CN109377703A - A kind of forest fireproofing early warning system and its method based on machine vision - Google Patents
A kind of forest fireproofing early warning system and its method based on machine vision Download PDFInfo
- Publication number
- CN109377703A CN109377703A CN201811488530.XA CN201811488530A CN109377703A CN 109377703 A CN109377703 A CN 109377703A CN 201811488530 A CN201811488530 A CN 201811488530A CN 109377703 A CN109377703 A CN 109377703A
- Authority
- CN
- China
- Prior art keywords
- image
- flame
- module
- pixel
- video
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004079 fireproofing Methods 0.000 title claims abstract description 18
- 238000000034 method Methods 0.000 title claims abstract description 17
- 238000001514 detection method Methods 0.000 claims abstract description 29
- 238000000605 extraction Methods 0.000 claims abstract description 21
- 239000000203 mixture Substances 0.000 claims abstract description 16
- 238000012544 monitoring process Methods 0.000 claims abstract description 10
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 9
- 238000012545 processing Methods 0.000 claims description 25
- 238000007781 pre-processing Methods 0.000 claims description 12
- 239000000284 extract Substances 0.000 claims description 10
- 230000000007 visual effect Effects 0.000 claims description 10
- 238000004891 communication Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 8
- 230000000877 morphologic effect Effects 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 2
- 108010046685 Rho Factor Proteins 0.000 claims 1
- 239000002023 wood Substances 0.000 abstract description 4
- 230000006399 behavior Effects 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 5
- 238000000926 separation method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 241000872198 Serjania polyphylla Species 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 239000003086 colorant Substances 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000002265 prevention Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 244000025254 Cannabis sativa Species 0.000 description 1
- 244000249914 Hemigraphis reptans Species 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B17/00—Fire alarms; Alarms responsive to explosion
- G08B17/005—Fire alarms; Alarms responsive to explosion for forest fires, e.g. detecting fires spread over a large or outdoors area
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
- G06V20/42—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
-
- 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
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B17/00—Fire alarms; Alarms responsive to explosion
- G08B17/12—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions
- G08B17/125—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions by using a video camera to detect fire or smoke
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- 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
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biodiversity & Conservation Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Closed-Circuit Television Systems (AREA)
- Image Analysis (AREA)
- Alarm Systems (AREA)
Abstract
The invention discloses a kind of forest fireproofing early warning system and its method based on machine vision, forest fireproofing early warning system of the invention, including collection terminal and mainly by moving target foreground detection module, flame characteristic extraction module, flame tracking and identification module, the monitor supervision platform that alarm module is constituted carries out the monitoring of 24 hours continual fire the condition of a disasters to realize to wood land, and the generation of fire can be recognized accurately and flame is tracked, fire alarm signal can be issued in time by monitor supervision platform, relevant staff is set accurately to make Disaster Relief Measures, it is practical.The present invention identifies whether occur the things moved in video image when detecting to video image using mixture Gaussian background model, modeling algorithm is perfect, has the extraction of target in video with performance well, 92% or more can achieve to the discrimination of fire, the recognition accuracy of fire the condition of a disaster generation can be significantly improved.
Description
Technical field
The present invention relates to a kind of fire prevention early warning system, especially a kind of forest fireproofing early warning system based on machine vision and
Its method.
Background technique
Since the generation of fire brings huge loss to the life and property of the people, the early warning before fire generation
Work is particularly important.The fire early-warning system that China uses at present, mostly be using ground patrol formula forest fire protection,
The monitor modes such as Wang Tai is monitored, aviation is patrolled work to complete fire alarm.And the above-mentioned three kinds of modes used cannot achieve
24 hours real time monitorings, in the biggish region of forest range, particularly dry season, if the initial stage occurred in the condition of a disaster
The rapid sprawling that cannot be found in time and corresponding fire suppression measures is taken to will lead to fire, causes the loss for being difficult to retrieve.
To overcome drawbacks described above existing for forest fire protection, forest fireproofing early warning system is applied and is given birth to.Forest fire protection early warning system
System, which is one, carries out forest with modernization advanced technology (such as Image Acquisition, image identifying and processing, intelligent control etc.)
The intellectualizing system of monitoring free of discontinuities for 24 hours, use process it can find fire and quasi- in the initial stage that fire occurs
It really identifies size and the fire spot of fire, while promptly relevant staff being reminded to carry out fire extinguishing disaster relief work.Example
Such as the utility model patent that Authorization Notice No. is CN201707774U, a kind of forest fireproofing early warning system is disclosed, is by front end
Monitoring arrangement, communication and control server, intelligent fire behavior search or fire behavior differentiate server, monitor client, GIS geography information
System composition, communication differentiate server signal with front end monitoring arrangement and intelligent fire behavior search or fire behavior respectively with control server
Connection, monitor client differentiate server and GIS geography letter with control server, the search of intelligent fire behavior or fire behavior with communicate respectively
Cease system signal connection.The search of its intelligent fire behavior used or fire behavior differentiate server for the identification of visible light pyrotechnics and infra red flame
Identification, although how can specifically be differentiated with round-the-clock uninterrupted monitoring for fire, how equal the accuracy of differentiation is
It is not disclosed in detail, can not precisely be handled especially for the picture collected, flame can not be tracked, be caused
Real-time online can not be accomplished to the judgement of the condition of a disaster.
Summary of the invention
Goal of the invention of the invention is, in view of the above-mentioned problems, providing a kind of forest fire protection early warning system based on machine vision
System, this system can uninterruptedly carry out the monitoring of fire the condition of a disaster in 24 hours to wood land, and the hair of fire can be recognized accurately
It gives birth to and the condition of a disaster is tracked, issue fire alarm signal in time, relevant staff is enable accurately to make Disaster Relief Measures.
In order to achieve the above objectives, the technical scheme adopted by the invention is that:
A kind of forest fireproofing early warning system based on machine vision, including collection terminal and monitor supervision platform, the collection terminal point
Cloth is at forest zone scene to be monitored, for obtaining the video image at forest zone scene in real time and being sent to monitor supervision platform, the prison
Module is connect control platform with collection terminal by wireless communication, and the monitor supervision platform includes moving target foreground detection module, image
Preprocessing module, flame characteristic extraction module, flame tracking and identification module and alarm module,
The moving target foreground detection module receives the video image that collection terminal is sent, and first by being integrated in fortune
Foreground detection unit in moving-target foreground detection module detects video image: by the present frame picture in video image
It is set as foreground image, while the former frame picture of present frame is set as background image;Then background model is constructed, background mould is passed through
Whether type there is the things of movement in the video image to differentiate input: if the pixel of foreground image and background image is without frame
When difference variation, then background image is updated;Conversely, if the pixel of foreground image and background image is considered as when generating the variation of frame difference
There is moving object in frequency, and foreground image generation target image is sent to flame characteristic extraction module;
The flame characteristic extraction module first passes through what image pre-processing module sent moving target foreground detection module
Target image carries out image procossing, and target image is then converted to HSI color identification model by rgb space image, passes through meter
Differentiate whether the location of pixels is flame pixels point after calculating the saturation degree component value S in HSI color identification model, is the then figure
It is on the contrary then be shown as black as pixel is shown as white and extracts feature;
Flame tracking and identification module identify in video image according to HSI color identification model there are after flame,
Flame characteristic is constantly extracted by flame characteristic extraction module, and is regarded according to obtained flame characteristic position is extracted using opencv
Feel that the API in image processing module draws flame contours, and locating and tracking lookup is carried out to flame contours, is issued to alarm module
Fire alarm signal.
Preferably, the forest fireproofing early warning system may also include mobile terminal, and the mobile terminal is logical
Communication module is crossed to connect with alarm module.
Preferably, the input terminal of described image preprocessing module and moving target foreground detection module connect
It connects, the output end of image pre-processing module is connect with flame characteristic extraction module.
Preferably, the background model that the moving target foreground detection module uses is mixed Gaussian background
Model is the description for being fitted background by establishing multiple Gaussian Profiles to realize to multimodal distribution background.
Preferably, the collection terminal includes n video camera, and n video camera is respectively distributed to each of forest zone
Corner, and each video camera is respectively connected with a network-control holder, the network-control holder receives what monitor supervision platform was sent
Actuated camera obtains the video image of respective angles, the integer that n is >=10 according to control instruction after control instruction.
The present invention also provides a kind of forest fire protection method for early warning based on machine vision, includes the following steps:
S1, video image obtain: controlling the video image that collection terminal obtains forest zone scene by monitor supervision platform;
S2, target prospect detection: first detecting the step S1 video image sent by foreground detection unit, will
Present frame picture in video image is set as foreground image, while the former frame picture of present frame is set as background image;Then
Background model is constructed, whether occurs the thing of movement in video image of the obtained background model to differentiate input by constructing
Object updates background image if the pixel of foreground image and background image changes without frame difference;Conversely, if foreground image and back
The pixel of scape image is considered as in video when generating the variation of frame difference moving object, and foreground image is generated target image;
S3, flame characteristic extract: after the target image that step S2 is obtained carries out image procossing, then by target image by
Rgb space image is converted to HSI color identification model, then proceeds as follows:
S31, tri- channels R, G, B in RGB are separated respectively, sets the saturation threshold values of red component R in RGB image
For RT, and R > RT;
S32, according to R > R in step S1TCondition judges the channel R;
S33, the saturation degree component value S that target image is calculated based on flame pixels index (1-1) and formula (1-2), are led to
Degree of supersaturation component value S determines whether the location of pixels is flame pixels point, is, the image slices vegetarian refreshments be shown as white and
It extracts feature and carries out position mark, it is on the contrary then be shown as black;
R > G > B (1-1)
S≥((255-R)*ST/RT (1-2)
R is the saturation value of red component in RGB, R in formulaTIt is the threshold value of red component R, S is in HSI color identification model
Saturation degree component value, STIt is the saturation degree threshold value of S;
S4, flame locating and tracking: it identifies that there are flames in video image by step S2, and is constantly extracted by step S3
Behind the home position of obtained flame characteristic and corresponding flame characteristic, using the API in opencv visual pattern processing module
Flame contours are drawn out, and call the locating query algorithm for drawing rectangle frame in opencv visual pattern processing module in video
Flame region in image carries out the locating and tracking of pixel, while being marked conflagration area with rectangle frame;
S5, alarm: the flame locating and tracking result of step S4 is sent to alarm module.
Further, in step S2, the background model is mixture Gaussian background model, the mixture Gaussian background model
Modeling algorithm it is as follows:
In formula, k is distribution pattern sum, ε (xt, μI, t, τI, t) it is i-th of Gaussian Profile of t moment, μI, tFor its mean value,
τI, tFor its covariance matrix, δI, tFor variance, I is three-dimensional unit matrix, WI, tFor the weight of t moment i-th Gaussian Profile.
It is further preferred that the mixture Gaussian background model differentiates in the video image inputted the thing moved whether occur
The method of object is as follows:
S21, each new pixel value Xt are pressed with current k model | Xt- εI, t=1|≤2.5σI, t=1Condition formula is calculated
Compare, until there is a new pixel to occur, i.e., pixel deviations guarantee in 2.5 σ;
If S22, the matched mode of institute meet context request, which belongs to background, otherwise belongs to prospect;
S23, each schema weight update as follows, and wherein α is learning rate, for matched mode Mk,t=1, it is no
Then Mk,t=0, then the weight of each mode is normalized:
wK, t=(1- ρ) * wK, t-1+ρ*MK, t (2-4)
S24, the mean μ of non-match pattern and standard deviation sigma are constant, and the parameter of match pattern is updated according to following, by as follows
Formula (2-5), (2-6), (2-7) are updated;
ρ=α * ε (Xt|μk, σk) (2-5)
μt=(1- ρ) * μt-1+ρ*Xt (2-6)
In formula, k is the distribution of distribution pattern sum, ε (xt|μk, σk) indicate the kth moment Gaussian Profile, μtFor t moment
The mean value of pixel, σtFor the variance of t moment pixel, σkIndicate the variance of k moment pixel, μkWhat is indicated is the kth moment
The mean value of pixel, α indicate updating factor, and ρ indicates the related coefficient of X and ρ;
If S25, step S22 are not carried out, pixel will carry out descending arrangement according to weight size.
Further, in step S3, described image processing the step of it is as follows:
S311, colored target image is subjected to gray proces, that is, calculates the average value in tri- channels R, G, B in RGB,
Then the average value being calculated is attached to tri- components of R, G, B again respectively;
S312, the gray scale picture for obtaining step S01 carry out binary conversion treatment, that is, set the threshold value of red component R as RT,
RTValue estimate to obtain by the histogram of image, be greater than RTThe pixel of threshold range is designated as 1, i.e., white, is less than RTThe picture of threshold value
Vegetarian refreshments is designated as 0, is black;
S313, Morphological Gradient is carried out to the image that step S02 is handled using opencv visual pattern processing module
After processing, the red component in image made is more obvious, convenient for the subsequent RGB triple channel of flame characteristic extraction module
Separation.
Due to the adoption of the above technical scheme, the invention has the following advantages:
1. forest fireproofing early warning system of the invention, including collection terminal and mainly by moving target foreground detection module, fire
The monitor supervision platform that flame characteristic extracting module, flame tracking are constituted with identification module, alarm module carries out wood land to realize
The monitoring of 24 hours continual fire the condition of a disasters, and the generation of fire can be recognized accurately and flame is tracked, it can pass through
Monitor supervision platform issues fire alarm signal in time, and relevant staff is enable accurately to make Disaster Relief Measures, practical.
2. the present invention is to identify in video image using mixture Gaussian background model when detecting to video image
No the things moved occur, modeling algorithm is perfect, has the extraction of target in video with performance well, passes through the applicant
Realistic simulation 92% or more can achieve to the discrimination of fire using this forest fireproofing early warning system, fire can be significantly improved
The recognition accuracy that calamity the condition of a disaster occurs.
3. the present invention knows for the feature extraction combination RGB color model and HSI color identification model of flame, and in flame
Not with using the opencv visual pattern processing module based on machine vision, improve the same of flame characteristic recognition accuracy when tracking
When flame can also be tracked, data processing method is simple, system sensitivity is high.
4, the present invention is additionally provided with image pre-processing module, carries out ash respectively to video image by image pre-processing module
Degree processing, binary conversion treatment and Morphological Gradient processing, wherein the gray proces of image can reduce the calculation amount of system, to ash
It is black with white two kinds of colors that degree image progress binary conversion treatment can not only be such that the color of image is only left, and the profile of image is more clear
It is clear, it is convenient that the profile of figure is further analyzed, the primitiveness of image can also be kept;At the Morphological Gradient of image
It manages and can remove noise jamming, so that the color profile of image is more clear the separation for obviously facilitating and carrying out RGB triple channel, make image
Profile highlight.
Detailed description of the invention
Fig. 1 is a kind of system block diagram of the forest fireproofing early warning system based on machine vision of the present invention.
Fig. 2 is a kind of flow diagram of the forest fire protection method for early warning based on machine vision of the present invention.
Fig. 3 is present invention tracking and signal execution block diagram when identification flame.
Specific embodiment
It is further illustrated below in conjunction with specific implementation of the attached drawing to invention.
As shown in Figure 1, a kind of forest fireproofing early warning system based on machine vision, including collection terminal, monitor supervision platform and shifting
Dynamic terminal.The collection terminal is distributed in forest zone scene to be monitored, for obtaining video image and the transmission at forest zone scene in real time
To monitor supervision platform.Module is connect the monitor supervision platform with collection terminal by wireless communication.
The monitor supervision platform include moving target foreground detection module, image pre-processing module, flame characteristic extraction module,
Flame tracking and identification module and alarm module.The mobile terminal is connect by communication module with alarm module.The communication
Module is specially gsm module, GPRS module, any one being wirelessly connected in chip, is used for fire alert information with short message
Or voice mail or the mode of wireless transmission are sent on mobile terminal.
The moving target foreground detection module passes through be integrated in first for receiving the video image that collection terminal is sent
Foreground detection unit in moving target foreground detection module detects video image: the present frame in video image is drawn
Face is set as foreground image, while the former frame picture of present frame is set as background image;Then background model is constructed, background is passed through
Model whether occur in the video image to differentiate input movement things: if the pixel of foreground image and background image without
When frame difference changes, then background image is updated;Conversely, if the pixel of foreground image and background image is considered as when generating the variation of frame difference
There is moving object in video, and foreground image generation target image is sent to flame characteristic extraction module.
The flame characteristic extraction module first passes through what image pre-processing module sent moving target foreground detection module
Target image carries out image procossing, and target image is then converted to HSI color identification model by rgb space image, passes through meter
Differentiate whether the location of pixels is flame pixels point after calculating the saturation degree component value S in HSI color identification model, is the then figure
It is on the contrary then be shown as black as pixel is shown as white and extracts feature.
Flame tracking and identification module identify in video image according to HSI color identification model there are after flame,
Flame characteristic is constantly extracted by flame characteristic extraction module, and is regarded according to obtained flame characteristic position is extracted using opencv
Feel that the API in image processing module draws flame contours, and locating and tracking lookup is carried out to flame contours, is issued to alarm module
Fire alarm signal.
Specifically, the input terminal of described image preprocessing module is connect with moving target foreground detection module, and image is located in advance
The output end of reason module is connect with flame characteristic extraction module.
The background model that the moving target foreground detection module uses is more by establishing for mixture Gaussian background model
A Gaussian Profile is come the description that is fitted background to realize to multimodal distribution background.
The collection terminal includes n video camera, and n video camera is respectively distributed to each corner in forest zone, and each is imaged
Machine is respectively connected with a network-control holder, and navigation module is equipped on video camera.It is flat that the network-control holder receives monitoring
Actuated camera obtains the video image of respective angles according to control instruction after the control instruction that platform is sent, n be >=10 it is whole
Number.Each network-control holder is correspondingly arranged on coding.
As shown in Figure 2 and Figure 3, the present invention also provides a kind of forest fire protection method for early warning based on machine vision, including such as
Lower step:
S1, video image obtain: controlling the video image that collection terminal obtains forest zone scene by monitor supervision platform.
S2, target prospect detection: first detecting the step S1 video image sent by foreground detection unit, will
Present frame picture in video image is set as foreground image, while the former frame picture of present frame is set as background image;Then
Background model is constructed, whether occurs the thing of movement in video image of the obtained background model to differentiate input by constructing
Object updates background image if the pixel of foreground image and background image changes without frame difference;Conversely, if foreground image and back
The pixel of scape image is considered as in video when generating the variation of frame difference moving object, and foreground image is generated target image.
S3, flame characteristic extract: after the target image that step S2 is obtained carries out image procossing, then by target image by
Rgb space image is converted to HSI color identification model, then proceeds as follows:
S31, tri- channels R, G, B in RGB are separated respectively, sets the saturation threshold values of red component R in RGB image
For RT, and R > RT;
S32, according to R > R in step S1TCondition judges the channel R;
S33, the saturation degree component value S that target image is calculated based on flame pixels index (1-1) and formula (1-2), are led to
Degree of supersaturation component value S determines whether the location of pixels is flame pixels point, is, the image slices vegetarian refreshments be shown as white and
It extracts feature and carries out position mark, it is on the contrary then be shown as black;
R > G > B (1-1)
S≥((255-R)*ST/RT (1-2)
R is the saturation value of red component in RGB, R in formulaTIt is the threshold value of red component R, S is in HSI color identification model
Saturation degree component value, STIt is the saturation degree threshold value of S.Above-mentioned steps are executed, due to only having used the S in HSI points in formula (1-2)
Amount, so without applying color model transfer function, directly calculating S component, so that calculation amount and computational complexity drop
It is low.
Wherein, pixel value (0~255) the statistics appearance in tonal range to whole image using image pixel histogram
The frequency, red line represent the frequency of R pixel appearance, and green represents G pixel, and what blue represented is B pixel, thus
It can estimate the threshold size in the channel R.
S4, flame locating and tracking: it identifies that there are flames in video image by step S2, and is constantly extracted by step S3
Behind the home position of obtained flame characteristic and corresponding flame characteristic, using the API in opencv visual pattern processing module
Flame contours are drawn out, and call the locating query algorithm for drawing rectangle frame in opencv visual pattern processing module in video
Flame region in image carries out the locating and tracking of pixel, while being marked conflagration area with rectangle frame;
S5, alarm: the flame locating and tracking result of step S4 is sent to alarm module.
Further, in step S2, the background model is mixture Gaussian background model.In mixture Gaussian background model
In, it is believed that the colouring information between pixel is irrelevant, is all independent from each other to the processing of each pixel.For video image
Each of pixel, variation of the value in sequence image be considered as constantly generate pixel value random process.For
Each pixel of multimodal Gaussian distribution model, image is modeled by the superposition of multiple Gaussian Profiles of different weights, and every kind
Corresponding one of Gaussian Profile there may be the state of the presented color of pixel, the weight and distribution parameter of each Gaussian Profile with
Time updates.When handling color image, it is assumed that tri- chrominance channels image slices vegetarian refreshments R, G, B are mutually indepedent and side having the same
Difference.
Then for the observation data set { x of stochastic variable X1,x2,…,xn, xt=(Rt, Gt, Bt) is the sample of t moment pixel
This, then its Gaussian mixtures probability density function obeyed of single sampled point Xt is expressed as P (Xt), i.e., the described mixed Gaussian back
The modeling algorithm of scape model is as follows:
In formula, k is distribution pattern sum, ε (xt, μI, t, τI, t) it is i-th of Gaussian Profile of t moment, μI, tFor its mean value,
τI, tFor its covariance matrix, δI, tFor variance, I is three-dimensional unit matrix, WI, tFor the weight of t moment i-th Gaussian Profile.Institute
Stating e is the natural number truth of a matter, and e specifically takes 2.72 in the present embodiment.
Formula (2-3) is substituted into formula (2-2), the modeling algorithm of the mixture Gaussian background model
Further formula (2-2) is substituted into, is obtained
I.e. by the weighted value W of known t moment i-th Gaussian ProfileI, t, mean μI, t, covariance matrix value τI, t, three-dimensional
Unit matrix value I, distribution pattern sum k are substituted into above-mentioned formula (I) respectively, and it is high to obtain the mixing that single sampled point Xt is obeyed
This distribution probability density function P (Xt)。
It is further preferred that the mixture Gaussian background model differentiates in the video image inputted the thing moved whether occur
The method of object is as follows:
S21, each new pixel value Xt are pressed with current k model | Xt- εI, t=1|≤2.5σI, t=1Condition formula is calculated
Compare, until there is a new pixel to occur, i.e., pixel deviations guarantee in 2.5 σ;
If S22, the matched mode of institute meet context request, which belongs to background, otherwise belongs to prospect;
S23, each schema weight update as follows, and wherein α is learning rate, for matched mode Mk,t=1, it is no
Then Mk,t=0, then the weight of each mode is normalized:
wK, t=(1- ρ) * wK, t-1+ρ*MK, t (2-4)
S24, the mean μ of non-match pattern and standard deviation sigma are constant, and the parameter of match pattern is updated according to following, by as follows
Formula (2-5), (2-6), (2-7) are updated;
ρ=α * ε (Xt|μk, σk) (2-5)
μt=(1- ρ) * μt-1+ρ*Xt (2-6)
In formula, k is the distribution of distribution pattern sum, ε (xt|μk, σk) indicate the kth moment Gaussian Profile, μtFor t moment
The mean value of pixel, σtFor the variance of t moment pixel, σkIndicate the variance of k moment pixel, μkWhat is indicated is the kth moment
The mean value of pixel, α indicate updating factor, and ρ indicates the related coefficient of X and ρ.
If S25, step S22 are not carried out, pixel will carry out descending arrangement according to weight size.
In S21 step, if the smallest mode of weight is replaced without any pattern match, i.e. the mean value of the mode is
Current pixel value, standard deviation are initial the larger value, and weight is smaller value.
Further, in step S3, described image processing the step of it is as follows:
S311, colored target image is subjected to gray proces, that is, calculates the average value in tri- channels R, G, B in RGB,
Then the average value being calculated is attached to tri- components of R, G, B again respectively.Purpose color image gray processing be in order to
The calculation amount of reduction system, calculation amount is as few as possible when making to handle picture.
S312, the gray scale picture for obtaining step S01 carry out binary conversion treatment, that is, set the threshold value of red component R as RT,
RTValue estimate to obtain by the histogram of image, be greater than RTThe pixel of threshold range is designated as 1, i.e., white, is less than RTThe picture of threshold value
Vegetarian refreshments is designated as 0, is black.After carrying out binaryzation to gray level image, it is left the color of image only black with white two kinds of colors, figure
The profile of picture is more clear, convenient to be further analyzed to the profile of figure, and maintains the primitiveness of image.
S313, Morphological Gradient is carried out to the image that step S02 is handled using opencv visual pattern processing module
After processing, the red component in image made is more obvious, convenient for the subsequent RGB triple channel of flame characteristic extraction module
Separation.The morphological operation of image is to obtain more accurately image to remove noise jamming.Image after morphological operation
Also some environmental factor bring interference, such as wind can be removed.The color profile of image is more clear obviously after being operated, and is
Followed by channel separation reduce calculation amount-effect.
It, can be with the present invention is based on the fire identification rate of the forest fire of machine vision fire prevention early warning system by repetition test
Reach 94%, as shown in table 1, as can be seen from the table, in the case where threshold value is constant, incendiary material either withered grass or wood
Material, system can successfully track and recognize flame.
The case where 1 four selecting video image Flames of table identify table:
Table 1
Above description is the detailed description for the present invention preferably possible embodiments, but embodiment is not limited to this hair
Bright patent claim, it is all the present invention suggested by technical spirit under completed same changes or modifications change, should all belong to
In the covered the scope of the patents of the present invention.
Claims (9)
1. a kind of forest fireproofing early warning system based on machine vision, including collection terminal and monitor supervision platform, the collection terminal distribution
At forest zone scene to be monitored, for obtaining the video image at forest zone scene in real time and being sent to monitor supervision platform, the monitoring
Module is connect platform with collection terminal by wireless communication, it is characterised in that: the monitor supervision platform includes moving target foreground detection
Module, image pre-processing module, flame characteristic extraction module, flame tracking and identification module and alarm module,
The moving target foreground detection module receives the video image that collection terminal is sent, and first by being integrated in movement mesh
Foreground detection unit in mark foreground detection module detects video image: the present frame picture in video image is set as
Foreground image, while the former frame picture of present frame is set as background image;Then construct background model, by background model come
Differentiate whether occur the things of movement in the video image of input: if the pixel of foreground image and background image becomes without frame difference
When change, then background image is updated;Conversely, if the pixel of foreground image and background image is considered as in video when generating the variation of frame difference
There is moving object, and foreground image generation target image is sent to flame characteristic extraction module;
The flame characteristic extraction module first passes through the target that image pre-processing module sends moving target foreground detection module
Image carries out image procossing, target image is then converted to HSI color identification model by rgb space image, by calculating
Differentiate whether the location of pixels is flame pixels point after saturation degree component value S in HSI color identification model, is the then image slices
Vegetarian refreshments is shown as white and extracts feature, on the contrary then be shown as black;
The flame tracking and identification module identify in video image according to HSI color identification model there are after flame, pass through
Flame characteristic extraction module constantly extracts flame characteristic, and uses opencv vision figure according to obtained flame characteristic position is extracted
As the API drafting flame contours in processing module, and locating and tracking lookup is carried out to flame contours, issues fire to alarm module
Pre-warning signal.
2. the forest fireproofing early warning system according to claim 1 based on machine vision, it is characterised in that: further include movement
Terminal, the mobile terminal are connect by communication module with alarm module.
3. the forest fireproofing early warning system according to claim 1 based on machine vision, it is characterised in that: described image is pre-
The input terminal of processing module is connect with moving target foreground detection module, and output end and the flame characteristic of image pre-processing module mention
The connection of modulus block.
4. the forest fireproofing early warning system according to claim 1 based on machine vision, it is characterised in that: the movement mesh
The background model that mark foreground detection module uses is that back is fitted by establishing multiple Gaussian Profiles for mixture Gaussian background model
Scape is to realize the description to multimodal distribution background.
5. the forest fireproofing early warning system according to claim 1 based on machine vision, it is characterised in that: the collection terminal
Including n video camera, n video camera is respectively distributed to each corner in forest zone, and each video camera is respectively connected with a network control
Holder processed, actuated camera is obtained according to control instruction after the network-control holder receives the control instruction that monitor supervision platform is sent
Take the video image of respective angles, the integer that n is >=10.
6. a kind of forest fire protection method for early warning based on machine vision, which comprises the steps of:
S1, video image obtain: controlling the video image that collection terminal obtains forest zone scene by monitor supervision platform;
S2, target prospect detection: the step S1 video image sent is detected by foreground detection unit first, by video
Present frame picture in image is set as foreground image, while the former frame picture of present frame is set as background image;Then it constructs
Whether background model there is the things of movement by constructing in video image of the obtained background model to differentiate input, if
When the pixel of foreground image and background image changes without frame difference, background image is updated;Conversely, if foreground image and background image
Pixel be considered as in video and have moving object when generating the variation of frame difference, and foreground image is generated into target image;
S3, flame characteristic extract: after the target image that step S2 is obtained carries out image procossing, then by target image by RGB sky
Between image be converted to HSI color identification model, then proceed as follows:
S31, tri- channels R, G, B in RGB are separated respectively, sets the saturation threshold values of red component R in RGB image as RT,
And R > RT;
S32, according to R > R in step S1TCondition judges the channel R;
S33, the saturation degree component value S that target image is calculated based on flame pixels index (1-1) and formula (1-2), by full
Determine whether the location of pixels is flame pixels point with degree component value S, be, which is shown as white and extracts
Feature carries out position mark, on the contrary then be shown as black;
R > G > B (1-1)
S≥((255-R)*ST/RT (1-2)
R is the saturation value of red component in RGB, R in formulaTIt is the threshold value of red component R, S is full in HSI color identification model
With degree component value, STIt is the saturation degree threshold value of S;
S4, flame locating and tracking: it identifies that there are flames in video image by step S2, and is constantly extracted and obtained by step S3
Flame characteristic and corresponding flame characteristic home position after, drawn using the API in opencv visual pattern processing module
Flame contours out, and call the locating query algorithm for drawing rectangle frame in opencv visual pattern processing module in video image
In flame region carry out the locating and tracking of pixel, while conflagration area being marked with rectangle frame;
S5, alarm: the flame locating and tracking result of step S4 is sent to alarm module.
7. the forest fire protection method for early warning according to claim 6 based on machine vision, it is characterised in that: in step S2,
The background model is mixture Gaussian background model, and the modeling algorithm of the mixture Gaussian background model is as follows:
In formula, k is distribution pattern sum, ε (xt, μI, t, τI, t) it is i-th of Gaussian Profile of t moment, μI, tFor its mean value, τI, tFor it
Covariance matrix, δI, tFor variance, I is three-dimensional unit matrix, WI, tFor the weight of t moment i-th Gaussian Profile.
8. the forest fire protection method for early warning based on machine vision according to claim 7, which is characterized in that described
Mixture Gaussian background model differentiates that the method for whether occurring the things moved in the video image inputted is as follows:
S21, each new pixel value Xt are pressed with current k model | Xt- εI, t=1|≤2.5σI, t=1Condition formula is calculated and compared,
Until there is a new pixel to occur, i.e., pixel deviations guarantee in 2.5 σ;
If S22, the matched mode of institute meet context request, which belongs to background, otherwise belongs to prospect;
S23, each schema weight update as follows, and wherein α is learning rate, for matched mode MK, t=1, otherwise MK, t
=0, then the weight of each mode is normalized:
wK, t=(1- ρ) * wK, t-1+ρ*MK, t (2-4)
S24, the mean μ of non-match pattern and standard deviation sigma are constant, and the parameter of match pattern is updated according to following, as follows
(2-5), (2-6), (2-7) are updated;
ρ=α * ε (Xt|μk, σk) (2-5)
μt=(1- ρ) * μt-1+ρ*Xt (2-6)
In formula, k is the distribution of distribution pattern sum, ε (xt|μk, σk) indicate the kth moment Gaussian Profile, μtFor t moment pixel
Mean value, σtFor the variance of t moment pixel, σkIndicate the variance of k moment pixel, μkWhat is indicated is kth moment pixel
Mean value, α indicate updating factor, ρ indicate X and ρ related coefficient;
If S25, step S22 are not carried out, pixel will carry out descending arrangement according to weight size.
9. the forest fire protection method for early warning based on machine vision according to claim 6, which is characterized in that step
In S3, described image processing the step of it is as follows:
S311, colored target image is subjected to gray proces, that is, calculates the average value in tri- channels R, G, B in RGB, then
The average value being calculated is attached to tri- components of R, G, B again respectively;
S312, the gray scale picture for obtaining step S01 carry out binary conversion treatment, that is, set the threshold value of red component R as RT, RT's
Value is estimated to obtain by the histogram of image, is greater than RTThe pixel of threshold range is designated as 1, i.e., white, is less than RTThe pixel of threshold value
It is designated as 0, is black;
S313, Morphological Gradient processing is carried out to the image that step S02 is handled using opencv visual pattern processing module
Afterwards, the red component in image made is more obvious, convenient for point of the subsequent RGB triple channel of flame characteristic extraction module
From.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811488530.XA CN109377703A (en) | 2018-12-06 | 2018-12-06 | A kind of forest fireproofing early warning system and its method based on machine vision |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811488530.XA CN109377703A (en) | 2018-12-06 | 2018-12-06 | A kind of forest fireproofing early warning system and its method based on machine vision |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109377703A true CN109377703A (en) | 2019-02-22 |
Family
ID=65376213
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811488530.XA Pending CN109377703A (en) | 2018-12-06 | 2018-12-06 | A kind of forest fireproofing early warning system and its method based on machine vision |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109377703A (en) |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109886231A (en) * | 2019-02-28 | 2019-06-14 | 重庆科技学院 | A kind of garbage burning factory Combustion Flame Recognition Using method |
CN110119730A (en) * | 2019-06-03 | 2019-08-13 | 齐鲁工业大学 | A kind of monitor video processing method, system, terminal and storage medium |
CN110517435A (en) * | 2019-09-08 | 2019-11-29 | 天津大学 | The portable instant fire prevention early warning of one kind and Information Collecting & Processing early warning system and method |
CN110991361A (en) * | 2019-12-06 | 2020-04-10 | 衢州学院 | Multi-channel multi-modal background modeling method for high-definition high-speed video |
CN111047818A (en) * | 2019-11-01 | 2020-04-21 | 浙江省林业技术推广总站(浙江省林业信息宣传中心) | Forest fire early warning system based on video image |
CN111274896A (en) * | 2020-01-15 | 2020-06-12 | 深圳市守行智能科技有限公司 | Smoke and fire identification algorithm |
CN111666834A (en) * | 2020-05-20 | 2020-09-15 | 哈尔滨理工大学 | Forest fire automatic monitoring and recognizing system and method based on image recognition technology |
CN111830924A (en) * | 2020-08-04 | 2020-10-27 | 郑州信大先进技术研究院 | Unified management and linkage control system and method for internal facilities of building engineering |
CN112312081A (en) * | 2020-09-08 | 2021-02-02 | 深圳中核普达测量科技有限公司 | Fire scene intelligent monitoring method and system |
CN112556655A (en) * | 2020-12-09 | 2021-03-26 | 武汉云图互联科技股份有限公司 | Forestry fire prevention monocular positioning method and system |
CN112966668A (en) * | 2021-04-06 | 2021-06-15 | 中交三公局第一工程有限公司 | Intelligent fire-fighting early warning system |
CN113435373A (en) * | 2021-07-05 | 2021-09-24 | 西安科技大学 | Mine fire video image intelligent recognition device and method |
CN113570802A (en) * | 2021-06-25 | 2021-10-29 | 浙江大华技术股份有限公司 | Camera warning method, warning device and computer readable storage medium |
CN114152347A (en) * | 2021-09-30 | 2022-03-08 | 国网黑龙江省电力有限公司电力科学研究院 | Transformer substation power equipment fault positioning and fire research and judgment comprehensive detection method |
WO2022121060A1 (en) * | 2020-12-12 | 2022-06-16 | 浙江工业大学之江学院 | Visual perception three-dimensional reconstruction technology-based fire intelligent early warning system and method |
CN114882447A (en) * | 2022-07-12 | 2022-08-09 | 南通森田消防装备有限公司 | Fire-proof rolling door real-time early warning method and system based on visual perception |
CN114885119A (en) * | 2022-03-29 | 2022-08-09 | 西北大学 | Intelligent monitoring alarm system and method based on computer vision |
CN115083096A (en) * | 2022-05-05 | 2022-09-20 | 上海电机学院 | Fire early warning and positioning system based on multi-sensor information fusion |
CN116630843A (en) * | 2023-04-13 | 2023-08-22 | 安徽中科数智信息科技有限公司 | Fire prevention supervision and management method and system for fire rescue |
CN117523499A (en) * | 2023-12-29 | 2024-02-06 | 广东邦盛北斗科技股份公司 | Forest fire prevention monitoring method and system based on Beidou positioning and sensing |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120195462A1 (en) * | 2011-01-27 | 2012-08-02 | Chang Jung Christian University | Flame identification method and device using image analyses in hsi color space |
CN103617414A (en) * | 2013-11-09 | 2014-03-05 | 中国科学技术大学 | Fire disaster color model and fire disaster flame and smog identification method based on maximum margin criterion |
CN104794486A (en) * | 2015-04-10 | 2015-07-22 | 电子科技大学 | Video smoke detecting method based on multi-feature fusion |
CN106845443A (en) * | 2017-02-15 | 2017-06-13 | 福建船政交通职业学院 | Video flame detecting method based on multi-feature fusion |
-
2018
- 2018-12-06 CN CN201811488530.XA patent/CN109377703A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120195462A1 (en) * | 2011-01-27 | 2012-08-02 | Chang Jung Christian University | Flame identification method and device using image analyses in hsi color space |
CN103617414A (en) * | 2013-11-09 | 2014-03-05 | 中国科学技术大学 | Fire disaster color model and fire disaster flame and smog identification method based on maximum margin criterion |
CN104794486A (en) * | 2015-04-10 | 2015-07-22 | 电子科技大学 | Video smoke detecting method based on multi-feature fusion |
CN106845443A (en) * | 2017-02-15 | 2017-06-13 | 福建船政交通职业学院 | Video flame detecting method based on multi-feature fusion |
Non-Patent Citations (3)
Title |
---|
彭文健: "《智能视频***中的火焰检测算法研究》", 《电子科技大学》 * |
郭伟: "《一款用于森林防火的小型无人机设计》", 《南京航空航天大学》 * |
雷帮军: "《视频目标跟踪***分步详解》", December 2015, 国防工业出版社 * |
Cited By (29)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109886231A (en) * | 2019-02-28 | 2019-06-14 | 重庆科技学院 | A kind of garbage burning factory Combustion Flame Recognition Using method |
CN110119730A (en) * | 2019-06-03 | 2019-08-13 | 齐鲁工业大学 | A kind of monitor video processing method, system, terminal and storage medium |
CN110517435A (en) * | 2019-09-08 | 2019-11-29 | 天津大学 | The portable instant fire prevention early warning of one kind and Information Collecting & Processing early warning system and method |
CN111047818A (en) * | 2019-11-01 | 2020-04-21 | 浙江省林业技术推广总站(浙江省林业信息宣传中心) | Forest fire early warning system based on video image |
CN110991361B (en) * | 2019-12-06 | 2021-01-15 | 衢州学院 | Multi-channel multi-modal background modeling method for high-definition high-speed video |
CN110991361A (en) * | 2019-12-06 | 2020-04-10 | 衢州学院 | Multi-channel multi-modal background modeling method for high-definition high-speed video |
CN111274896A (en) * | 2020-01-15 | 2020-06-12 | 深圳市守行智能科技有限公司 | Smoke and fire identification algorithm |
CN111274896B (en) * | 2020-01-15 | 2023-09-26 | 深圳市守行智能科技有限公司 | Smoke and fire recognition algorithm |
CN111666834A (en) * | 2020-05-20 | 2020-09-15 | 哈尔滨理工大学 | Forest fire automatic monitoring and recognizing system and method based on image recognition technology |
CN111830924A (en) * | 2020-08-04 | 2020-10-27 | 郑州信大先进技术研究院 | Unified management and linkage control system and method for internal facilities of building engineering |
CN111830924B (en) * | 2020-08-04 | 2021-06-11 | 郑州信大先进技术研究院 | Unified management and linkage control system and method for internal facilities of building engineering |
CN112312081A (en) * | 2020-09-08 | 2021-02-02 | 深圳中核普达测量科技有限公司 | Fire scene intelligent monitoring method and system |
CN112312081B (en) * | 2020-09-08 | 2023-04-07 | 深圳中核普达测量科技有限公司 | Fire scene intelligent monitoring method and system |
CN112556655A (en) * | 2020-12-09 | 2021-03-26 | 武汉云图互联科技股份有限公司 | Forestry fire prevention monocular positioning method and system |
CN112556655B (en) * | 2020-12-09 | 2022-04-26 | 武汉云图互联科技股份有限公司 | Forestry fire prevention monocular positioning method and system |
WO2022121060A1 (en) * | 2020-12-12 | 2022-06-16 | 浙江工业大学之江学院 | Visual perception three-dimensional reconstruction technology-based fire intelligent early warning system and method |
CN112966668A (en) * | 2021-04-06 | 2021-06-15 | 中交三公局第一工程有限公司 | Intelligent fire-fighting early warning system |
CN113570802A (en) * | 2021-06-25 | 2021-10-29 | 浙江大华技术股份有限公司 | Camera warning method, warning device and computer readable storage medium |
CN113570802B (en) * | 2021-06-25 | 2022-12-23 | 浙江大华技术股份有限公司 | Camera warning method, warning device and computer readable storage medium |
CN113435373A (en) * | 2021-07-05 | 2021-09-24 | 西安科技大学 | Mine fire video image intelligent recognition device and method |
CN114152347A (en) * | 2021-09-30 | 2022-03-08 | 国网黑龙江省电力有限公司电力科学研究院 | Transformer substation power equipment fault positioning and fire research and judgment comprehensive detection method |
CN114885119A (en) * | 2022-03-29 | 2022-08-09 | 西北大学 | Intelligent monitoring alarm system and method based on computer vision |
CN115083096A (en) * | 2022-05-05 | 2022-09-20 | 上海电机学院 | Fire early warning and positioning system based on multi-sensor information fusion |
CN114882447A (en) * | 2022-07-12 | 2022-08-09 | 南通森田消防装备有限公司 | Fire-proof rolling door real-time early warning method and system based on visual perception |
CN114882447B (en) * | 2022-07-12 | 2022-09-20 | 南通森田消防装备有限公司 | Fire-proof rolling door real-time early warning method and system based on visual perception |
CN116630843A (en) * | 2023-04-13 | 2023-08-22 | 安徽中科数智信息科技有限公司 | Fire prevention supervision and management method and system for fire rescue |
CN116630843B (en) * | 2023-04-13 | 2024-05-17 | 安徽中科数智信息科技有限公司 | Fire prevention supervision and management method and system for fire rescue |
CN117523499A (en) * | 2023-12-29 | 2024-02-06 | 广东邦盛北斗科技股份公司 | Forest fire prevention monitoring method and system based on Beidou positioning and sensing |
CN117523499B (en) * | 2023-12-29 | 2024-03-26 | 广东邦盛北斗科技股份公司 | Forest fire prevention monitoring method and system based on Beidou positioning and sensing |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109377703A (en) | A kind of forest fireproofing early warning system and its method based on machine vision | |
CN109271554B (en) | Intelligent video identification system and application thereof | |
EP3869459B1 (en) | Target object identification method and apparatus, storage medium and electronic apparatus | |
CN106600888B (en) | Automatic forest fire detection method and system | |
CN109299683B (en) | Security protection evaluation system based on face recognition and behavior big data | |
CN103208126B (en) | Moving object monitoring method under a kind of physical environment | |
CN108062349A (en) | Video frequency monitoring method and system based on video structural data and deep learning | |
CN108053427A (en) | A kind of modified multi-object tracking method, system and device based on KCF and Kalman | |
CN110502965A (en) | A kind of construction safety helmet wearing monitoring method based on the estimation of computer vision human body attitude | |
CN113516076B (en) | Attention mechanism improvement-based lightweight YOLO v4 safety protection detection method | |
CN108009473A (en) | Based on goal behavior attribute video structural processing method, system and storage device | |
CN109558840A (en) | A kind of biopsy method of Fusion Features | |
CN108229458A (en) | A kind of intelligent flame recognition methods based on motion detection and multi-feature extraction | |
CN106131376B (en) | A kind of indoor and outdoor scene determines method and device | |
CN109145689A (en) | A kind of robot fire detection method | |
CN110070530A (en) | A kind of powerline ice-covering detection method based on deep neural network | |
CN108921039A (en) | The forest fire detection method of depth convolution model based on more size convolution kernels | |
CN110378865A (en) | A kind of greasy weather visibility intelligence hierarchical identification method and system under complex background | |
CN113903081A (en) | Visual identification artificial intelligence alarm method and device for images of hydraulic power plant | |
CN109118548A (en) | A kind of comprehensive intelligent water quality recognition methods | |
CN106339657B (en) | Crop straw burning monitoring method based on monitor video, device | |
CN111401310B (en) | Kitchen sanitation safety supervision and management method based on artificial intelligence | |
CN109241847A (en) | The Oilfield Operation District safety monitoring system of view-based access control model image | |
CN110135476A (en) | A kind of detection method of personal safety equipment, device, equipment and system | |
CN109935080A (en) | The monitoring system and method that a kind of vehicle flowrate on traffic route calculates in real time |
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 |
Application publication date: 20190222 |
|
RJ01 | Rejection of invention patent application after publication |