CN105844295B - A kind of video smoke sophisticated category method based on color model and motion feature - Google Patents
A kind of video smoke sophisticated category method based on color model and motion feature Download PDFInfo
- Publication number
- CN105844295B CN105844295B CN201610162064.0A CN201610162064A CN105844295B CN 105844295 B CN105844295 B CN 105844295B CN 201610162064 A CN201610162064 A CN 201610162064A CN 105844295 B CN105844295 B CN 105844295B
- Authority
- CN
- China
- Prior art keywords
- smoke
- feature
- region
- image block
- suspicious
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The video smoke sophisticated category method based on color model and motion feature that the present invention provides a kind of, this method comprises: based on gauss hybrid models and background difference and color model is constructed, detection movement pixel and progress preliminary screening；Moving target connected region is obtained, motion change feature is extracted, threshold value setting is carried out and discriminates whether as suspicious smoke region；Suspicious smoke region is pre-processed, SIFT feature is extracted based on image block；Optimize random forests algorithm in conjunction with SVM, be trained based on suspicious smoke region image block, to realize the sophisticated category to smog such as fire hazard aerosol fog, smoke from cigarette, aqueous vapors.The detection that video smoke is carried out the present invention is based on color model and motion feature, towards sophisticated category, realizes real-time quick Smoke Detection, effectively eliminates the influence for having similar effect interference source with fire hazard aerosol fog, improve the efficiency and accuracy rate of detection.
Description
Technical field
The present invention relates to Computer Image Processing, intelligent recognition field, more particularly to one kind is based on color model and movement
The video smoke sophisticated category method of feature.
Background technique
In recent years, it is popularized with laying special stress on protecting unit with video monitoring system in city and fire-fighting, artificial intelligence and mode
Identification technology constantly develops, and the fire detection method based on video is more and more taken seriously.Video fire hazard detection method has
Efficiently, in real time, intelligence, low cost, the advantages that facilitating, realize fire detection in real time, for detect earlier fire to
Bigger loss is avoided to have very positive meaning.
Video fire hazard detection method mainly includes flame detecting method and smog detection method.Smoke detection method is to visit
The smog that fire detecting calamity generates detects fire, judges whether there are also smoke regions in video image.General fire early period of origination
Experience is glowed the stage, and smoldering in this stage is the main characterization of fire.Obviously, this method can be found that the rank of glowing of fire
Section, can detect fire earlier, can more reduce the loss of people's property, therefore the detection of the video fire hazard based on smog
Technology is more able to satisfy the demand in market, also becomes research hotspot instantly.
In the implementation of the present invention, the inventor finds that the existing technology has at least the following problems: although existing at present
The correlative study of many video smoke detections, and certain achievement is achieved, but influenced by technical maturity and cost, it is based on
The video fire hazard detection technique of smog can't be used widely.Existing algorithm has similar effect to do to fire hazard aerosol fog
Source (such as smoke from cigarette, mosquito-repellent incense smog, water vapour) is disturbed, algorithm there is no to may be implemented preferably to exclude effect at present, and it is this
Interference source cannot exclude, that just fundamentally can not further promote accuracy rate and reduce false alert rate；The more complicated ring of background
Border is extracted and is detected to smoke region, is affected, existing algorithm effect when handling this background is poor；It is more existing to have
The algorithm required time complexity of preferable detection performance can not fully meet the real time handling requirement of monitoring system, thus
Lose the most important meaning of video smoke detection technique.
Summary of the invention
In view of the above problems, propose the present invention in order to provide one kind overcome the above problem or at least be partially solved or
Person slows down the video smoke sophisticated category method based on color model and motion feature of the above problem.
A kind of the technical solution adopted by the present invention are as follows: video smoke sophisticated category side based on color model and motion feature
Method, the method steps are as follows:
Step (1) obtains monitoring video flow in real time, detects the movement pixel in current video frame, and construct color mould
Type carries out preliminary screening to movement pixel；
Step (2) obtains connection moving region, and calculates its kinematic parameter, and differentiate to obtain suspicious smog according to threshold value
Region；
Step (3) generates image block after carrying out image preprocessing to suspicious smoke region, extracts SIFT feature, then foundation
The vision code book given birth to is mapped as histogram, obtains the feature vector of image block；
Step (4) carries out feature training based on image block, in conjunction with svm classifier algorithm to the node point of random forests algorithm
The process of splitting optimizes, the ballot parallelization to decision tree, to realize the sophisticated category of all kinds of smog.
Wherein, the step (1) further comprises:
Step (a) is directed to video frame, based on gauss hybrid models and background difference algorithm, carries out moving target respectively
It extracts, result is merged, to cope with different environment；
Step (b), building color model, carry out principium identification to movement pixel and retain suspicious smog pixel.
The step (b) further comprises:
The translucence of smog makes smog its color and background object color when more thin very nearly the same, therefore first
The comparison for first carrying out foreground and background colour to the moving region detected differentiates: for pixel x (i, j), calculating foreground picture
With the difference of Background respective channel, diff is obtainedR、diffG、diffBIf meeting:
diffR± α=diffG± α=diffB±α
Then think that the pixel in foreground picture is suspicious smog pixel, otherwise continue to differentiate according to the following steps: not by
If background object color is interfered, smog color is generally canescence, and under rgb space, triple channel value is very close, and big
In a certain threshold value, even meet:
R ± β=G ± β=B ± β, R (G, B) > γ
Then think that the pixel is suspicious smog pixel, otherwise excludes.
Wherein, the step (2) further comprises:
Step (a) takes a series of region contour processing methods, obtains connection moving region；
Step (b), the connection moving region obtained according to processing, calculate its kinematic parameter, and differentiated according to threshold value,
Obtain suspicious smoke region.
The step (b) further comprises:
It selects motion target area area change rate as a kinematic parameter: calculating one-shot change rate every N frame sampling, that
In t0When, the smog area change rate sampled between adjacent two field pictures may be expressed as:
Select motion target area geometry mass center change rate as a kinematic parameter: in t0When, vertical direction mass center becomes
The average value of rate may be expressed as:
Wherein, the step (3) further comprises:
Step (a), vision code book generation phase pass through the pretreatment to training image collection, extraction SIFT feature, Jin Erju
Class generates vision code book；
Step (b), the present frame suspicious smoke region character representation stage, by the way that image block is pre-processed, generated to it, is extracted
SIFT feature obtains the map histograms of image block according to vision code book, then obtain the corresponding feature in suspicious smoke region to
Amount.
The step (b) further comprises:
Remember that suspicious smoke region generates P image block after pretreatment, carries out SIFT feature extraction, obtain image block
SIFT feature vector, calculate Euclidean under each of image block SIFT feature to its vision code book between vision word away from
From, and be mapped as apart from nearest vision word, i.e., by the correspondence word frequency+1 of the vision word, after completing this step, often
One image block just maps for a histogram corresponding with vision word sequence, the final suspicious smoke region of testing image
As soon as each image block be represented as the feature vector of K dimension, then this image has P K dimensional feature vector.
Wherein, the step (4) further comprises:
Step (a) selects strong two-value classifier SVM in the fission process of node, with this classification for enhancing decision tree
Reliability；
Step (b), the prediction classification speed in order to improve random forests algorithm, the present invention are pre- in each decision tree of random forest
It surveys during classification results, using the implementation method of multi-threaded parallel.
The step (a) further comprises:
Traditional random forests algorithm during generating decision tree, for decision tree nodes division using weak
Classifier, direct given threshold, by comparing the characteristic value of current selected feature and the size of threshold value, by the sample in present node
Originally it is divided into left and right child node.The present invention intends selecting strong two-value classifier SVM in the fission process of node, enhances decision with this
The reliability of the classification of tree.Each intermediate node in decision tree is trained by SVM classifier, what is obtained in this way determines
Plan tree is strongr.
The major advantage of the present invention compared with the conventional method is:
A kind of video smoke sophisticated category method based on color model and motion feature provided by the invention, passes through optimization
Moving object detection algorithm constructs suitable smog color model, in conjunction with more careful regional processing method, obtains movement and connects
Motion change feature is simultaneously extracted in logical region, and merges random forest and SVM progress sophisticated category, realizes real-time quick smog inspection
It surveys, effectively eliminates the influence for having similar effect interference source with fire hazard aerosol fog, improve the efficiency and accuracy rate of detection.
The above description is only an overview of the technical scheme of the present invention, in order to better understand the technical means of the present invention,
And it can be implemented in accordance with the contents of the specification, and in order to allow above and other objects of the present invention, feature and advantage can
It is clearer and more comprehensible, the followings are specific embodiments of the present invention.
Detailed description of the invention
By reading the following detailed description of the preferred embodiment, various other advantages and benefits are common for this field
Technical staff will become clear.The drawings are only for the purpose of illustrating a preferred embodiment, and is not considered as to the present invention
Limitation.In the accompanying drawings:
Fig. 1 shows basic flow chart of the invention；
Fig. 2 shows the flow charts of smoke region detection algorithm of the invention；
Fig. 3 shows the flow chart of image feature representation of the invention.
Specific embodiment
Below with reference to the accompanying drawings, the embodiment of the present invention is described in detail.
Method overall process of the invention is illustrated first.
Fig. 1 shows the exemplary basic flow chart of the present invention, is therefrom expressly understood that four committed steps of the invention:
(1) monitoring video flow is obtained in real time, detects the movement pixel in current video frame, and constructs color model and pixel is carried out
Preliminary screening；(2) connection moving region is obtained, and calculates its kinematic parameter, and differentiate to obtain suspicious smoke region according to threshold value；
(3) image block is generated after carrying out image preprocessing to suspicious smoke region, extracts SIFT feature, then according to the vision given birth to
Code book is mapped as histogram, obtains the feature vector of image block；(4) feature training is carried out based on image block, in conjunction with SVM points
Class algorithm optimizes the node split process of random forests algorithm, the ballot parallelization to decision tree, to realize all kinds of
The sophisticated category of smog.
Fig. 2 shows the processes of movement the pixel detection algorithm and suspicious smoke region detection algorithm of the embodiment of the present invention
Figure.
As can be seen from the figure the first step for moving pixel detection algorithm is exactly: mixed based on Gauss for video frame
Molding type and background difference algorithm carry out the extraction of moving target respectively, merge to result, to cope with different environment.
Background modeling is carried out to current image frame using existing gauss hybrid models, test parameters chooses suitable parameter
To extract movement pixel；The recognition effect of more slow object is fixed and moved to moving direction in view of gauss hybrid models
Poor, the present invention uses background subtraction also to extract movement pixel.Above two moving object detection algorithm is actually being answered
There is respectively suitable environmental condition in, is two in conjunction with the moving parameter information of subsequent acquisition, and monitoring shooting time point
Different weights are arranged in the result that kind detection method obtains.
In this step, it is a more crucial step that background frames, which update, and static background frame is selected in conjunction with frame differential method,
If the pixel number in the difference image of continuous N number of consecutive frame is less than specified threshold, that selected present frame is updated to background frames.Prospect
Background pixel point determines: the optimal threshold of present frame is generated using otsu maximum variance between clusters, greater than the pixel of the threshold value
It is confirmed as foreground pixel point.
Fig. 2 shows movement pixel detection algorithms followed by building color model, to moving object detection algorithm
The movement pixel detected is differentiated, suspicious smog pixel is retained.
It is found in investigation, indoor smog color is generally grey white, and it is reasonable to need to construct when detecting smog pixel
Smog color model.The translucence of smog makes smog its color when more thin with background object color differ nothing
It is several, therefore the comparison for carrying out foreground and background colour to the moving region detected first differentiates:
For pixel x (i, j), the difference of foreground picture and Background respective channel is calculated, diff is obtainedR、diffG、
diffBIf meeting:
diffR± α=diffG± α=diffB±α
Then think that the pixel in foreground picture is suspicious smog pixel, otherwise continues to differentiate according to (b) step.Not by
If background object color is interfered, smog color is generally canescence, and under rgb space, triple channel value is very close, and big
In a certain threshold value, even meet:
R ± β=G ± β=B ± β
R (G, B) > γ
Then think that the pixel is suspicious smog pixel, otherwise excludes.
It whether is being that suspicious smog pixel carries out just to movement pixel by color model according to shown by Fig. 2
After step judgement, into suspicious smoke region detection algorithm process, connection moving region is obtained.
A series of processing are carried out to suspicious smog pixel point set to obtain motion communication region, include intermediate value in these processing
Filtering, Morphological scale-space, profile processing etc..Noise is removed using median filtering, using morphologic opening and closing operation by phase
Neighbouring region merges, and on this basis, is handled using profile each region entirety, not only removes those profiles and is less than regulation
The connected region of threshold value, and those satisfactory profile convex closures or polynomial fitting curve are replaced, to extract
Complete motion communication region.
The second step of suspicious smoke region detection algorithm is in Fig. 2 example: the connection motor area obtained according to processing
Domain calculates its kinematic parameter, and is differentiated according to threshold value, obtains suspicious smoke region, provides for next sophisticated category
More efficiently data.
Select motion target area area change rate as a kinematic parameter: target area area can pass through statistic mixed-state
The bianry image pixel number obtained afterwards is calculated.The problem of calculation amount is considered in actual treatment, every N frame sampling meter
One-shot change rate is calculated, in addition remembers that the area for obtaining target area from present frame is St(t=1,2 ..., i), then in t0When, sampling
Smog area change rate between adjacent two field pictures may be expressed as:
IfGreater than its given threshold, then can add up into statistic, as a factor for being determined as smog.When tired
Adduction is more than after its given threshold, it is believed that the frame sequence meets this motion feature of region area change rate.
Select motion target area geometry mass center change rate as a kinematic parameter: t frame motion target area is corresponding
Centroid position (xo, yo) calculate such as formula:
DR (t) indicates the motion target area of t frame in formula.Then in t0When, the average value of vertical direction mass center change rate
It may be expressed as:
H is picture frame vertical height in formula, ifGreater than its given threshold, then it can add up into statistic, as sentencing
Not Wei smog a factor, after the statistic is cumulative is greater than its given threshold, it is believed that the sequence frame meets moving target area
Domain geometry mass center vertical direction change rate.
By the investigation to sophisticated category method, the present invention uses the precise image based on code book (Codebook-based)
Classification mode, every piece image in training data is first mainly divided into many part patch (can be by rigid segmentation
It is also possible to based on SIFT critical point detection), in this way, each image is just indicated by many patch, then to image
Each patch carries out feature extraction, and each feature then passes through the side of visual word packet Bag-of-Words as a vocabulary, image
Method is indicated.Dictionary/code of one group of visual vocabulary comprising characteristics of image is determined further according to the feature of all training images
Sheet (Visual Dictionary or Codebook) is to the distribution situation of its visual vocabulary of the image study of each classification, so
The intermediate semantic of category image is learnt by statistical model or generative probabilistic model afterwards, the last classification according to image is semantic
Constitute the classification for determining image.
The present invention is based on image blocks to carry out feature extraction training, and in the feature of image block, the present invention is used to image
Scaling, target rotation, affine transformation, illumination variation etc. have the SIFT feature of invariance, establish and are based on image block SIFT
The BOW model (Bag of words model) of feature.Such as Fig. 3, for the suspicious smoke region that differentiation obtains, the present invention by its into
The step of row character representation includes two Main Stages: vision code book generation phase and input picture character representation stage.
K- is then carried out to SIFT feature extraction is carried out after training image collection pretreatment in vision code book generation phase
Means cluster, is gathered into the biggish some clusters of quantity, similarity with higher in cluster, and similarity is lower between cluster, wherein often
A cluster is counted as a vision word, which can be used for indicating certain that the deposited key point inside the race has jointly
Local mode.Cluster is divided into K cluster, and cluster centre is also K, and the length of code book is also just K.The step for available training
The vision code book of image set.
In the input picture character representation stage, image preprocessing, the suspicious cigarette being aligned are carried out to suspicious smoke region
Fog-zone generates image block behind domain by the way of even partition again, using consistent fixed mesh, what such each image obtained
Image block all has same size.Remember that suspicious smoke region generates P image block after pretreatment, carry out SIFT feature extraction,
Obtain the SIFT feature vector of image block.Calculate under each of image block SIFT feature to its vision code book vision word it
Between Euclidean distance, and be mapped as apart from nearest vision word, i.e., by the correspondence word frequency+1 of the vision word, complete this
As soon as after step, each image block maps for a histogram corresponding with vision word sequence, and final testing image can
As soon as each image block for doubting smoke region is represented as a feature vector for K dimension, then this image has P K dimensional feature
Vector.
By exemplary step shown in Fig. 3, the present invention is based on code books to have carried out feature extraction expression to suspicious smoke region, with
This provides effective smoke characteristics model for sophisticated category.Next the present invention just uses random forests algorithm algorithm, based on figure
As block carries out feature training.
Traditional random forests algorithm during generating decision tree, for decision tree nodes division using weak
Classifier, direct given threshold, by comparing the characteristic value of current selected feature and the size of threshold value, by the sample in present node
Originally it is divided into left and right child node.The present invention selects strong two-value classifier SVM in the fission process of node, enhances decision tree with this
Classification reliability.Each intermediate node in decision tree, the decision obtained in this way are trained by SVM classifier
It sets strongr.
In addition, for a test sample, traditional random forests algorithm is using serial Voting Model, that is, often
A decision tree classifier give one by one the test sample vote, such as first decision tree classifier to test sample poll closing it
Afterwards, second decision tree classifier starts to vote to the test sample again, and so on.Serial ballot is when comparing consumption
Between, especially when the Numerous of decision tree classifier and test sample huge amount.It is random gloomy in order to improve
The prediction classification speed of woods algorithm is intended during each decision tree prediction classification results of random forest, simultaneously using multithreading herein
The implementation method of rowization.
In conclusion a kind of video smoke sophisticated category side based on color model and motion feature provided by the invention
Method constructs suitable smog color model, in conjunction with more careful regional processing side by optimizing moving object detection algorithm
Method obtains motion communication region and extracts motion change feature, and merges random forest and SVM progress sophisticated category, realizes
Real-time quick Smoke Detection effectively eliminates the influence for having similar effect interference source with fire hazard aerosol fog, improves the effect of detection
Rate and accuracy rate.
Through the above description of the embodiments, those skilled in the art can be apparent from other advantages
And modification.The above embodiments are only used to illustrate the present invention, and not limitation of the present invention, the common skill in relation to technical field
Art personnel can also make a variety of changes and modification without departing from the spirit and scope of the present invention, therefore all etc.
Same technical solution also belongs to scope of the invention, and scope of patent protection of the invention should be defined by the claims.
Claims (5)
1. a kind of video smoke sophisticated category method based on color model and motion feature, it is characterised in that steps are as follows:
Step (1) obtains monitoring video flow in real time, detects the movement pixel in current video frame, and construct color model, right
It moves pixel and carries out preliminary screening；
Step (2) obtains connection moving region, and calculates its kinematic parameter, and differentiate to obtain suspicious smoke region according to threshold value；
Step (3) generates image block after carrying out image preprocessing to suspicious smoke region, extracts SIFT feature, then according to
The vision code book given birth to is mapped as histogram, obtains the feature vector of image block；
The step (3) further comprises:
Step (31), vision code book generation phase by the pretreatment to training image collection, extraction SIFT feature, and then cluster
Generate vision code book；
Step (32), the present frame suspicious smoke region character representation stage, by the way that image block is pre-processed, generated to it, is extracted
SIFT feature obtains the map histograms of image block according to vision code book, then obtain the corresponding feature in suspicious smoke region to
Amount；
Step (4) carries out feature training based on image block, in conjunction with svm classifier algorithm to the node split mistake of random forests algorithm
Journey optimizes, the ballot parallelization to decision tree, to realize the sophisticated category of all kinds of smog；
The step (4) further comprises:
Step (41) selects strong two-value classifier SVM in the fission process of node, with the reliable of this classification for enhancing decision tree
Property；
Step (42), the prediction classification speed in order to improve random forests algorithm, in each decision tree prediction classification knot of random forest
During fruit, using the implementation method of multi-threaded parallel.
2. a kind of video smoke sophisticated category method based on color model and motion feature according to claim 1,
It is characterized in that, the step (1) further comprises:
Step (11) is directed to video frame, based on gauss hybrid models and background difference algorithm, carries out mentioning for moving target respectively
It takes, result is merged, to cope with different environment；
Step (12), building color model, carry out principium identification to movement pixel and retain suspicious smog pixel.
3. a kind of video smoke sophisticated category method based on color model and motion feature according to claim 1,
It is characterized in that, the step (2) further comprises:
Step (21) takes a series of region contour processing methods, obtains connection moving region；
Step (22), the connection moving region obtained according to processing, calculate its kinematic parameter, and differentiated according to threshold value, obtain
To suspicious smoke region.
4. a kind of video smoke sophisticated category method based on color model and motion feature according to claim 1,
It is characterized in that, the step (32) further comprises:
Remember that suspicious smoke region generates P image block after pretreatment, carry out SIFT feature extraction, the SIFT for obtaining image block is special
Vector is levied, calculates the Euclidean distance under each of image block SIFT feature to its vision code book between vision word, and will
It is mapped as apart from nearest vision word, i.e., by the correspondence word frequency+1 of the vision word, after completing this step, each image
Block just maps for a histogram corresponding with vision word sequence, each figure of the final suspicious smoke region of testing image
As soon as block is represented as the feature vector that a K is tieed up, then this image has P K dimensional feature vector.
5. a kind of video smoke sophisticated category method based on color model and motion feature according to claim 1,
It is characterized in that, the step (41) further comprises:
Strong two-value classifier SVM is selected to pass through in the fission process of node with the reliability of this classification for enhancing decision tree
SVM classifier trains each intermediate node in decision tree, and the decision tree obtained in this way is strongr.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610162064.0A CN105844295B (en) | 2016-03-21 | 2016-03-21 | A kind of video smoke sophisticated category method based on color model and motion feature |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610162064.0A CN105844295B (en) | 2016-03-21 | 2016-03-21 | A kind of video smoke sophisticated category method based on color model and motion feature |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105844295A CN105844295A (en) | 2016-08-10 |
| CN105844295B true CN105844295B (en) | 2019-09-27 |
Family
ID=56588151
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610162064.0A Active CN105844295B (en) | 2016-03-21 | 2016-03-21 | A kind of video smoke sophisticated category method based on color model and motion feature |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105844295B (en) |
Families Citing this family (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106446957B (en) * | 2016-10-08 | 2019-02-22 | 常熟理工学院 | A kind of haze image classification method based on random forest |
| CN106355566A (en) * | 2016-10-10 | 2017-01-25 | 浙江树人大学 | Smoke and flame detection method applied to fixed camera dynamic video sequence |
| CN106599874B (en) * | 2016-12-26 | 2019-07-30 | 清华大学苏州汽车研究院（吴江） | A kind of group's mist detection method based on video analysis |
| CN106682635B (en) * | 2016-12-31 | 2019-01-04 | 中国科学技术大学 | A kind of smog detection method based on random forest feature selecting |
| CN107609470B (en) * | 2017-07-31 | 2020-09-01 | 成都信息工程大学 | Method for detecting early smoke of field fire by video |
| CN110120142B (en) * | 2018-02-07 | 2021-12-31 | 中国石油化工股份有限公司 | Fire smoke video intelligent monitoring early warning system and early warning method |
| CN108319964B (en) * | 2018-02-07 | 2021-10-22 | 嘉兴学院 | Fire image recognition method based on mixed features and manifold learning |
| CN108765833A (en) * | 2018-04-27 | 2018-11-06 | 天津大学 | Based on the incipient fire detection algorithm for improving mixed Gaussian and machine learning |
| CN108428324A (en) * | 2018-04-28 | 2018-08-21 | 温州大学激光与光电智能制造研究院 | The detection device of smog in a kind of fire scenario based on convolutional network |
| CN109086647B (en) * | 2018-05-24 | 2022-01-07 | 苏州飞搜科技有限公司 | Smoke detection method and device |
| CN110658118B (en) * | 2018-06-29 | 2023-05-02 | 九阳股份有限公司 | Cooking smoke detection method and smoke machine |
| CN109740639B (en) * | 2018-12-15 | 2021-02-19 | 中国科学院深圳先进技术研究院 | Wind cloud satellite remote sensing image cloud detection method and system and electronic equipment |
| CN110135374A (en) * | 2019-05-21 | 2019-08-16 | 吉林大学 | It is identified using image block characteristics and returns the fire hazard smoke detecting method classified |
| CN110503006A (en) * | 2019-07-29 | 2019-11-26 | 恒大智慧科技有限公司 | A kind of community smoking management-control method, system and its storage medium |
| CN112580396A (en) * | 2019-09-29 | 2021-03-30 | 东北林业大学 | Forest fire recognition method |
| CN112784642B (en) * | 2019-11-08 | 2024-06-21 | 杭州海康威视数字技术股份有限公司 | Vehicle detection method and device |
| CN111553405B (en) * | 2020-04-24 | 2023-08-18 | 青岛杰瑞工控技术有限公司 | Group fog recognition algorithm based on pixel density K-means clustering |
| CN112163121B (en) * | 2020-11-03 | 2021-03-23 | 万得信息技术股份有限公司 | Video content information intelligent analysis processing method based on big data |
| CN113593170B (en) * | 2021-08-11 | 2022-08-30 | 张佳 | Intelligent early warning system based on remote smoke detection |
| CN116416200B (en) * | 2022-01-12 | 2023-10-27 | 中国消防救援学院 | Forest fire smoke detection method based on region matching |
| CN118397475A (en) * | 2024-05-11 | 2024-07-26 | 中睿合和物联网技术(北京)有限公司 | Forest fire intelligent monitoring method based on remote sensing image |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101908141A (en) * | 2010-08-04 | 2010-12-08 | 丁天 | Video smoke detection method based on mixed Gaussian model and morphological characteristics |
| CN102708651A (en) * | 2012-05-23 | 2012-10-03 | 无锡蓝天电子有限公司 | Image type smoke fire disaster detection method and system |
| CN104408745A (en) * | 2014-11-18 | 2015-03-11 | 北京航空航天大学 | Real-time smog scene detection method based on video image |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101237089B1 (en) * | 2011-10-12 | 2013-02-26 | 계명대학교 산학협력단 | Forest smoke detection method using random forest classifier method |
-
2016
- 2016-03-21 CN CN201610162064.0A patent/CN105844295B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101908141A (en) * | 2010-08-04 | 2010-12-08 | 丁天 | Video smoke detection method based on mixed Gaussian model and morphological characteristics |
| CN102708651A (en) * | 2012-05-23 | 2012-10-03 | 无锡蓝天电子有限公司 | Image type smoke fire disaster detection method and system |
| CN104408745A (en) * | 2014-11-18 | 2015-03-11 | 北京航空航天大学 | Real-time smog scene detection method based on video image |
Non-Patent Citations (2)
| Title |
|---|
| Erratum to: Video Fire Smoke Detection Using Motion and Color Features;C Yu 等;《Fire Technology》;20100731;第46卷(第3期);第763-763页 * |
| 基于运动和颜色的视频烟雾检测算法;郑璐 等;《计算机工程与设计》;20101116;第31卷(第21期);第4650-4652页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105844295A (en) | 2016-08-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105844295B (en) | A kind of video smoke sophisticated category method based on color model and motion feature | |
| CN112052797B (en) | MaskRCNN-based video fire disaster identification method and MaskRCNN-based video fire disaster identification system | |
| CN110298297B (en) | Flame identification method and device | |
| CN104008370B (en) | A kind of video face identification method | |
| KR101964397B1 (en) | Information processing apparatus and information processing method | |
| CN108268859A (en) | A kind of facial expression recognizing method based on deep learning | |
| CN109919122A (en) | A kind of timing behavioral value method based on 3D human body key point | |
| CN106934386B (en) | A kind of natural scene character detecting method and system based on from heuristic strategies | |
| CN106228137A (en) | A kind of ATM abnormal human face detection based on key point location | |
| CN103020992B (en) | A kind of video image conspicuousness detection method based on motion color-associations | |
| CN108229458A (en) | A kind of intelligent flame recognition methods based on motion detection and multi-feature extraction | |
| CN109063559A (en) | A kind of pedestrian detection method returned based on improvement region | |
| CN104504362A (en) | Face detection method based on convolutional neural network | |
| CN104992142A (en) | Pedestrian recognition method based on combination of depth learning and property learning | |
| CN105095884B (en) | A kind of pedestrian's identifying system and processing method based on random forest support vector machines | |
| CN110717896A (en) | Plate strip steel surface defect detection method based on saliency label information propagation model | |
| CN105389556B (en) | A kind of high-resolution remote sensing image vehicle checking method for taking shadow region into account | |
| CN105513053B (en) | One kind is used for background modeling method in video analysis | |
| CN102902967A (en) | Method for positioning iris and pupil based on eye structure classification | |
| CN107316036A (en) | A kind of insect recognition methods based on cascade classifier | |
| CN109918971A (en) | Number detection method and device in monitor video | |
| CN103136504A (en) | Face recognition method and device | |
| CN102136059A (en) | Video- analysis-base smoke detecting method | |
| CN103020596A (en) | Method for identifying abnormal human behaviors in power production based on block model | |
| CN113239801B (en) | Cross-domain action recognition method based on multi-scale feature learning and multi-level domain alignment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 100080 c-95, 4th floor, 48 Haidian West Street, Haidian District, Beijing Patentee after: Beijing Zhongke ningkun Technology Co.,Ltd. Address before: 100080 c-95, 4th floor, 48 Haidian West Street, Haidian District, Beijing Patentee before: Beijing Zhongke Jinshu Technology Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210415 Address after: 100080 c-95, 4th floor, 48 Haidian West Street, Haidian District, Beijing Patentee after: Beijing Zhongke Jinshu Technology Co.,Ltd. Address before: 100191 No. 37, Haidian District, Beijing, Xueyuan Road Patentee before: BEIHANG University |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220906 Address after: A14-5, 13th Floor, Building A, Building J1, Phase II, Innovation Industrial Park, No. 2800, Chuangxin Avenue, High-tech Zone, Hefei, China (Anhui) Pilot Free Trade Zone, 230088 Patentee after: Anhui aiguan Vision Technology Co.,Ltd. Address before: 100080 c-95, 4th floor, 48 Haidian West Street, Haidian District, Beijing Patentee before: Beijing Zhongke ningkun Technology Co.,Ltd. |