CN107273880A - A kind of multi-storied garage safety-protection system and method based on intelligent video monitoring - Google Patents
A kind of multi-storied garage safety-protection system and method based on intelligent video monitoring Download PDFInfo
- Publication number
- CN107273880A CN107273880A CN201710638562.2A CN201710638562A CN107273880A CN 107273880 A CN107273880 A CN 107273880A CN 201710638562 A CN201710638562 A CN 201710638562A CN 107273880 A CN107273880 A CN 107273880A
- Authority
- CN
- China
- Prior art keywords
- storied garage
- main frame
- web camera
- video image
- image information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000012544 monitoring process Methods 0.000 title claims abstract description 33
- 238000004458 analytical method Methods 0.000 claims abstract description 7
- 238000001514 detection method Methods 0.000 claims description 23
- 238000004422 calculation algorithm Methods 0.000 claims description 18
- 238000012706 support-vector machine Methods 0.000 claims description 12
- 230000002159 abnormal effect Effects 0.000 claims description 10
- 238000004891 communication Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 230000005856 abnormality Effects 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000001960 triggered effect Effects 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000009471 action Effects 0.000 abstract description 2
- 239000000463 material Substances 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 14
- 238000012549 training Methods 0.000 description 10
- 230000000694 effects Effects 0.000 description 8
- 230000008859 change Effects 0.000 description 7
- 230000003287 optical effect Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 238000005457 optimization Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 238000013507 mapping Methods 0.000 description 4
- 238000003909 pattern recognition Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000007812 deficiency Effects 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- GOLXNESZZPUPJE-UHFFFAOYSA-N spiromesifen Chemical compound CC1=CC(C)=CC(C)=C1C(C(O1)=O)=C(OC(=O)CC(C)(C)C)C11CCCC1 GOLXNESZZPUPJE-UHFFFAOYSA-N 0.000 description 3
- 230000004438 eyesight Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 206010000117 Abnormal behaviour Diseases 0.000 description 1
- 241001465754 Metazoa Species 0.000 description 1
- 241000256626 Pterygota <winged insects> Species 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000036760 body temperature Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000011410 subtraction method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000007306 turnover Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- H04N7/181—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a plurality of remote sources
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of multi-storied garage safety-protection system and method based on intelligent video monitoring, multi-storied garage applied to lift-sliding, the multi-storied garage safety-protection system includes button, at least one web camera, router, main frame and controller, in the case where not needing human intervention, the image sequence of video camera recording is automatically analyzed using machine vision and the method for video analysis, realize the positioning and identification to target in dynamic scene, and the species of target is analyzed and judged on this basis, draw the understanding to picture material implication and the explanation to objective scene, so as to instruct and planning action.Emphasis of the present invention is studied the key technology of computer intelligence video monitoring system and is applied to the security protection in multi-storied garage Transport Vehicle region.
Description
Technical field
The invention belongs to technical field of video monitoring, more particularly it relates to which a kind of be based on intelligent video monitoring
Multi-storied garage safety-protection system and method.
Background technology
With the continuous improvement developed rapidly with living standards of the people of China's economy, increasing people have purchased private savings
Car, life but frequently encounters the problem that vehicle is nowhere parked while convenience.The data display that National Development and Reform Committee in 2017 announces,
Conservative estimation China parking stall breach is more than 50,000,000, and " parking pain " turns into city common fault.
Multi-storied garage, is a kind of mechanical device for being accessed for maximum and storing vehicle, is allowed using mechanically and electrically system
Stop into vehicle automatic putting to vacant locations, it is with the small unique property of its average bicycle floor space, as effective solution
The important channel of domestic city parking problem.By the end of the year 2015, the city that the whole nation possesses mechanical carport reaches 491,
Become basically universal to all one, two, three line cities, built mechanical carport project more than 14.4 ten thousand, berth total amount more than 3,360,000
It is individual.Over 30 years, mechanical stereo garage industry experienced evolution from scratch, from small to large, from weak to strong, China's mesh
Before have become mechanical stereo garage in the world and have the call and production scale also maximum country.
Mechanical stereo garage is broadly divided into lifting cross sliding type, vertical circulation class, horizontal cyclic class, multilayer circulation class, flat
The types such as face movement class, tunnel stacking class, vertical lifting, simple lifting class, wherein, lifting cross sliding type multi-storied garage uses mould
Blockization is designed, and parking stall can be used from several to hundreds of, using a variety of site conditions and can use multiple combinations mode, have
Effect utilizes place existing space, can also be widely used in the transformation of underground parking and Surface parking lots, be mainly used in house
The Public Parking of cell, institutional settings, commercial office building, the above and below ground parking lot in hotel.At present, lifting cross sliding type is three-dimensional
Garage is most widely used, in multi-storied garage because its type is more, scalable, place strong adaptability, the low outstanding feature of cost
The occupation rate in market is more than 70%.
However, lifting cross sliding type multi-storied garage still suffers from larger potential safety hazard at present, cause many people dare not be by vehicle
Relievedly dock in garage, the vacancy rate of many parking systems is higher.Some easy lifting cross sliding type parking systems
Safety is seen whether mainly by garage keeper or driver's eyesight to determine to start mobile vehicle, but if observer is not careful enough
Or still someone or pet stops in observation blind area, it is possible to which life security is threatened.For this problem, take at present
Major measure be that human body heat-releasing infrared inductor and network monitoring camera head are installed, these measures are respectively provided with certain effect,
But respectively there are different defects.
Human body heat-releasing infrared inductor is that the infrared signal for sending human body is gathered on inductor and produces ac signal,
Cause this AC signal to produce fluctuation to produce response in control circuit by human motion again.Therefore, the detection side
Method is applied to the personnel in multi-storied garage Transport Vehicle region and the security protection of pet is primarily present following deficiency:
(1) personnel are moved through hour and can't detect;
(2) there is certain monitoring dead angle;
(3) when being vulnerable to the too high interference of environment, especially indoor temperature or larger indoor/outdoor temperature-difference, mistake is all easily caused
When report, environment temperature and close human body temperature, detection and sensitivity are decreased obviously, and even cause failure in short-term sometimes;
(4) it is only capable of judging whether there are personnel in detection zone, but the location of personnel can not be judged;
(5) toy on ground can not be detected, the setting height(from bottom) of detector is to the toy one on ground in investigative range
As do not produce alarm.
As can be seen that by human body heat-releasing infrared inductor be can not reliably judge garage door close after whether someone or its
What his toy was present.
In addition, small part lifting cross sliding type multi-storied garage Transport Vehicle region there is also mounted and focus infrared monitoring camera, 24
Hour non-stop run, vision signal is collected CCC and is simultaneously shown on monitor, then by security personnel couple
Video image carries out analysis and draws judgement.However, existing video monitoring system merely provide the capture of video, storage, point
The simple functions such as hair, it is impossible to the effect for monitoring of taking the initiative, the judgement to video content can only be completed by people, therefore, when
During unexpected generation, it is impossible to signal is fed back into controller in time, it is to avoid the state of affairs aggravates.
Therefore, it is necessary to find significantly more efficient solution, multi-storied garage is monitored, enhanced prison is made it have
Control ability, reduction hidden danger, while the resource that uses manpower and material resources sparingly, investment reduction.
The content of the invention
The deficiency existed for the security protection measure of current lifting cross sliding type multi-storied garage Transport Vehicle region, the present invention is proposed
A kind of multi-storied garage safety-protection system based on intelligent video monitoring, multi-storied garage is monitored using image capture device, can be few
On the premise of amount increase cost, the security of multi-storied garage is obviously improved, the safety deposit of vehicle is realized and removes.
To achieve these goals, technical solution of the present invention is as follows:
A kind of multi-storied garage safety-protection system based on intelligent video monitoring, it is described applied to the multi-storied garage of lift-sliding
Multi-storied garage safety-protection system includes button, at least one web camera, router, main frame and controller, wherein:
The button connects send to main frame to main frame and move car instruction;
The web camera is arranged on the back upper place of multi-storied garage, and monitored picture covers whole multi-storied garage, for adopting
Collect the video image information in multi-storied garage;
The router is located between web camera and main frame, respectively with the web camera and the two-way company of main frame
Connect, be used in uplink communication to main frame transmitting video image information, be used to trigger web camera operation in downlink communication;
The main frame is unidirectionally connected to the controller, and main frame judges whether to move according to the video image information of collection
Car, and send a signal to the controller;
The controller performs to move car instruction or perform according to the signal received to be forbidden moving car instruction.
Preferably, the multi-storied garage safety-protection system also includes display, and the display is connected to the main frame, is used for
Show the video image in multi-storied garage.
Preferably, the multi-storied garage safety-protection system also includes warning system, and the warning system is connected to the main frame,
For sending alarm when noting abnormalities.
According to a preferred embodiment of the invention, the web camera is infrared high-definition network camera, respectively
Installed in multi-storied garage back upper place both sides.
The invention also proposes a kind of multi-storied garage safety protection method based on intelligent video monitoring, comprise the following steps:
S1. button is pressed, shifting car instruction is sent to main frame;
S2. main frame is received after shifting car instruction, is sent signal to web camera by router, is made web camera quilt
Triggering brings into operation;
S3. web camera transmits the video image information of collection to main frame by router;
S4. main frame is handled and analyzed to video image information, is judged in multi-storied garage with the presence or absence of abnormal;
If S5. normal, main frame sends shifting car to controller and instructed, and normally moves car, if there is exception, main frame is sent out to controller
Go out to forbid to move car instruction, forbid moving car.
Preferably, methods described also includes, if there is exception, main frame sends police instruction to warning system, is by alarm
System sends alarm sounds.
Further, the video to collection is included to the processing and analysis of video image information in the step S4 of methods described
Image information carries out moving object detection, feature extraction and moving object classification.
Moving object detection refers to come out moving target recognition from video monitoring scene or image sequence, is follow-up reality
The key of existing moving object classification.According to the principle of detection target, conventional object detection method is divided into three classes:Inter-frame difference
Method, optical flow method and background subtraction method.
Frame differential method is also known as time differencing method, and it makees the two continuous frames in sequence of video images or multiframe pixel point value
Difference, obtains the profile of moving target in monitoring scene, realizes the detection to target.The major advantage of this method is principle letter
Single, computation complexity is low, real-time;Sensitivity to moving object detection is high;To slow illumination variation strong adaptability.It is main
The deficiency to be deposited is, when target speed is slower, cavity to be produced inside the prospect of extraction;Two be the target wheel detected
, there is pseudo- target point in wide and realistic objective slightly deviation.
Optical flow method refers to the instantaneous velocity that its surface pixels point as caused by object of which movement is produced, and optical flow method passes through meter
The change for calculating corresponding pixel points in each two field picture characterizes moving target.Optical flow method detects that the principle of moving target is:Work as monitoring
In scene during without motion object, flow vector consecutive variations in monitoring range, and when there is moving target in monitoring scene, light stream
Field can change, and moving target light stream has significant change, by detecting that the light stream vector of significant changes can realize motion mesh
Target is detected.Under actual conditions, the gray scale conservation of optical flow constraint equation is assumed to disclosure satisfy that, causes optical flow field and sports ground
There is deviation.Optical flow method need to calculate the change of each pixel of monitoring scene, require higher to hardware support, more difficult in real-time
Meet.
Background subtraction by set up background model and by each frame of sequence of video images and background image make difference come
Extract the foreground target of motion.Background subtraction principle is simple, and amount of calculation is small, it is adaptable to the occasion that monitoring scene is fixed.It is existing
Background subtraction can be broadly divided into it is based on model, feature based and based on pixel value difference three major types.
Further, main frame is using video image information of the background modeling method based on Gaussianmixture model to collection
Carry out moving object detection.
Further, support vector machines grader is designed in main frame to classify to moving target.
Further, support vector machines grader is optimized using improved universal gravitation algorithm IGSA, if
IGSA-SVM graders are counted, the accuracy rate classified to moving target is effectively increased.
The multi-storied garage safety-protection system based on intelligent video monitoring and method of the present invention has advantages below:
1. IP Camera can automatically be triggered by signal, it is not necessary to run within 24 hours, extended equipment life, reduce
Human cost;
2. intelligent video monitoring system is integrated with powerful image-capable, and the intelligent algorithm for having upper strata is supported,
The abnormal behavior in User Defined monitoring scene can be supported, the operating efficiency of staff is drastically increased, has
The reduction of effect ground is failed to report and rate of false alarm;
3. being filtered out hash using classifier design, automatic identification abnormal object can effectively improve system
Response speed;
4. using the background modeling method detection moving target of Gaussianmixture model, improve the robustness of system;
5. devising the pattern-recognition that the grader based on IGSA-SVM carries out abnormal image, the accurate of classification is improved
Rate;
6. abnormal conditions are fed back into controller in the form of switching value, forbidden moves vehicle while alarm has
Effect ensure that life security.
Brief description of the drawings
Fig. 1 is the structured flowchart of the multi-storied garage safety-protection system based on intelligent video monitoring according to the present invention;
Fig. 2 is the workflow diagram of the multi-storied garage safety protection method based on intelligent video monitoring according to the present invention;
Fig. 3 carries out the schematic diagram of moving object detection and classification for the main frame of the present invention to video image information;
Fig. 4 is support vector machine classifier classification schematic diagram.
Embodiment
The present invention will be described in detail below in conjunction with the accompanying drawings.
Fig. 1 shows the structured flowchart of the multi-storied garage safety-protection system based on intelligent video monitoring according to the present invention.
In the present embodiment, the multi-storied garage safety-protection system based on intelligent video monitoring includes button, web camera, router, master
Machine, display, warning system and controller, wherein, the button connects to main frame send to main frame and move car instruction;The net
Network video camera is arranged on the back upper place both sides of multi-storied garage, and monitored picture covers whole multi-storied garage, for gathering multi-storied garage
Interior video image information.In the present embodiment, web camera is infrared high-definition network camera, is separately mounted to solid
Garage back upper place both sides, can effectively reduce monitoring blind area, and can be shot in the environment of dark.
The router is located between web camera and main frame, respectively with the web camera and the two-way company of main frame
Connect, be used in uplink communication to main frame transmitting video image information, be used to trigger web camera operation in downlink communication.
Router one end can connect multiple web cameras, one end connection main frame, the letter that router can gather web camera
Breath is transferred in main frame, the instruction of main frame can also be passed into web camera, it is started working.
The main frame is unidirectionally connected to the controller, and main frame receives the video image information gathered by web camera,
And judged whether that car can be moved according to the video image information of collection, and the controller is sent a signal to, if stereoscopic vehicle
Unusual condition is not detected in storehouse, then sends shifting car instruction to controller, if detecting unusual condition, for example, is existed dynamic
Thing, then send to controller and forbid moving car instruction, the controller is performed according to the signal received to be moved car instruction or perform taboo
Only move car instruction.
The display is connected to the main frame, for showing the video image in the multi-storied garage that video camera is shot, with
Just staff can be intuitive to see the situation in multi-storied garage, so as to artificially control multi-storied garage safety-protection system.
The warning system is connected to the main frame, for sending alarm when noting abnormalities, and the warning system can be by
In multi-storied garage, video monitoring room can also be arranged on.
Fig. 2 shows the workflow diagram of the multi-storied garage safety protection method based on intelligent video monitoring of the present invention.Specifically
Ground, comprises the following steps:
When leaving multi-storied garage after car owner's parking is finished, car owner or garage keeper need to press button, are sent out to main frame
Go out to move car instruction;Main frame is received after shifting car instruction, is sent trigger signal to web camera by router, is made network shooting
Machine, which is triggered, to bring into operation;Web camera transmits the video image information of collection to main frame by router;Main frame to regarding
Frequency image information is handled and analyzed, and analysis and processing of the main frame to the video image information of collection are main to be included to motion mesh
Target detects and the moving target Land use models knowledge method for distinguishing of detection is classified, so as to judge whether deposited in multi-storied garage
In exception, if human or animal, then it is determined as abnormal conditions;If winged insect, refuse bag etc., then be determined as normal condition.
If normal, main frame sends shifting car to controller and instructed, and normally moves car, if there is exception, by abnormal conditions with the shape of switching signal
Formula feeds back to controller, forbids vehicle to move, and can be sent out alarm pointed out.After abnormal conditions are disposed, after
Continuous to carry out image variants, controller starts normally to move car operation.
The video image information that main frame is gathered to web camera is handled and analyzed including moving object detection and right
Moving target is classified.The image sequence of video camera recording is divided automatically using machine vision and the method for video analysis
Analysis, realizes positioning, the identification to target in dynamic scene, and analyzes and judge the behavior of target on this basis, draws to figure
As the understanding and the explanation to objective scene of content implication, so as to instruct and planning action.Carried out referring to Fig. 3 and Fig. 4
It is described in detail.
First, it is sequence of video images that main frame, which will obtain collection sample changeover, and carries out image preprocessing to it, is treated
Detection image, then carries out moving object detection to image to be detected.Herein on the basis of studying target detection technique, carry
A kind of background modeling method based on Gaussianmixture model is gone out, has been established using the method for statistics based on color and color gradient
Background model, and background model is updated in real time, finally both background models is considered target is carried out
Effective detection.The problems such as algorithm preferably solves the extraction of background model, renewal, background perturbation, extraneous illumination variation.
Key step is as follows:
1) background modeling:Automatically the advantage that background model is background difference algorithm is set up, can be different according to scene, learning
The habit stage inputs training video image to algorithm, is trained by certain time, is automatically learned the parameter of background model;
2) sport foreground is detected:As soon as often carrying out two field picture, present image and background model are carried out difference, i.e., with background mould
Type is matched.It can allow the size of algorithm dynamic adjusted threshold according to scene, be prevented effectively from missing inspection and the flase drop of moving region,
The robustness of algorithm is improved simultaneously;
3) post-processing:Result to background difference is further processed, and obtains more accurate complete moving target
Region, such as removes shade;
4) target detection:Moving target scope is confined in original image;
5) background model updates:According to background difference and the result of target detection, determine whether background model needs renewal,
Algorithm is adapted to the change of various complex scenes, improve the accuracy rate of detection, it is to have influence on background subtraction to select suitable turnover rate
Divide the key of effect.
Using moving target at background model identification, and feature extraction is made to moving target, carried out according to the feature of extraction
Moving object classification.Assorting process is divided into offline training process and online decision process, offline training process needs pair
The model of foundation is trained, and can be applied to online decision process.
The present invention carries out pattern-recognition using SVMs, judges whether it is dangerous abnormal image, choose stabilization and
It can reflect that the feature of object essence, as input vector, filters out smaller moving target, eliminate noise, effectively reduce the instruction of model
Practice the time.
SVMs is a new technology in data mining, is to solve Machine Learning Problems by means of optimal method
New tool.It is based on Statistical Learning Theory, and its structural risk minimization avoids " cross and learn " phenomenon of neutral net,
And with good generalization ability.It is minimum etc. real that supporting vector function preferably solves small sample, non-linear, high dimension drawn game portion
Border problem.When handling nonlinear problem, it is translated into the linear problem in higher dimensional space first, then with a core letter
Count to replace the inner product operation in higher dimensional space, so as to dexterously solve complicated calculations problem.SVM is wide in the application of each field
It is general, it is adapted in pattern-recognition build grader, its basic thought is:A hyperplane is constructed by two different set point
Open, as shown in Figure 3.Sample can be divided for two classes in plane, the task of machine learning is to find straight line, can not only be two
Class sample is separated, and ensures that class interval is maximum.So-called class interval refers to from this straight line into two class samples nearest sample
This apart from sum, and these minimum distance samples are SVMs.Complicated classification problem can first pass through non-linear
The input space is transformed to a higher dimensional space by mapping, then obtains optimal classification surface in the higher dimensional space.SVMs
The schematic diagram of classification is as shown in Figure 3.For given training set:T={ (x1,y1),…,(xn,yn)}∈(X×Y)n, wherein, xi
∈ X=Rn,yi∈ Y={ 1, -1 } (i=1,2 ..., n), xiVector is characterized, there is straight line g (x)=wTX+b makes all yi
=-1 point falls in g (x)<0 one side, all yi=+1 point falls in g (x)>0 one side.And maximum apart from both sides border
It is known as optimal hyperlane.Standard supporting vector grader now is:
yi[w·xi+ b] -1 >=0i=1,2 ..., N (1)
Supporting hyperplane is to the distance of optimal hyperlane:D=1/ | | w | |, the distance between two Optimal Separating Hyperplanes
Margin is two times of d, i.e.,:Margin=2d=2/ | | w | |.It is known as the problem of optimal hyperlane is to solve for:
However, in the case that sample data is complicated and changeable, optimal hyperlane is difficult that all samples are done into one accurately
Divide, therefore, introducing relaxation factor ξi, its purpose is to loosen restrictive condition, it is allowed to certain wrong distribution life, then formula
(1) sorter model is changed into:
yi[w·xi+b]-1+ξi>=0i=1,2 ..., N (3)
Now, the quadratic programming problem for solving optimal hyperlane is just changed into:
In above formula, c is the punishment parameter of error, wrong point sample proportion and algorithm complex can be traded off, true
C value is small in fixed proper subspace represents small to the punishment of experience error, and the complexity of Learning machine is small and empiric risk
Value is larger;C takes infinity, then all constraints must all is fulfilled for, it means that training sample must classify exactly.Often
Individual proper subspace at least make it that SVM Generalization Abilities are best in the presence of a suitable c.
Sample in Practical Project is often nonlinear, it is therefore desirable to which input space x is passed through into certain Nonlinear MappingA high-dimensional feature space is mapped to, linear optimal separating hyper plane is reconstructed in this space.This mapping is usual
Realized with designing the method for kernel function, using different kernel functions with regard to various forms of Nonlinear Support Vector Machines can be obtained.
Conventional kernel function has following four type:
(1) linear kernel function:K (x, x ')=(xx ')
(2) Polynomial kernel function:K (x, x ')=((xx ')+1)q
(3) RBF kernel functions:
(4) Sigmoid functions:K (x, x ')=((xx ')+1)q
The sorter model constructed in space after Nonlinear Mapping is as follows:
Now, it is desirable to which the object function of solution is changed into:
In summary, SVM basic thought is exactly that the nonlinear change defined by interior Product function converts the input space
To a higher dimensional space, then optimal classification surface is sought in this new space.
The emulation experiment of the present invention is to be carried out based on open source software LIBSVM with MATLAB2016b platforms, and LIBSVM is platform
Simple, easy to use and fast and effectively SVM pattern-recognitions and the recurrence that the exploitation such as gulf university professor Lin Zhiren is designed
Software kit.The software not only provide it is compiled can Windows serial systems execution file, additionally provide source code, side
Just improve, change and applied in other operating systems.
When being trained to grader, the penalty parameter c and nuclear parameter g of its training pattern are (with the σ in kernel function each other
It is reciprocal) selection crucial effect is played to the accuracy of classification results, but using net the selection of the two parameters so far more
Lattice are searched for and cross-validation method.However, grid search be included in certain scope according to artificially given density it is all
Possible parameter combination, then finds out the minimum parameter combination of predicated error as its optimal solution, its amount of calculation is huge again.Intersect
Then each subset is trained set checking collection, Jin Erfa by proof method each other again by being N number of subset by training sample random division
Existing optimal parameter combination.But due to be every time it is random split training data, therefore the Optimal Parameters obtained every time are also not to the utmost
Identical, along with amount of calculation is relatively large, when training samples number increases, search procedure excessively expends the time.Therefore, this hair
The bright parameter combination optimal using improved gravitation searching algorithm Automatic-searching, builds optimal decision function so that SVM points
The accuracy highest of class, then obtains improved universal gravitation algorithm-support vector machine classifier, i.e. IGSA-SVM graders.
Gravitation searching algorithm (Gravitational Search Algorithms, GSA) is that one kind is come to physics
Gravitation in carries out the swarm intelligence optimization method of simulation generation, as a kind of meta-heuristic algorithm, the search in GSA
Particle is to act on attracting each other by gravitation between one group of object in space motion, object, and the motion of object follows dynamic
Laws of Mechanics.Because gravitational effect can cause objects to be moved towards the maximum object of quality, and the maximum thing of quality
Body occupies optimal location, so as to obtain the optimal solution of optimization problem.The algorithm passes through the gravitation phase interaction between object
With the shared of optimization information is realized, guiding colony is searched for optimal solution regional implementation.
The present invention proposes a kind of improved gravitation searching algorithm (IGSA).By introducing backward learning strategy, essence
English strategy and boundary mutation strategy, make basic gravitation searching algorithm have higher optimization precision and stable
Property, tested finally by 13 reference functions, to verify that improved gravitation searching algorithm has preferably optimization property
Energy.Support vector machines grader is optimized using improved universal gravitation algorithm IGSA, SVM training moulds can be sought
The penalty parameter c of type and nuclear parameter g optimal solution, realization are classified to moving target.
Moreover, it will be appreciated that although the present specification is described in terms of embodiments, not each embodiment is only wrapped
Containing an independent technical scheme, this narrating mode of specification is only that for clarity, those skilled in the art should
Using specification as an entirety, the technical solutions in the various embodiments may also be suitably combined, forms those skilled in the art
It may be appreciated other embodiment.
Claims (10)
1. a kind of multi-storied garage safety-protection system based on intelligent video monitoring, described vertical applied to the multi-storied garage of lift-sliding
Body garage safety-protection system includes button, at least one web camera, router, main frame and controller, wherein:
The button connects send to main frame to main frame and move car instruction;
The web camera is arranged on the back upper place of multi-storied garage, and monitored picture covers whole multi-storied garage, vertical for gathering
Video image information in body garage;
The router is located between web camera and main frame, is bi-directionally connected respectively with the web camera and main frame,
It is used for during uplink communication to main frame transmitting video image information, is used to trigger web camera operation in downlink communication;
The main frame is unidirectionally connected to the controller, and main frame judges whether that car can be moved according to the video image information of collection,
And send a signal to the controller;
The controller performs to move car instruction or perform according to the signal received to be forbidden moving car instruction.
2. multi-storied garage safety-protection system according to claim 1, it is characterised in that the multi-storied garage safety-protection system is also wrapped
Display is included, the display is connected to the main frame, for showing the video image in multi-storied garage.
3. multi-storied garage safety-protection system according to claim 1 or 2, it is characterised in that the multi-storied garage safety-protection system
Also include warning system, the warning system is connected to the main frame, for sending alarm when noting abnormalities.
4. multi-storied garage safety-protection system according to claim 1, it is characterised in that the web camera is infrared high definition
Web camera, is separately mounted to multi-storied garage back upper place both sides.
5. a kind of multi-storied garage safety protection method based on intelligent video monitoring, it is characterised in that comprise the following steps:S1. press
Button, shifting car instruction is sent to main frame;
S2. main frame is received after shifting car instruction, is sent signal to web camera by router, web camera is triggered
Bring into operation;
S3. web camera transmits the video image information of collection to main frame by router;
S4. main frame is handled and analyzed to video image information, is judged in multi-storied garage with the presence or absence of abnormal;
If S5. normal, main frame sends shifting car to controller and instructed, and normally moves car, if there is exception, main frame sends taboo to controller
Car instruction is only moved, forbids moving car.
6. multi-storied garage safety protection method according to claim 5, it is characterised in that also include, if exist it is abnormal, main frame to
Warning system sends police instruction, and alarm sounds are sent by warning system.
7. the multi-storied garage safety protection method according to claim 5 or 6, it is characterised in that right in the step S4 of methods described
The processing and analysis of video image information include carrying out moving object detection, feature extraction and fortune to the video image information of collection
Moving-target is classified.
8. multi-storied garage safety protection method according to claim 7, it is characterised in that main frame is used based on mixing Gauss moulds
The background modeling method of type carries out moving object detection to the video image information of collection.
9. multi-storied garage safety protection method according to claim 8, it is characterised in that SVMs is designed in main frame
SVM classifier is classified to moving target.
10. multi-storied garage safety protection method according to claim 9, it is characterised in that utilize improved universal gravitation algorithm
IGSA is optimized to support vector machines grader, is designed IGSA-SVM graders, is improved and moving target is classified
Accuracy rate.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710638562.2A CN107273880A (en) | 2017-07-31 | 2017-07-31 | A kind of multi-storied garage safety-protection system and method based on intelligent video monitoring |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710638562.2A CN107273880A (en) | 2017-07-31 | 2017-07-31 | A kind of multi-storied garage safety-protection system and method based on intelligent video monitoring |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107273880A true CN107273880A (en) | 2017-10-20 |
Family
ID=60076097
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710638562.2A Pending CN107273880A (en) | 2017-07-31 | 2017-07-31 | A kind of multi-storied garage safety-protection system and method based on intelligent video monitoring |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107273880A (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108181840A (en) * | 2017-12-29 | 2018-06-19 | 贵州航天南海科技有限责任公司 | A kind of stereo garage monitoring control system |
CN108196479A (en) * | 2017-12-29 | 2018-06-22 | 贵州航天南海科技有限责任公司 | A kind of three-dimensional garage control system |
CN108830136A (en) * | 2018-04-26 | 2018-11-16 | 杨春花 | A kind of stereo garage safety protection method based on big data analysis |
CN109445315A (en) * | 2018-08-21 | 2019-03-08 | 成都理工大学 | High-voltage test hall safety guard and method of controlling security |
CN110909689A (en) * | 2019-11-26 | 2020-03-24 | 浪潮软件股份有限公司 | Kitchen monitoring method and system |
CN110942205A (en) * | 2019-12-05 | 2020-03-31 | 国网安徽省电力有限公司 | Short-term photovoltaic power generation power prediction method based on HIMVO-SVM |
CN111310736A (en) * | 2020-03-26 | 2020-06-19 | 上海同岩土木工程科技股份有限公司 | Rapid identification method for unloading and piling of vehicles in protected area |
CN111931587A (en) * | 2020-07-15 | 2020-11-13 | 重庆邮电大学 | Video anomaly detection method based on interpretable space-time self-encoder |
CN112668433A (en) * | 2020-12-22 | 2021-04-16 | 成都睿畜电子科技有限公司 | Farm management method, farm management device, farm management medium and farm management equipment |
CN113593250A (en) * | 2021-07-12 | 2021-11-02 | 浙江工贸职业技术学院 | Illegal parking detection system based on visual identification |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102955461A (en) * | 2011-08-31 | 2013-03-06 | 鸿富锦精密工业(深圳)有限公司 | System and method for monitoring mechanical-type parking spaces |
CN104951775A (en) * | 2015-07-15 | 2015-09-30 | 攀钢集团攀枝花钢钒有限公司 | Video technology based secure and smart recognition method for railway crossing protection zone |
CN204691363U (en) * | 2015-06-04 | 2015-10-07 | 四川五新智能设备有限公司 | A kind of obstacle avoidance system being applied to multi-storied garage parking position jacking equipment |
-
2017
- 2017-07-31 CN CN201710638562.2A patent/CN107273880A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102955461A (en) * | 2011-08-31 | 2013-03-06 | 鸿富锦精密工业(深圳)有限公司 | System and method for monitoring mechanical-type parking spaces |
CN204691363U (en) * | 2015-06-04 | 2015-10-07 | 四川五新智能设备有限公司 | A kind of obstacle avoidance system being applied to multi-storied garage parking position jacking equipment |
CN104951775A (en) * | 2015-07-15 | 2015-09-30 | 攀钢集团攀枝花钢钒有限公司 | Video technology based secure and smart recognition method for railway crossing protection zone |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108181840A (en) * | 2017-12-29 | 2018-06-19 | 贵州航天南海科技有限责任公司 | A kind of stereo garage monitoring control system |
CN108196479A (en) * | 2017-12-29 | 2018-06-22 | 贵州航天南海科技有限责任公司 | A kind of three-dimensional garage control system |
CN108830136A (en) * | 2018-04-26 | 2018-11-16 | 杨春花 | A kind of stereo garage safety protection method based on big data analysis |
CN109445315A (en) * | 2018-08-21 | 2019-03-08 | 成都理工大学 | High-voltage test hall safety guard and method of controlling security |
CN110909689A (en) * | 2019-11-26 | 2020-03-24 | 浪潮软件股份有限公司 | Kitchen monitoring method and system |
CN110942205A (en) * | 2019-12-05 | 2020-03-31 | 国网安徽省电力有限公司 | Short-term photovoltaic power generation power prediction method based on HIMVO-SVM |
CN110942205B (en) * | 2019-12-05 | 2022-12-06 | 国网安徽省电力有限公司 | Short-term photovoltaic power generation power prediction method based on HIMVO-SVM |
CN111310736A (en) * | 2020-03-26 | 2020-06-19 | 上海同岩土木工程科技股份有限公司 | Rapid identification method for unloading and piling of vehicles in protected area |
CN111931587A (en) * | 2020-07-15 | 2020-11-13 | 重庆邮电大学 | Video anomaly detection method based on interpretable space-time self-encoder |
CN112668433A (en) * | 2020-12-22 | 2021-04-16 | 成都睿畜电子科技有限公司 | Farm management method, farm management device, farm management medium and farm management equipment |
CN113593250A (en) * | 2021-07-12 | 2021-11-02 | 浙江工贸职业技术学院 | Illegal parking detection system based on visual identification |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107273880A (en) | A kind of multi-storied garage safety-protection system and method based on intelligent video monitoring | |
CN106241533B (en) | Elevator occupant's comprehensive safety intelligent control method based on machine vision | |
CN104123544B (en) | Anomaly detection method and system based on video analysis | |
CN106219367B (en) | A kind of elevator O&M monitoring method based on intelligent vision light curtain | |
KR20180135898A (en) | Systems and methods for training object classifiers by machine learning | |
CN110428522A (en) | A kind of intelligent safety and defence system of wisdom new city | |
CN102156880B (en) | Method for detecting abnormal crowd behavior based on improved social force model | |
CN106571014A (en) | Method for identifying abnormal motion in video and system thereof | |
CN106006266B (en) | A kind of machine vision method for building up applied to elevator safety monitoring | |
CN107220603A (en) | Vehicle checking method and device based on deep learning | |
CN105426820B (en) | More people's anomaly detection methods based on safety monitoring video data | |
CN106128053A (en) | A kind of wisdom gold eyeball identification personnel stay hover alarm method and device | |
CN106241534B (en) | More people's boarding abnormal movement intelligent control methods | |
CN110298278A (en) | A kind of underground parking garage Pedestrians and vehicles monitoring method based on artificial intelligence | |
KR20200071799A (en) | object recognition and counting method using deep learning artificial intelligence technology | |
CN105744232A (en) | Method for preventing power transmission line from being externally broken through video based on behaviour analysis technology | |
CN104166861A (en) | Pedestrian detection method | |
CN109948455B (en) | Detection method and device for left-behind object | |
CN103069434A (en) | Multi-mode video event indexing | |
CN108363953B (en) | Pedestrian detection method and binocular monitoring equipment | |
TW200905575A (en) | Method for finding paths in video | |
CN109255298A (en) | Safety cap detection method and system in a kind of dynamic background | |
CN104134364B (en) | Real-time traffic sign identification method and system with self-learning capacity | |
CN102163290A (en) | Method for modeling abnormal events in multi-visual angle video monitoring based on temporal-spatial correlation information | |
CN103632427B (en) | A kind of gate cracking protection method and gate control system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171020 |