CN103473791B - Abnormal speed event automatic identifying method in monitor video - Google Patents
Abnormal speed event automatic identifying method in monitor video Download PDFInfo
- Publication number
- CN103473791B CN103473791B CN201310410135.0A CN201310410135A CN103473791B CN 103473791 B CN103473791 B CN 103473791B CN 201310410135 A CN201310410135 A CN 201310410135A CN 103473791 B CN103473791 B CN 103473791B
- Authority
- CN
- China
- Prior art keywords
- target
- foreground target
- foreground
- moving target
- image
- 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.)
- Active
Links
Abstract
The invention discloses abnormal speed event automatic identifying method in a kind of monitor video, including step:(1) two field picture is read from monitor video, and carries out image procossing, so as to obtain the image sequence of needs;(2) background model is set up using statistical average method and mixed Gaussian background modeling, and foreground target set is detected with background subtraction;(3) cross-matched detection method and Mean shift algorithms are combined, the moving target corresponding to each foreground target in the foreground target set is tracked, to obtain the positional information of each moving target;(4) speed of each moving target is calculated, when the speed of each moving target exceedes predetermined threshold value, then alarm is sent;Otherwise return and perform step(1).Using the present invention, the corresponding foreground target of moving target can be relatively accurately detected, and maximally utilise foreground target and occur without larger deviation ensureing the position of moving target, relatively accurately identify abnormal speed event.
Description
Technical field
The present invention relates to computer intelligence monitor video process field, more particularly in monitor video abnormal speed event from
Dynamic recognition methods.
Background technology
In recent years, with the continuous progressive and development of society, increasing public place even private site all starts
Assembling monitoring device.And the consciousness of safety precaution is strengthened constantly along with people, market is to record and warning system
Demand is also growing day by day, and this causes that video monitoring has obtained application widely in production and living each side.But it is general
Video monitoring system needs are more to be accomplished manually, so often there is careless omission because of the fatigue of monitoring personnel, or even in fact
Most of video monitoring system does not have that personnel are on the scene to be monitored on border, and this video monitoring system is recorded video figure
Picture, can only be used as post-mordem forensics, and the real-time and initiative of monitoring are not given full play to.On the other hand, in present video
In monitoring field, the particularly monitoring of public arena typically all round-the-clock 24 hours is monitored, and because monitoring sets
The continuous reduction of standby cost, the quantity of monitoring is also lifted constantly, and common manual monitoring has been difficult to meet growing
Demand, it is therefore desirable to video monitoring can automatic identification anomalous event reach the real-time and initiative of monitoring.
Usual intelligent monitoring is the movable information by moving target obtained with monitoring and tracking moving target,
Wherein, moving target is the movement entity in video, and foreground target is prospect of the moving target in a certain two field picture.
Relative to manual monitoring, intelligent video monitoring has three advantages:One be not in due to staff fatigue and
The monitoring for causing is slipped up;Two can be quick search and statistics;Three can be the event that rapidly notes abnormalities, there is provided report in real time
It is alert, the monitoring personnel very first time is reacted and is processed." record thing occurs " of tradition monitoring is changed into " active defense ", can
Relatively reliable guarantee is provided with for social stability.Using the powerful calculating ability of computer, we can not only provide in advance
Trend prediction, Realtime Alerts, can also realize the function of quick search afterwards.
With continuing to develop for monitoring trade, intelligent monitoring application is inherently penetrated into all trades and professions, greatly to bank's weight
Want the monitoring in place, small to family's anti-thefting monitoring, intelligent monitoring can play its distinctive effect.Certainly, it is intelligent
Monitoring will ensure that society's peace provides more strong support to create safe social environment.
But, also there is certain deficiency in current intelligent video monitoring, it is impossible to effectively and accurately identify abnormal speed
The generation of thing.
The content of the invention
Embodiment of the present invention technical problem to be solved is, there is provided abnormal speed event is automatic in a kind of monitor video
Recognition methods.It is an object of the invention to carry out detect and track to the moving target inside monitor video, moving target is extracted
Information, automatic identification abnormal speed event and can send warning, give full play to the real-time and initiative of monitoring.
In order to solve the above-mentioned technical problem, the embodiment of the present invention uses mixed Gaussian background modeling and background subtraction to preceding
Scape goal set detected, and combined with Mean-shift algorithms with cross-matched detection method and realize to the prospect
The tracking of the corresponding moving target of each foreground target in goal set, finally recognizes abnormal speed event and sends alarm, has
Body includes step:
(1) two field picture is read from monitor video, image procossing is carried out to the two field picture, so as to obtain the image of needs
Sequence;
(2) using the described image sequence for obtaining, background mould is set up using statistical average method and mixed Gaussian background modeling
Type, and foreground target set is detected with background subtraction based on the background model and the two field picture, and store described
Foreground target set;
(3) cross-matched detection method and Mean-shift algorithms are combined, to the foreground target set in it is each before
Moving target corresponding to scape target is tracked, to obtain the positional information of each moving target;
(4) positional information of the described each moving target by obtaining, calculates the speed of each moving target, when
When the speed of each moving target exceedes predetermined threshold value, then alarm is sent;Otherwise return and perform step (1).
Further, step (1) specifically includes step:
(1-1) reads two field picture from monitor video, and the two field picture is RGB image;
The RGB image is converted into HSV images by (1-2), and extracts the H information in the HSV images, to set up H figures
Picture;
(1-3) removes the H picture noises using gaussian filtering, obtains described image sequence.
Wherein, step (1-1) is specially:The RGB image is converted into HSV images, and uses mask image mask pairs
The HSV images are processed, the pixel too low to remove V brightness values.
Further, step (2) specifically includes following steps:
(2-1) is directed to described image sequence, and background model is set up using statistical average method and mixed Gaussian background modeling;
(2-2) is split foreground target pixel set from background subtraction for the background model and the two field picture
Out;
(2-3) carries out the polygonal segments of profile to the foreground target pixel set using designated precision, before extracting
Scape objective contour;
(2-4) travels through the foreground target contour images pixel, and diffusion to find and the foreground target profile diagram
As the pixel set that pixel is connected, the pixel set is the foreground target set in current frame image.
Further, the step (2) also includes step after step (2-4):
(2-5) detects whether the pixel of the foreground target set marginal portion changes, if so, then return will be described
The two field picture that step (1) reads carries out image procossing again after carrying out Gaussian Blur treatment first, and continues executing with step (2), directly
Not changed to the pixel for detecting the foreground target set marginal portion.
Wherein, step (2-2) is specifically included:Will be preceding from background subtraction based on the background model and the two field picture
Scape object pixel set-partition out, and operates and closes behaviour to the foreground target pixel set using opening in morphological operation
Make to eliminate noise.
Further, step (3) specifically includes following steps:
Prospect mesh in foreground target set and current frame image in the previous frame image that (3-1) obtains storage from step (2)
Mark set;
Foreground target set in foreground target set and current frame image, searches mutual in (3-2) traversal previous frame image
The quantity of intersection, i.e., the situation of each foreground target windows overlay;
(3-3) is directed to the situation of the windows overlay of each foreground target, using cross-matched detection method and Mean-
Shift algorithms, are tracked to the corresponding moving target of each foreground target.
Wherein, step (3-3) specifically includes step:
(3-3-1) be not when having foreground target with current foreground target phase in the foreground target set in the previous frame image
During intersection, then it is assumed that new moving target occur, and determine that the current foreground target is that the new moving target is being worked as
Position in previous frame;
(3-3-2) when have in the foreground target set in the previous frame image and only one of which foreground target with it is current before
When scape target intersects, then the histogrammic matching value of the two foreground targets is measured, if above threshold value, it is determined that described to work as
Preceding foreground target is its corresponding moving target position in the current frame;Wherein, the histogram represents that foreground target institute is right
The color information of the moving target answered, and measure two histogrammic matching values of foreground target, bar using Pasteur's distance
Family name is apart from smaller, and matching value is bigger.
(3-3-3) in the foreground target set in the previous frame image when having multiple foreground targets with current foreground target
When intersecting, then each foreground target intersected with it is examined closely, by Mean-shift algorithms, search the prospect mesh of its matching
Mark, to determine the corresponding moving target of current foreground target position in the current frame.
Further, step (3-3) also includes step:
(3-3-4) is when the foreground target not matched in the presence of residue in the foreground target set in the previous frame image
When, for the foreground target that the residue is not matched, found using Mean-shift algorithms, the region that extraction is searched out
Histogram, and measure the histogrammic matching value of foreground target that the histogram and the residue in the region are not matched, it is low
Think that the corresponding moving target of the foreground target disappears when threshold value, otherwise it is assumed that static.
Further, step (4) specifically includes step:
The barycenter that (4-1) extracts the moving target carries out velocity measuring, and using frame-to-frame differences vnAs the motion mesh
It is marked on the speed of n frames, frame-to-frame differences vnComputing formula be:vn=| sn+1-sn|;
Wherein, snIt is the moving target in the position of t frames, sn+1It is the moving target in the position of t+1 frames;
(4-2) calculates the difference of speed of the moving target between two seconds, when more than predetermined threshold value, is judged as abnormal fast
Degree event occurs, and sends alarm;Wherein, the difference of speed of the moving target between two seconds is calculated by below equation:
|vt+1-vt| > h
Wherein, vtIt is the moving target in the speed of t seconds, vt+1It is the moving target in the speed of t+1 seconds;H is pre-
If threshold value, and determined by the ratio of width to height of video.
Implement the embodiment of the present invention, have the advantages that:
1st, the embodiment of the present invention can carry out monitor in real time, be not in that the monitoring caused due to staff's fatigue is lost
By mistake;
2nd, the embodiment of the present invention can rapidly note abnormalities event, there is provided Realtime Alerts, make the monitoring personnel very first time
React and process;
3rd, the present invention can relatively accurately detect the corresponding foreground target of moving target, and maximally utilise prospect
Target occurs without larger deviation ensureing the position of moving target, relatively accurately identifies abnormal speed event.
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 embodiment or existing
The accompanying drawing to be used needed for having technology description is briefly described, it should be apparent that, drawings in the following description are only this
Some embodiments of invention, for those of ordinary skill in the art, on the premise of not paying creative work, can be with
Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 is the flow chart of abnormal speed event automatic identifying method in monitor video provided in an embodiment of the present invention;
Fig. 2 is the particular flow sheet of step S1 in Fig. 1;
Fig. 3 is the particular flow sheet of step S2 in Fig. 1;
Fig. 4 is the particular flow sheet of step S3 in Fig. 1;
Fig. 5 is the particular flow sheet of step S4 in Fig. 1.
Specific 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
Site preparation is described, it is clear that described embodiment is only a part of embodiment of the invention, rather than whole embodiments.It is based on
Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under the premise of creative work is not made
Embodiment, belongs to the scope of protection of the invention.
As shown in Figure 1:Abnormal speed event automatic identifying method includes step in monitor video provided in an embodiment of the present invention
Suddenly:
S1:Two field picture is read from monitor video, image procossing is carried out to the two field picture, so as to obtain the image of needs
Sequence;
S2:Using the described image sequence for obtaining, background mould is set up using statistical average method and mixed Gaussian background modeling
Type, and foreground target set is detected with background subtraction based on the background model and the two field picture, and store described
Foreground target set;
S3:With reference to cross-matched detection method and Mean-shift algorithms, to the foreground target set in it is each before
Moving target corresponding to scape target is tracked, to obtain the positional information of each moving target;
S4:The positional information of the described each moving target by obtaining, calculates the speed of each moving target, when
When the speed of each moving target exceedes predetermined threshold value, then alarm is sent;Otherwise return and perform step S1.
Further, as shown in Fig. 2 step S1 specifically includes step:
S11:Two field picture is read from monitor video, the two field picture is RGB image;
S12:The RGB image is converted into HSV images, and extracts the H information in the HSV images, to set up H figures
Picture;
S13:The H picture noises are removed using gaussian filtering, described image sequence is obtained.
Preferably, RGB image is first switched into HSV images using the cvCvtColor functions in openCV storehouses, is then made again
Go out H information to set up H images with cvSplit method separation and Extractions.
Wherein, step S11 is specially:The RGB image is converted into HSV images, and using mask image mask to institute
State HSV images to be processed, the pixel too low to remove V brightness values.Using mask image mask to the HSV image procossings
The interference that shade is detected to foreground target set can be reduced.
Further, as shown in figure 3, step S2 specifically includes following steps:
S21:For described image sequence, background model is set up using statistical average method and mixed Gaussian background modeling;
Statistical average method and mixed Gaussian background modeling set up background model is united by continuous image sequence
Meter, and obtain average value and set up background model, while needing to set up a polynary model, the independent model of original pixel is expanded
Exhibition is the composite model comprising adjacent pixel;Because background model, background mould will be re-established for each two field picture
Type can be constantly updated;
S22:Foreground target pixel set is split from background subtraction for the background model and the two field picture
Out;
Wherein, background subtraction, exactly carries out difference using the background model having built up and current frame image, is worth phase
With pixel be considered as background, value it is different be considered as prospect, you can distinguish foreground target pixel set;Specifically, in the step
In, foreground target pixel set is split from background subtraction based on the background model and the two field picture, and it is right
The foreground target pixel set eliminates noise using open operation and the closed operation in morphological operation.Can be with using operation is opened
Make less pixel cluster disappear, but so can also make the edge of the target of needs disappear, therefore reuse closed operation
Rebuild due to opening the part that operation is lost;
S23:The polygonal segments of profile are carried out using designated precision to the foreground target pixel set, prospect is extracted
Objective contour;
Preferably, polygonal segments are using the cvApproxPoly methods in openCV storehouses;
S24:The foreground target contour images pixel, and diffusion is traveled through to find and the foreground target contour images
The pixel set that pixel is connected, the pixel set is the foreground target set in current frame image;
S25:Detect whether the pixel of the foreground target set marginal portion changes, if so, then return will be described
The two field picture that step S1 reads carries out image procossing again after carrying out Gaussian Blur treatment first, and continues executing with step S2, until
Detect the foreground target set marginal portion pixel do not change untill.
Further, as shown in figure 4, step S3 specifically includes following steps:
S31:Foreground target in foreground target set and current frame image in the previous frame image for obtaining storage from step S2
Set;
S32:Foreground target set in foreground target set and current frame image, searches and mutually hands in traversal previous frame image
The quantity of fork, i.e., the situation of each foreground target windows overlay;
S33:For the situation of the windows overlay of each foreground target, using cross-matched detection method and Mean-shift
Algorithm, is tracked to the corresponding moving target of each foreground target.
Wherein, in step S33, for each foreground target, the situation of windows overlay potentially includes following three kinds of situations:
1st, when there is no foreground target to be intersected with current foreground target in the foreground target set in the previous frame image
When, then it is assumed that there is new moving target, and determine that the current foreground target is the new moving target in present frame
In position;
2nd, when having in the foreground target set in the previous frame image and only one of which foreground target and current prospect mesh
Mark then measures the histogrammic matching value of the two foreground targets, if above threshold value when intersecting, it is determined that it is described it is current before
Scape target is its corresponding moving target position in the current frame;Wherein, the histogram is represented corresponding to foreground target
The color information of moving target, and two histogrammic matching values of foreground target are measured using Pasteur's distance, Pasteur away from
From smaller, matching value is bigger;
3rd, when thering are multiple foreground targets to intersect with current foreground target in the foreground target set in the previous frame image
During fork, then each foreground target intersected with it is examined closely, by Mean-shift algorithms, search the foreground target of its matching,
To determine the corresponding moving target of current foreground target position in the current frame;
In addition, when there is the foreground target that residue is not matched in the foreground target set in the previous frame image,
For the foreground target that the residue is not matched, found using Mean-shift algorithms, the region that extraction is searched out
Histogram, and the histogrammic matching value of foreground target that the histogram and the residue in the region are not matched is measured, it is less than
Think that the corresponding moving target of the foreground target disappears during threshold value, otherwise it is assumed that static.
Further, as shown in figure 5, step S4 specifically includes step:
S41:Extracting the barycenter of the moving target carries out velocity measuring, and using frame-to-frame differences vnAs the moving target
In the speed of n frames, frame-to-frame differences vnComputing formula be:vn=| sn+1-sn|;
Wherein, snIt is the moving target in the position of t frames, sn+1It is the moving target in the position of t+1 frames;
S42:The difference of speed of the moving target between two seconds is calculated, when more than predetermined threshold value, is judged as abnormal fast
Degree event occurs, and sends alarm;Wherein, the difference of speed of the moving target between two seconds is calculated by below equation:
|vt+1-vt| > h
Wherein, vtIt is target in the speed of t seconds, vt+1It is target in the speed of t+1 seconds;H is predetermined threshold value, and by video
The ratio of width to height determine.
Implement the embodiment of the present invention, have the advantages that:
1st, the embodiment of the present invention can carry out monitor in real time, be not in that the monitoring caused due to staff's fatigue is lost
By mistake;
2nd, the embodiment of the present invention can rapidly note abnormalities event, there is provided Realtime Alerts, make the monitoring personnel very first time
React and process;
3rd, the present invention can relatively accurately detect the corresponding foreground target of moving target, and maximally utilise prospect
Target occurs without larger deviation ensureing the position of moving target, relatively accurately identifies abnormal speed event.
Above disclosed is only a kind of preferred embodiment of the invention, can not limit the power of the present invention with this certainly
Sharp scope, therefore the equivalent variations made according to the claims in the present invention, still belong to the scope that the present invention is covered.
Claims (5)
1. abnormal speed event automatic identifying method in a kind of monitor video, it is characterised in that including step:
(1) two field picture is read from monitor video, image procossing is carried out to the two field picture, so as to obtain the image sequence of needs
Row;
(2) using the described image sequence for obtaining, background model is set up using statistical average method and mixed Gaussian background modeling, and
Foreground target set is detected with background subtraction based on the background model and the two field picture, and stores the prospect mesh
Mark set;
(3) cross-matched detection method and Mean-shift algorithms are combined, to the foreground target set in each prospect mesh
The corresponding moving target of mark is tracked, to obtain the positional information of each moving target;
Step (3) specifically includes following steps:
Foreground target collection in foreground target set and current frame image in the previous frame image that (3-1) obtains storage from step (2)
Close;
Foreground target set in foreground target set and current frame image in (3-2) traversal previous frame image, lookup intersects
Quantity, i.e., the situation of each foreground target windows overlay;
(3-3) is calculated for the situation of the windows overlay of each foreground target using cross-matched detection method and Mean-shift
Method, is tracked to the corresponding moving target of each foreground target;
Step (3-3) specifically includes step:
(3-3-1) in the foreground target set in the previous frame image when not having foreground target to be intersected with current foreground target
When, then it is assumed that there is new moving target, and determine that the current foreground target is the new moving target in present frame
In position;
(3-3-2) in the foreground target set in the previous frame image when having and only one of which foreground target and current prospect mesh
Mark then measures the histogrammic matching value of the two foreground targets, if above threshold value when intersecting, it is determined that it is described it is current before
Scape target is its corresponding moving target position in the current frame;Wherein, the histogram is represented corresponding to foreground target
The color information of moving target, and two histogrammic matching values of foreground target are measured using Pasteur's distance, Pasteur away from
From smaller, matching value is bigger;
(3-3-3) in the foreground target set in the previous frame image when having multiple foreground targets to intersect with current foreground target
During fork, then each foreground target intersected with it is examined closely, by Mean-shift algorithms, search the foreground target of its matching,
To determine the corresponding moving target of current foreground target position in the current frame;
(4) positional information of the described each moving target by obtaining, calculates the speed of each moving target, when described
When the speed of each moving target exceedes predetermined threshold value, then alarm is sent;Otherwise return and perform step (1);
Wherein, step (1) specifically includes step:
(1-1) reads two field picture from monitor video, and the two field picture is RGB image;
The RGB image is converted into HSV images by (1-2), and extracts the H information in the HSV images, to set up H images;
(1-3) removes the H picture noises using gaussian filtering, obtains described image sequence;
Wherein, step (1-1) is specially:The RGB image is converted into HSV images, and using mask image mask to described
HSV images are processed, the pixel too low to remove V brightness values;
Step (2) specifically includes following steps:
(2-1) is directed to described image sequence, and background model is set up using statistical average method and mixed Gaussian background modeling;
Be partitioned into for foreground target pixel set from background subtraction for the background model and the two field picture by (2-2)
Come;
(2-3) carries out the polygonal segments of profile to the foreground target pixel set using designated precision, extracts prospect mesh
Mark profile;
(2-4) travels through the foreground target contour images pixel, and diffusion to find and the foreground target contour images picture
The pixel set that vegetarian refreshments is connected, the pixel set is the foreground target set in current frame image.
2. abnormal speed event automatic identifying method in monitor video as claimed in claim 1, it is characterised in that the step
(2) step is also included after step (2-4):
(2-5) detects whether the pixel of the foreground target set marginal portion changes, if so, then returning the step
(1) two field picture for reading carries out image procossing again after carrying out Gaussian Blur treatment first, and continues executing with step (2), until inspection
Measure the foreground target set marginal portion pixel do not change untill.
3. abnormal speed event automatic identifying method in monitor video as claimed in claim 1, it is characterised in that step (2-
2) specifically include:Foreground target pixel set is partitioned into from background subtraction based on the background model and the two field picture
Come, and to the foreground target pixel set using in morphological operation opening operation and closed operation eliminate noise.
4. abnormal speed event automatic identifying method in monitor video as claimed in claim 1, it is characterised in that step (3-3)
Also include step:
(3-3-4) when exist in the foreground target set in the previous frame image it is remaining do not match foreground target when, pin
The foreground target not matched to the residue, is found using Mean-shift algorithms, extraction search out region it is straight
Fang Tu, and the histogrammic matching value of foreground target that the histogram and the residue in the region are not matched is measured, less than threshold
Think that the corresponding moving target of the foreground target disappears during value, otherwise it is assumed that static.
5. abnormal speed event automatic identifying method in monitor video as claimed in claim 1, it is characterised in that step (4) has
Body includes step:
The barycenter that (4-1) extracts the moving target carries out velocity measuring, and using frame-to-frame differences vnAs the moving target in n
The speed of frame, frame-to-frame differences vnComputing formula be:vn=︱ sn+1–sn︱;
Wherein, snIt is the moving target in the position of t frames, sn+1It is the moving target in the position of t+1 frames;
(4-2) calculates the difference of the speed between described moving target t seconds and t+1 seconds, when more than predetermined threshold value, is judged as exception
Speed event occurs, and sends alarm;Wherein, the difference of the speed between described moving target t seconds and t+1 seconds passes through below equation
Calculate:
︱ vt+1–vt︱>h
Wherein, vtIt is the moving target in the speed of t seconds, vt+1It is the moving target in the speed of t+1 seconds;H is default threshold
Value, and determined by the ratio of width to height of video.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310410135.0A CN103473791B (en) | 2013-09-10 | 2013-09-10 | Abnormal speed event automatic identifying method in monitor video |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310410135.0A CN103473791B (en) | 2013-09-10 | 2013-09-10 | Abnormal speed event automatic identifying method in monitor video |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103473791A CN103473791A (en) | 2013-12-25 |
CN103473791B true CN103473791B (en) | 2017-06-30 |
Family
ID=49798625
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310410135.0A Active CN103473791B (en) | 2013-09-10 | 2013-09-10 | Abnormal speed event automatic identifying method in monitor video |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103473791B (en) |
Families Citing this family (24)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104506850B (en) * | 2014-12-16 | 2017-09-29 | 北京中星微电子有限公司 | A kind of video shelter determination methods and device |
CN104932683A (en) * | 2015-05-28 | 2015-09-23 | 重庆大学 | Game motion sensing control method based on vision information |
WO2017161544A1 (en) * | 2016-03-25 | 2017-09-28 | 深圳大学 | Single-camera video sequence matching based vehicle speed measurement method and system |
CN105933645A (en) * | 2016-04-15 | 2016-09-07 | 张志华 | Unmanned warehouse monitoring system |
CN105812743A (en) * | 2016-04-15 | 2016-07-27 | 张志华 | Moving object shooting method |
CN105933644A (en) * | 2016-04-15 | 2016-09-07 | 张志华 | Monitoring device for searching moving object |
CN105701843A (en) * | 2016-04-15 | 2016-06-22 | 张志华 | Unattended parking lot monitoring system |
CN105931265A (en) * | 2016-04-15 | 2016-09-07 | 张志华 | Intelligent monitoring early-warning device |
CN106981104B (en) * | 2017-03-24 | 2020-04-21 | 北京悦畅科技有限公司 | Charging monitoring method, device, server and system |
CN107424170B (en) * | 2017-06-17 | 2020-07-28 | 复旦大学 | Motion characteristic description method for automatically detecting local abnormal behaviors in monitoring video |
CN107490377A (en) * | 2017-07-17 | 2017-12-19 | 五邑大学 | Indoor map-free navigation system and navigation method |
CN107729866A (en) * | 2017-10-31 | 2018-02-23 | 武汉理工大学 | Ship based on timing diagram picture touches mark automatic detection device and method |
CN107911655A (en) * | 2017-11-17 | 2018-04-13 | 江苏艾思博信息技术有限公司 | Intelligence handling based on AR perform method and system |
WO2019155533A1 (en) * | 2018-02-06 | 2019-08-15 | 三菱電機株式会社 | Sequential data analysis apparatus, sequential data analysis method, and sequential data analysis program |
CN108830204B (en) * | 2018-06-01 | 2021-10-19 | 中国科学技术大学 | Method for detecting abnormality in target-oriented surveillance video |
CN108848366B (en) * | 2018-07-05 | 2020-12-18 | 盎锐(上海)信息科技有限公司 | Information acquisition device and method based on 3D camera |
CN109389152B (en) * | 2018-08-30 | 2022-05-13 | 广东工业大学 | Refined identification method for power transmission line falling object |
CN111325048B (en) * | 2018-12-13 | 2023-05-26 | 杭州海康威视数字技术股份有限公司 | Personnel gathering detection method and device |
CN110490902B (en) * | 2019-08-02 | 2022-06-14 | 西安天和防务技术股份有限公司 | Target tracking method and device applied to smart city and computer equipment |
CN112749596A (en) * | 2019-10-31 | 2021-05-04 | 顺丰科技有限公司 | Abnormal picture detection method and device, electronic equipment and storage medium |
CN111401135B (en) * | 2020-02-20 | 2023-09-26 | 长沙智能驾驶研究院有限公司 | Door opening anti-collision method and device, computer equipment and storage medium |
CN111601011A (en) * | 2020-04-10 | 2020-08-28 | 全景智联(武汉)科技有限公司 | Automatic alarm method and system based on video stream image |
CN113807127A (en) * | 2020-06-12 | 2021-12-17 | 杭州海康威视数字技术股份有限公司 | Personnel archiving method and device and electronic equipment |
CN112906456B (en) * | 2020-12-29 | 2024-02-27 | 周口师范学院 | Crowd abnormal behavior detection method and system based on inter-frame characteristics |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103248878A (en) * | 2013-05-23 | 2013-08-14 | 南车株洲电力机车有限公司 | Pattern recognition method, device and system of abnormal situation of fully mechanized coal mining face |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI492188B (en) * | 2008-12-25 | 2015-07-11 | Univ Nat Chiao Tung | Method for automatic detection and tracking of multiple targets with multiple cameras and system therefor |
-
2013
- 2013-09-10 CN CN201310410135.0A patent/CN103473791B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103248878A (en) * | 2013-05-23 | 2013-08-14 | 南车株洲电力机车有限公司 | Pattern recognition method, device and system of abnormal situation of fully mechanized coal mining face |
Non-Patent Citations (3)
Title |
---|
基于Directshow的智能视频监控系统研究;吕行;《中国优秀博硕士学位论文全文数据库 (硕士) 信息科技辑》;20070615(第06期);第25、42、43、44页,图3-3 * |
基于场景变化的运动目标实时检测与跟踪技术研究;白雪;《中国优秀硕士学位论文全文数据库 信息科技辑》;20111215(第S2期);第37、41页 * |
视频监控系统中运动目标检测与跟踪的研究;金克琼;《中国优秀硕士学位论文全文数据库信息科技辑》》;20101215(第12期);第16、24、28、30、31页,图3-3 * |
Also Published As
Publication number | Publication date |
---|---|
CN103473791A (en) | 2013-12-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103473791B (en) | Abnormal speed event automatic identifying method in monitor video | |
CN102663743B (en) | Personage's method for tracing that in a kind of complex scene, many Kameras are collaborative | |
US7787656B2 (en) | Method for counting people passing through a gate | |
EP2891990A1 (en) | Method and device for monitoring video digest | |
EP2801078B1 (en) | Context aware moving object detection | |
CN103279737B (en) | A kind of behavioral value method of fighting based on space-time interest points | |
CN104866827B (en) | Personnel based on video monitoring platform cross the detection method of behavior | |
JP5218906B2 (en) | Smoke detection device and smoke detection method | |
US10691947B2 (en) | Monitoring device | |
CN102938058A (en) | Method and system for video driving intelligent perception and facing safe city | |
CN105761261A (en) | Method for detecting artificial malicious damage to camera | |
CN101179713A (en) | Method of detecting single moving target under complex background | |
Salehi et al. | An automatic video-based drowning detection system for swimming pools using active contours | |
CN104866842B (en) | A kind of human body target intrusion detection method | |
CN102542246A (en) | Abnormal face detection method for ATM (Automatic Teller Machine) | |
CN107909599A (en) | A kind of object detecting and tracking system | |
CN108830204B (en) | Method for detecting abnormality in target-oriented surveillance video | |
WO2019172172A1 (en) | Object tracker, object tracking method, and computer program | |
CN102214359A (en) | Target tracking device and method based on hierarchic type feature matching | |
CN107067595A (en) | State identification method, device and the electronic equipment of a kind of indicator lamp | |
CN113297926A (en) | Behavior detection and identification method and system | |
TWI493510B (en) | Falling down detection method | |
CN103049748B (en) | Behavior monitoring method and device | |
US20200306585A1 (en) | Jump counting method for jump rope | |
Liang et al. | Methods of moving target detection and behavior recognition in intelligent vision monitoring. |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20181121 Address after: 516003 Sanxin 22 District, Jiangbei, Huizhou City, Guangdong Province (4th Floor, 2nd Building of Factory Building) Patentee after: Guangdong Anxin Technology Co., Ltd. Address before: 516007 Da Dao Da Avenue, Huicheng District, Huizhou City, Guangdong Province, No. 46 Patentee before: Huizhou College |