface wave, illumination variation) to moving target recognition, has
The interleaving problem solved in target motion process of effect, reduce false drop rate.Moreover, the above method can also be in low and middle-end chip
Run on (i.e. headend equipment uses low and middle-end chip), so as to greatly reduce the index and cost intelligently deployed to ensure effective monitoring and control of illegal activities.
Based on the inventive concept same with the above method, a kind of detection dress of moving target is also provided in the embodiment of the present invention
Put, can apply in headend equipment, the detection means of the moving target can be realized by software, can also by hardware or
The mode of person's software and hardware combining is realized.It is by where it as the device on a logical meaning exemplified by implemented in software
Corresponding computer program instructions in nonvolatile memory are read what operation in internal memory was formed by the processor of headend equipment.
For hardware view, as shown in Fig. 2 one kind for the headend equipment where the detection means of moving target proposed by the present invention
Hardware structure diagram, in addition to the processor shown in Fig. 2, network interface, internal memory and nonvolatile memory, headend equipment may be used also
With including other hardware, the forwarding chip of such as responsible processing message；From hardware configuration the headend equipment be also possible to be
Distributed apparatus, multiple interface cards may be included, to carry out the extension of Message processing in hardware view.
As shown in figure 3, the structure chart of the detection means for moving target proposed by the present invention, the device includes：
Extraction module 11, for extracting the first image from monitor video, described first image includes foreground image；First
Module 12 is obtained, for carrying out out operation to described first image, obtains the second image, and got from second image
The first object frame set of present frame, the first object frame set is interior to include multiple first object frames；Second obtains module 13,
For carrying out closed operation to described first image, the 3rd image is obtained, and gets from the 3rd image the of present frame
Two target frame set, the second target frame set is interior to include multiple second target frames；3rd obtains module 14, for utilizing institute
State the set of first object frame and the second target frame set obtains the prospect frame set of present frame；Detection module 15, for profit
Moving target is detected with the prospect frame set.
Described first obtains module 12, specifically for carrying out out operation to the first image, obtains the process of the second image
In, etching operation is carried out to the first image by the first Erodent Algorithm, obtains the image after etching operation；Pass through the first bulging die
Plate carries out expansive working to the image after the etching operation, obtains the second image；
Described second obtains module 13, specifically for carrying out closed operation to the first image, obtains the process of the 3rd image
In, expansive working is carried out to the first image by the second expansion template, obtains the image after expansive working；Pass through the second corrosion mode
Plate carries out etching operation to the image after the expansive working, obtains the 3rd image.
The second expansion template specifically includes：
Second Erodent Algorithm specifically includes：
In one example, the described 3rd module 14 is obtained, specifically for utilizing the first object frame set and the
During two target frame set obtain the prospect frame set of present frame, the prospect frame set of former frame is obtained, and described in utilization
The prospect frame set of the set of first object frame, the second target frame set and the former frame, before obtaining the present frame
Scape frame set.
In one example, the described 3rd module 14 is obtained, specifically for utilizing the first object frame set, described
Second target frame set and the prospect frame set of the former frame, during the prospect frame set for obtaining the present frame, pin
Following processing is performed to each second target frame in the second target frame set, obtains the prospect frame set of present frame；
Count in the first object frame set, fall the pixel in all first object frames of the second target inframe
Number sum, obtains pixel quantity corresponding to the second target frame；Obtain area corresponding to the second target frame；Described in acquisition
Second target frame and the beeline of the prospect frame set of the former frame；
Whether judge the second target frame using pixel quantity, area, beeline corresponding to the second target frame
Meet selection condition；If meeting, the second target frame is added to the prospect frame set of present frame；If not meeting, prohibit
The second target frame is only added to the prospect frame set of present frame.
Described 3rd obtains module 14, specifically for utilizing pixel quantity corresponding to the second target frame, area, most
During short distance judges whether the second target frame meets selection condition, if (the pixel quantity-default first number
Area described in value *)-beeline described in second value * is preset, more than or equal to default third value, and the area is more than
Default 4th numerical value, it is determined that the second target frame meets the selection condition, otherwise, it determines the second target frame is not inconsistent
Close the selection condition；Wherein, the first numerical value, default second value, default third value, default 4th numerical value is preset to be all higher than
0。
The detection module 15, specifically for during using prospect frame set detection moving target, being based on
Tracking box corresponding to the moving target, the angle point for counting the tracking box fall each prospect frame in the prospect frame set
Angle point number；If all prospect frames do not include the angle point of the tracking box, it is determined that the moving target is lost；If
There is the angle point that one or more prospect frame includes the tracking box, then obtain comprising the most prospect frame of angle point number, and will
The prospect frame of the acquisition is defined as the prospect frame to match with the moving target, and the prospect frame of the acquisition is updated into institute
State tracking box corresponding to moving target.
In one example, the detection module 15, specifically for being updated to the prospect frame of the acquisition to move mesh
During tracking box corresponding to mark, based on tracking box corresponding to multiple moving targets, if match with the moving target
The coordinate of prospect frame, and it is different from the coordinate for the prospect frame that other moving targets match, then by the prospect frame of the acquisition more
It is newly tracking box corresponding to the moving target；
The detection module, it is additionally operable to before being defined as matching with the moving target by the prospect frame of the acquisition
After scape frame, if the coordinate of the prospect frame to match with the moving target, and the prospect frame to match with other moving targets
Coordinate it is identical, then obtain the barycenter of all angle points of the moving target currently in corresponding tracking box, and with the barycenter
Centered on, update tracking box corresponding to the moving target.
In one example, the detection module 15, be additionally operable to using the prospect frame set detection moving target it
Afterwards, using the position of the moving target detected, whether the moving target is detected by mixing line；And/or described in detection
Whether moving target is swarmed into forbidden zone or leaves forbidden zone.
Wherein, the modules of apparatus of the present invention can be integrated in one, and can also be deployed separately.Above-mentioned module can close
And be a module, multiple submodule can also be further split into.
Through the above description of the embodiments, those skilled in the art can be understood that the present invention can be by
Software adds the mode of required general hardware platform to realize, naturally it is also possible to which by hardware, but the former is more in many cases
Good embodiment.Based on such understanding, what technical scheme substantially contributed to prior art in other words
Part can be embodied in the form of software product, and the computer software product is stored in a storage medium, if including
It is dry to instruct to cause a computer equipment (be personal computer, server, or network equipment etc.) to perform this hair
Method described in bright each embodiment.It will be appreciated by those skilled in the art that accompanying drawing is the schematic diagram of a preferred embodiment,
Module or flow in accompanying drawing are not necessarily implemented necessary to the present invention.
It will be appreciated by those skilled in the art that the module in device in embodiment can describe be divided according to embodiment
It is distributed in the device of embodiment, respective change can also be carried out and be disposed other than in one or more devices of the present embodiment.On
The module for stating embodiment can be merged into a module, can also be further split into multiple submodule.The embodiments of the present invention
Sequence number is for illustration only, does not represent the quality of embodiment.
Disclosed above is only several specific embodiments of the present invention, and still, the present invention is not limited to this, any ability
What the technical staff in domain can think change should all fall into protection scope of the present invention.
Claims (10)
1. a kind of detection method of moving target, it is characterised in that methods described includes：
The first image is extracted from monitor video, described first image includes foreground image；
Operation is carried out out to described first image, obtains the second image, and gets from second image the of present frame
One target frame set, the first object frame set is interior to include multiple first object frames；
Closed operation is carried out to described first image, obtains the 3rd image, and gets from the 3rd image the of present frame
Two target frame set, the second target frame set is interior to include multiple second target frames；
The prospect frame set of present frame is obtained using the first object frame set and the second target frame set；
Moving target is detected using the prospect frame set.
2. according to the method for claim 1, it is characterised in that
It is described that operation is carried out out to described first image, the process of the second image is obtained, is specifically included：
Etching operation is carried out to described first image by the first Erodent Algorithm, obtains the image after etching operation；
Expansive working is carried out to the image after the etching operation by the first expansion template, obtains the second image；
It is described that closed operation is carried out to described first image, the process of the 3rd image is obtained, is specifically included：
Expansive working is carried out to described first image by the second expansion template, obtains the image after expansive working；
Etching operation is carried out to the image after the expansive working by the second Erodent Algorithm, obtains the 3rd image.
3. according to the method for claim 2, it is characterised in that
The second expansion template specifically includes：
Second Erodent Algorithm specifically includes：
4. according to the method for claim 1, it is characterised in that described to utilize the first object frame set and the second target
Frame set obtains the process of the prospect frame set of present frame, specifically includes：
The prospect frame set of former frame is obtained, and utilizes the first object frame set, the second target frame set and described
The prospect frame set of former frame, obtain the prospect frame set of the present frame.
5. according to the method for claim 4, it is characterised in that
The prospect frame set using the first object frame set, the second target frame set and the former frame, obtain
The process of the prospect frame set of the present frame is obtained, is specifically included：For each second mesh in the second target frame set
Mark frame and perform following processing, obtain the prospect frame set of present frame；
Count in the first object frame set, fall all first object frames of the second target inframe number of pixels it
With obtain pixel quantity corresponding to the second target frame；Obtain area corresponding to the second target frame；Obtain described second
Target frame and the beeline of the prospect frame set of the former frame；
Judge whether the second target frame meets using pixel quantity, area, beeline corresponding to the second target frame
Selection condition；If meeting, the second target frame is added to the prospect frame set of present frame；If not meeting, forbid by
The second target frame is added to the prospect frame set of present frame.
6. according to the method for claim 5, it is characterised in that utilize pixel quantity, face corresponding to the second target frame
Product, beeline judge whether the second target frame meets selection condition, specifically include：
(if the pixel quantity-area described in default first numerical value *)-beeline described in default second value *, is more than or equal to
Default third value, and the area is more than default 4th numerical value, it is determined that and the second target frame meets the selection bar
Part, otherwise, it determines the second target frame does not meet the selection condition；Wherein, preset the first numerical value, default second value,
Default third value, default 4th numerical value are all higher than 0.
7. according to the method for claim 1, it is characterised in that based on tracking box corresponding to the moving target, the profit
With the process of prospect frame set detection moving target, specifically include：
The angle point for counting the tracking box falls the angle point number of each prospect frame in the prospect frame set；
If all prospect frames do not include the angle point of the tracking box, it is determined that the moving target is lost；
If one or more prospect frame includes the angle point of the tracking box, then obtain comprising the most prospect of angle point number
Frame, and the prospect frame that the prospect frame of the acquisition is defined as matching with the moving target, and by the prospect of the acquisition
Frame is updated to tracking box corresponding to the moving target.
8. according to the method for claim 7, it is characterised in that based on tracking box corresponding to multiple moving targets, by described in
The prospect frame of acquisition is updated to the process of tracking box corresponding to moving target, specifically includes：
If the coordinate of the prospect frame to match with the moving target, and the coordinate of the prospect frame to match with other moving targets
Difference, then the prospect frame of the acquisition is updated to tracking box corresponding to the moving target；
After the prospect frame for being defined as matching with the moving target by the prospect frame of the acquisition, methods described is also wrapped
Include：If the coordinate of the prospect frame to match with the moving target, and the coordinate of the prospect frame to match with other moving targets
It is identical, then the barycenter of all angle points of the moving target currently in corresponding tracking box is obtained, and centered on the barycenter,
Update tracking box corresponding to the moving target.
9. according to the method for claim 1, it is characterised in that
After the detection moving target using the prospect frame set, methods described further comprises：
Using the position of the moving target detected, whether the moving target is detected by mixing line；And/or detection institute
State whether moving target is swarmed into forbidden zone or leave forbidden zone.
10. a kind of detection means of moving target, it is characterised in that described device specifically includes：
Extraction module, for extracting the first image from monitor video, described first image includes foreground image；
First obtains module, for carrying out out operation to described first image, obtains the second image, and from second image
The first object frame set of present frame is got, includes multiple first object frames in the first object frame set；
Second obtains module, for carrying out closed operation to described first image, obtains the 3rd image, and from the 3rd image
The second target frame set of present frame is got, includes multiple second target frames in the second target frame set；
3rd obtains module, before obtaining present frame using the first object frame set and the second target frame set
Scape frame set；
Detection module, for utilizing prospect frame set detection moving target.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610598717.XA CN107657626B (en) | 2016-07-25 | 2016-07-25 | Method and device for detecting moving target |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610598717.XA CN107657626B (en) | 2016-07-25 | 2016-07-25 | Method and device for detecting moving target |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107657626A true CN107657626A (en) | 2018-02-02 |
| CN107657626B CN107657626B (en) | 2021-06-01 |
Family
ID=61126724
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610598717.XA Active CN107657626B (en) | 2016-07-25 | 2016-07-25 | Method and device for detecting moving target |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107657626B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108460353A (en) * | 2018-03-07 | 2018-08-28 | 杭州晶智能科技有限公司 | A kind of infrared image human body detecting method based on sequential tracking |
| CN108932496A (en) * | 2018-07-03 | 2018-12-04 | 北京佳格天地科技有限公司 | The quantity statistics method and device of object in region |
| CN110751134A (en) * | 2019-12-23 | 2020-02-04 | 长沙智能驾驶研究院有限公司 | Target detection method, storage medium and computer device |
| CN111079694A (en) * | 2019-12-28 | 2020-04-28 | 神思电子技术股份有限公司 | Counter assistant job function monitoring device and method |
| CN111091022A (en) * | 2018-10-23 | 2020-05-01 | 宏碁股份有限公司 | Machine vision efficiency evaluation method and system |
| CN113496500A (en) * | 2020-04-02 | 2021-10-12 | 杭州萤石软件有限公司 | Out-of-range detection method and device and storage medium |
| CN114866692A (en) * | 2022-04-19 | 2022-08-05 | 合肥富煌君达高科信息技术有限公司 | Image output method and system of large-resolution monitoring camera |
| CN115984335A (en) * | 2023-03-20 | 2023-04-18 | 华南农业大学 | Method for acquiring characteristic parameters of fog drops based on image processing |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101847265A (en) * | 2010-04-20 | 2010-09-29 | 上海理工大学 | Method for extracting moving objects and partitioning multiple objects used in bus passenger flow statistical system |
| CN102096925A (en) * | 2010-11-26 | 2011-06-15 | 中国科学院上海技术物理研究所 | Real-time closed loop predictive tracking method of maneuvering target |
| CN102509086A (en) * | 2011-11-22 | 2012-06-20 | 西安理工大学 | Pedestrian object detection method based on object posture projection and multi-features fusion |
| CN103093197A (en) * | 2013-01-15 | 2013-05-08 | 信帧电子技术（北京）有限公司 | Monitoring method and system for recognizing hanging behavior |
| 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 |
| CN103425960A (en) * | 2012-05-25 | 2013-12-04 | 信帧电子技术（北京）有限公司 | Method for detecting fast-moving objects in video |
| CN103778645A (en) * | 2014-01-16 | 2014-05-07 | 南京航空航天大学 | Circular target real-time tracking method based on images |
| CN103810499A (en) * | 2014-02-25 | 2014-05-21 | 南昌航空大学 | Application for detecting and tracking infrared weak object under complicated background |
| CN104331905A (en) * | 2014-10-31 | 2015-02-04 | 浙江大学 | Surveillance video abstraction extraction method based on moving object detection |
| CN104599502A (en) * | 2015-02-13 | 2015-05-06 | 重庆邮电大学 | Method for traffic flow statistics based on video monitoring |
| CN105224912A (en) * | 2015-08-31 | 2016-01-06 | 电子科技大学 | Based on the video pedestrian detection and tracking method of movable information and Track association |
| CN105357594A (en) * | 2015-11-19 | 2016-02-24 | 南京云创大数据科技股份有限公司 | Massive video abstraction generation method based on cluster and H264 video concentration algorithm |
| CN105678811A (en) * | 2016-02-25 | 2016-06-15 | 上海大学 | Motion-detection-based human body abnormal behavior detection method |
-
2016
- 2016-07-25 CN CN201610598717.XA patent/CN107657626B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101847265A (en) * | 2010-04-20 | 2010-09-29 | 上海理工大学 | Method for extracting moving objects and partitioning multiple objects used in bus passenger flow statistical system |
| CN102096925A (en) * | 2010-11-26 | 2011-06-15 | 中国科学院上海技术物理研究所 | Real-time closed loop predictive tracking method of maneuvering target |
| CN102509086A (en) * | 2011-11-22 | 2012-06-20 | 西安理工大学 | Pedestrian object detection method based on object posture projection and multi-features fusion |
| CN103425960A (en) * | 2012-05-25 | 2013-12-04 | 信帧电子技术（北京）有限公司 | Method for detecting fast-moving objects in video |
| CN103093197A (en) * | 2013-01-15 | 2013-05-08 | 信帧电子技术（北京）有限公司 | Monitoring method and system for recognizing hanging behavior |
| 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 |
| CN103778645A (en) * | 2014-01-16 | 2014-05-07 | 南京航空航天大学 | Circular target real-time tracking method based on images |
| CN103810499A (en) * | 2014-02-25 | 2014-05-21 | 南昌航空大学 | Application for detecting and tracking infrared weak object under complicated background |
| CN104331905A (en) * | 2014-10-31 | 2015-02-04 | 浙江大学 | Surveillance video abstraction extraction method based on moving object detection |
| CN104599502A (en) * | 2015-02-13 | 2015-05-06 | 重庆邮电大学 | Method for traffic flow statistics based on video monitoring |
| CN105224912A (en) * | 2015-08-31 | 2016-01-06 | 电子科技大学 | Based on the video pedestrian detection and tracking method of movable information and Track association |
| CN105357594A (en) * | 2015-11-19 | 2016-02-24 | 南京云创大数据科技股份有限公司 | Massive video abstraction generation method based on cluster and H264 video concentration algorithm |
| CN105678811A (en) * | 2016-02-25 | 2016-06-15 | 上海大学 | Motion-detection-based human body abnormal behavior detection method |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108460353A (en) * | 2018-03-07 | 2018-08-28 | 杭州晶智能科技有限公司 | A kind of infrared image human body detecting method based on sequential tracking |
| CN108460353B (en) * | 2018-03-07 | 2021-08-20 | 深圳市恒天伟焱科技股份有限公司 | Infrared image human body detection method based on time sequence tracking |
| CN108932496A (en) * | 2018-07-03 | 2018-12-04 | 北京佳格天地科技有限公司 | The quantity statistics method and device of object in region |
| CN108932496B (en) * | 2018-07-03 | 2022-03-25 | 北京佳格天地科技有限公司 | Method and device for counting number of target objects in area |
| CN111091022A (en) * | 2018-10-23 | 2020-05-01 | 宏碁股份有限公司 | Machine vision efficiency evaluation method and system |
| CN110751134A (en) * | 2019-12-23 | 2020-02-04 | 长沙智能驾驶研究院有限公司 | Target detection method, storage medium and computer device |
| CN110751134B (en) * | 2019-12-23 | 2020-05-12 | 长沙智能驾驶研究院有限公司 | Target detection method, target detection device, storage medium and computer equipment |
| CN111079694A (en) * | 2019-12-28 | 2020-04-28 | 神思电子技术股份有限公司 | Counter assistant job function monitoring device and method |
| CN113496500A (en) * | 2020-04-02 | 2021-10-12 | 杭州萤石软件有限公司 | Out-of-range detection method and device and storage medium |
| CN113496500B (en) * | 2020-04-02 | 2024-09-20 | 杭州萤石软件有限公司 | Out-of-range detection method, device and storage medium |
| CN114866692A (en) * | 2022-04-19 | 2022-08-05 | 合肥富煌君达高科信息技术有限公司 | Image output method and system of large-resolution monitoring camera |
| CN115984335A (en) * | 2023-03-20 | 2023-04-18 | 华南农业大学 | Method for acquiring characteristic parameters of fog drops based on image processing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107657626B (en) | 2021-06-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107657626A (en) | The detection method and device of a kind of moving target | |
| Vasconez et al. | Comparison of convolutional neural networks in fruit detection and counting: A comprehensive evaluation | |
| St-Charles et al. | Flexible background subtraction with self-balanced local sensitivity | |
| CN107123131B (en) | Moving target detection method based on deep learning | |
| López-Rubio et al. | Foreground detection in video sequences with probabilistic self-organizing maps | |
| CA3154025A1 (en) | Interactive behavior recognizing method, device, computer equipment and storage medium | |
| CN104867161B (en) | A kind of method for processing video frequency and device | |
| Hou et al. | Detection of the mobile object with camouflage color under dynamic background based on optical flow | |
| CN104077605B (en) | A kind of pedestrian's search recognition methods based on color topological structure | |
| CN102307274B (en) | Motion detection method based on edge detection and frame difference | |
| CN102982313B (en) | The method of Smoke Detection | |
| CN107527009A (en) | A kind of remnant object detection method based on YOLO target detections | |
| CN106203260A (en) | Pedestrian's recognition and tracking method based on multiple-camera monitoring network | |
| CN106446862A (en) | Face detection method and system | |
| CN107633226A (en) | A kind of human action Tracking Recognition method and system | |
| CN106815576B (en) | Target tracking method based on continuous space-time confidence map and semi-supervised extreme learning machine | |
| CN110176024A (en) | Method, apparatus, equipment and the storage medium that target is detected in video | |
| CN106570490A (en) | Pedestrian real-time tracking method based on fast clustering | |
| CN115497056B (en) | Method for detecting lost articles in region based on deep learning | |
| CN117541994A (en) | Abnormal behavior detection model and detection method in dense multi-person scene | |
| CN109948628B (en) | Target detection method based on discriminant region mining | |
| CN108364303A (en) | A kind of video camera intelligent-tracking method with secret protection | |
| US20180336694A1 (en) | System and Method for Passive Tracking Based on Color Features | |
| CN108961287B (en) | Intelligent shelf triggering method, intelligent shelf system, storage medium and electronic equipment | |
| Yang et al. | Combining Gaussian mixture model and HSV model with deep convolution neural network for detecting smoke in videos |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |