CN107463968A - Smog judges the detection method of the production method of code book, generation system and smog - Google Patents
Smog judges the detection method of the production method of code book, generation system and smog Download PDFInfo
- Publication number
- CN107463968A CN107463968A CN201710742248.9A CN201710742248A CN107463968A CN 107463968 A CN107463968 A CN 107463968A CN 201710742248 A CN201710742248 A CN 201710742248A CN 107463968 A CN107463968 A CN 107463968A
- Authority
- CN
- China
- Prior art keywords
- image block
- dimensional image
- represent
- smog
- current
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 35
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 20
- 230000003068 static effect Effects 0.000 claims abstract description 71
- 239000000779 smoke Substances 0.000 claims abstract description 31
- 239000011159 matrix material Substances 0.000 claims description 57
- 238000009826 distribution Methods 0.000 claims description 18
- 238000011524 similarity measure Methods 0.000 claims description 18
- 238000012512 characterization method Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 5
- 238000013507 mapping Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 2
- 238000000034 method Methods 0.000 description 16
- 230000014509 gene expression Effects 0.000 description 12
- 238000000926 separation method Methods 0.000 description 9
- 238000004458 analytical method Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 6
- 238000012880 independent component analysis Methods 0.000 description 5
- 238000000354 decomposition reaction Methods 0.000 description 4
- 238000005183 dynamical system Methods 0.000 description 4
- 235000019504 cigarettes Nutrition 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000003595 mist Substances 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 241001123248 Arma Species 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- 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
-
- 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
- G06V10/56—Extraction of image or video features relating to colour
-
- 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
- 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
- G06V10/469—Contour-based spatial representations, e.g. vector-coding
- G06V10/473—Contour-based spatial representations, e.g. vector-coding using gradient analysis
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses the detection method that a kind of smog judges the production method of code book, generation system and smog, and production method includes：Obtain the video of smoke region；Each two field picture of video is divided into by some four-dimensional image blocks using time slip-window, four dimensional features of four-dimensional image block include：The length of four-dimensional image block, width, the static nature of time span and four-dimensional image block；Extract the RGB feature value of each four-dimensional image block；According to the RGB feature value of each four-dimensional image block, based on linear dynamic system framework, high-order dynamic system corresponding to each four-dimensional image block is established；Each high-order dynamic system is clustered, determines that smog judges code book.Because smog provided by the invention judges that code book has taken into full account the RGB feature information of color video, therefore, it is possible to improve the reliability that smog judges code book.Further, judge that code book carries out Smoke Detection using smog provided by the invention, it is possible to increase the reliability and accuracy of detection of Smoke Detection.
Description
Technical field
The present invention relates to detection technique field, production method, the generation system of code book are judged more particularly to a kind of smog
And the detection method of smog.
Background technology
Fire is a kind of sudden natural calamity high, destructive power is strong, accidentally will be to people's with fire in production and life
Lives and properties cause serious threat.Initial Stage of Fire, smog occur prior to flame, and therefore, the smoke detection of high reliability can
For evacuating personnel and fight a fire and strive for more valuable times.
The existing smoke detection system based on smog static nature mainly by the color of smog, texture, circularity,
It is fuzzy to wait static nature or different static natures are combined to judge whether smog.But the above method is but neglected
The dynamic characteristic of smog in fire has been omited, it is relatively low so as to reduce the reliability of smoke detection system.
For drawbacks described above, there is smog dynamic detection system, the brightness value of the smog changed over time modeled,
Smog is judged whether by the change of brightness value.But the existing smog dynamic detection system based on brightness value is carried
The smoke characteristics taken are excessively simple, so as to cause its reliability relatively low.
Therefore, how to provide a kind of high reliability, for judging whether that the smog of smog judges code book, turn into this
The technical problem of art personnel's urgent need to resolve.
The content of the invention
It is an object of the invention to provide the detection side that a kind of smog judges the production method of code book, generation system and smog
Method, it is possible to increase smog judges the reliability of code book, and then can further improve the reliability of Smoke Detection.
To achieve the above object, the invention provides following scheme：
A kind of smog judges the production method of code book, and the production method includes：
Obtain the video of smoke region；
Each two field picture of the video is divided into by some four-dimensional image blocks, the four-dimensional image block using time slip-window
Four dimensional features include：The length of four-dimensional image block, the width of four-dimensional image block, the time span of four-dimensional image block and four-dimensional figure
As the static nature of block, the static nature of the four-dimensional image block includes the RGB feature of four-dimensional image block；
Extract the RGB feature value of each four-dimensional image block；
According to the RGB feature value of each four-dimensional image block, based on linear dynamic system framework, each described four are established
Tie up high-order dynamic system corresponding to image block；
Each high-order dynamic system is clustered, determines that smog judges code book, the smog judge code book for containing
The high-order dynamic system of smog information.
Optionally, the high-order dynamic system is：
xRGB(t+1)=AhxRGB(t)+v(t)
Wherein, xRGB(t) the four-dimensional image block during, time evolution low-profile corresponding with current four-dimensional image block is represented
Static nature value, t represent the time, xRGB(t)=(xR(t) xG(t) xB(t))T, xR(t) represent and current four-dimensional image block pair
The characteristic value of the R component of four-dimensional image block during answer, time evolution low-profile, xG(t) represent and current four-dimensional image block
The characteristic value of the G components of four-dimensional image block during corresponding, time evolution low-profile, xB(t) represent and current four-dimensional image
The characteristic value of the B component of four-dimensional image block corresponding to block, during time evolution low-profile, AhRepresent multidate information matrix, v
(t) Normal Distribution is representedNoise,Noise v (t) variance is represented,Represent current four-dimensional
The static nature value of image block,Represent to work as
The characteristic value of the R component of preceding four-dimensional image block,The characteristic value of the G components of current four-dimensional image block is represented,Table
Show the characteristic value of the B component of current four-dimensional image block,Represent the time span of time span and current four-dimensional image block
The average of the static nature of each four-dimensional image block of identical,
Time span and the average of the R component characteristic value of each four-dimensional image block of time span identical of current four-dimensional image block are represented,Represent the G component characterization values of time span and each four-dimensional image block of time span identical of current four-dimensional image block
Average,Represent the B component of time span and each four-dimensional image block of time span identical of current four-dimensional image block
The average of characteristic value, ChStatic information matrix is represented, w (t) represents Normal DistributionNoise,Expression is made an uproar
Sound w (t) variance.
Optionally, the static nature of the four-dimensional image block also histograms of oriented gradients including the four-dimensional image block is special
Sign.
Optionally, the high-order dynamic system is：
xRGBH(t+1)=AhxRGBH(t)+v(t)
Wherein, xRGBH(t) the quiet of the four-dimensional image block during, time evolution low-profile corresponding with current four-dimensional image block is represented
State characteristic value, t represent time, xRGBH(t)=(xR(t) xG(t) xB(t) xH(t))T, xR(t) represent corresponding with current four-dimensional image block
, the characteristic value of the R component of four-dimensional image block during time evolution low-profile, xG(t) represent it is corresponding with current four-dimensional image block,
The characteristic value of the G components of four-dimensional image block during time evolution low-profile, xB(t) represent it is corresponding with current four-dimensional image block, when
Between four-dimensional image block during evolution low-profile B component characteristic value, xH(t), the time corresponding with current four-dimensional image block is represented
The characteristic value of the histograms of oriented gradients of four-dimensional image block during evolution low-profile, AhMultidate information matrix is represented, v (t) is represented
Normal DistributionNoise,Noise v (t) variance is represented,Represent current four-dimensional image block
Static nature value,
The characteristic value of the R component of current four-dimensional image block is represented,The characteristic value of the G components of current four-dimensional image block is represented,The characteristic value of the B component of current four-dimensional image block is represented,Represent the direction gradient of current four-dimensional image block
The characteristic value of histogram,Represent each four-dimensional image of time span identical of time span and current four-dimensional image block
The average of the static nature of block,During expression
Between span and current four-dimensional image block each four-dimensional image block of time span identical R component characteristic value average,
Time span and the average of the G component characterization values of each four-dimensional image block of time span identical of current four-dimensional image block are represented,Represent the B component characteristic value of time span and each four-dimensional image block of time span identical of current four-dimensional image block
Average,Represent the direction ladder of time span and each four-dimensional image block of time span identical of current four-dimensional image block
Spend the average of histogram feature value, ChStatic information matrix is represented, w (t) represents Normal DistributionNoise,
Represent noise w (t) variance.
A kind of smog judges the generation system of code book, and the generation system includes：
Smog video acquiring module, for obtaining the video of smoke region；
Image block division module, for each two field picture of the video to be divided into some four-dimensional figures using time slip-window
As block, four dimensional features of the four-dimensional image block include：The length of four-dimensional image block, the width of four-dimensional image block, four-dimensional image
The time span of block and the static nature of four-dimensional image block, the static nature of the four-dimensional image block include four-dimensional image block
RGB feature；
Characteristics extraction module, for extracting the RGB feature value of each four-dimensional image block；
High-order dynamic system establishes module, for the RGB feature value according to each four-dimensional image block, based on linear dynamic
State system framework, establish high-order dynamic system corresponding to each four-dimensional image block；
Code book determining module, for each high-order dynamic system to be clustered, determine that smog judges code book, the cigarette
Mist judges code book for the high-order dynamic system containing smog information.
A kind of detection method of smog, the detection method include：
The linear dynamic system of video and smog video to be detected is obtained, the video to be detected includes each frame mapping to be checked
Picture；
The characteristic value of the static nature of image to be detected described in each frame is extracted, the static nature of described image to be detected includes
The RGB feature of image and the histograms of oriented gradients feature of image；
According to the characteristic value of the static nature of image to be detected described in each frame, based on linear dynamic system framework, establish
High-order dynamic system corresponding to image to be detected described in each frame；
According to formula：RTPR-P=-QTQ, determine that high-order dynamic system corresponding to each described image to be detected is sentenced with smog
The similarity measure values of any high-order dynamic system in short in size book, wherein, the smog judge code book for according to claim 1~
Smog caused by production method described in 4 any one judges code book,Treated described in expression
The multidate information matrix to be measured of high-order dynamic system corresponding to detection image,Represent that high-order corresponding to described image to be detected moves
The static information matrix to be measured of state system,Represent that smog judges the judgement multidate information square of any high-order dynamic system in code book
Battle array,Represent the judgement static information matrix of high-order dynamic system corresponding with the judgement multidate information matrix；
The similarity measure values of the similar threshold value less than setting are judged whether, obtain the first judged result；
If the first judged result represents the similarity measure values that the similar threshold value less than setting be present, it is determined that described to be checked
Smog be present in altimetric image；
If the first judged result represents that the similarity measure values of the similar threshold value less than setting are not present, it is determined that described to treat
Smog is not present in detection image；
Optionally, it is described according to formula：RTPR-P=-QTQ, determine high-order dynamic corresponding to each described image to be detected
System judges the similarity measure values of any high-order dynamic system in code book with smog, specifically includes：
According to formula：RTPR-P=-QTQ, the multidate information matrix to be measured, the static information matrix to be measured, described sentence
Disconnected multidate information matrix and the judgement static information matrix, determine matrix P,
According to formula：Similarity measure values are determined, wherein,Represent similarity measure values, λiTable
Show matrixIth feature value.
According to specific embodiment provided by the invention, the invention discloses following technique effect：
Each two field picture of smog video is divided into multiple four-dimensional image blocks by the present invention using time slip-window, is then established
The high-order dynamic system of each four-dimensional image block comprising RGB feature, passes through the high-order dynamic of each four-dimensional image block of clustering
System, obtain for judging whether that the smog of smog judges code book.Because smog provided by the invention judges that code book is abundant
The RGB feature information of color video is considered, therefore, it is possible to improve the reliability that smog judges code book.Further, using this
The smog that invention provides judges that code book carries out Smoke Detection, it is possible to increase the reliability and accuracy of detection of Smoke Detection.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to institute in embodiment
The accompanying drawing needed to use is briefly described, it should be apparent that, drawings in the following description are only some implementations of the present invention
Example, for those of ordinary skill in the art, on the premise of not paying creative work, can also be obtained according to these accompanying drawings
Obtain other accompanying drawings.
Fig. 1 is the flow chart for the production method that the smog that the embodiment of the present invention 1 provides judges code book；
Fig. 2 is the structured flowchart for the generation system that the smog that the embodiment of the present invention 2 provides judges code book；
Fig. 3 is the flow chart of the detection method for the smog that the embodiment of the present invention 3 provides；
Fig. 4 is the structural representation for the four-dimensional image block that the embodiment of the present invention 3 provides.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other under the premise of creative work is not made
Embodiment, belong to the scope of protection of the invention.
It is an object of the invention to provide the detection side that a kind of smog judges the production method of code book, generation system and smog
Method, it is possible to increase smog judges the reliability of code book, and then can further improve the reliability of Smoke Detection.
In order to facilitate the understanding of the purposes, features and advantages of the present invention, it is below in conjunction with the accompanying drawings and specific real
Applying mode, the present invention is further detailed explanation.
Embodiment 1：
As shown in figure 1, a kind of smog judges that the production method of code book includes：
Step 11：Obtain the video of smoke region；
Step 12：Each two field picture of the video is divided into by some four-dimensional image blocks using time slip-window, described four
Four dimensional features of dimension image block include：The length of four-dimensional image block, the width of four-dimensional image block, the time span of four-dimensional image block
With the static nature of four-dimensional image block, the static nature of the four-dimensional image block includes the RGB feature of four-dimensional image block；
Step 13：Extract the RGB feature value of each four-dimensional image block；
Step 14：According to the RGB feature value of each four-dimensional image block, based on linear dynamic system framework, establish every
High-order dynamic system corresponding to the one four-dimensional image block；
Step 15：Each high-order dynamic system is clustered, determines that smog judges code book, the smog judges code book
For the high-order dynamic system containing smog information.In the present embodiment, using K- central points (K-medoids) algorithm by each institute
The high-order dynamic system for stating four-dimensional image block carries out clustering, and obtained cluster result includes three classes, and one kind is free of smog, separately
One kind contain smog, also have it is a kind of include from color and move composition similar with smog for above, using the cluster containing smog as
Final smog judges code book.
Wherein, the high-order dynamic system established in step 14 is：
xRGB(t+1)=AhxRGB(t)+v(t)
Wherein, xRGB(t) the four-dimensional image block during, time evolution low-profile corresponding with current four-dimensional image block is represented
Static nature value, t represent the time, xRGB(t)=(xR(t) xG(t) xB(t))T, xR(t) represent and current four-dimensional image block pair
The characteristic value of the R component of four-dimensional image block during answer, time evolution low-profile, xG(t) represent and current four-dimensional image block
The characteristic value of the G components of four-dimensional image block during corresponding, time evolution low-profile, xB(t) represent and current four-dimensional image
The characteristic value of the B component of four-dimensional image block corresponding to block, during time evolution low-profile, AhRepresent multidate information matrix, v
(t) Normal Distribution is representedNoise,Noise v (t) variance is represented,Represent current four
The static nature value of image block is tieed up,Table
Show the characteristic value of the R component of current four-dimensional image block,The characteristic value of the G components of current four-dimensional image block is represented,
The characteristic value of the B component of current four-dimensional image block is represented,Represent the time span of time span and current four-dimensional image block
The average of the static nature of each four-dimensional image block of identical,
Time span and the average of the R component characteristic value of each four-dimensional image block of time span identical of current four-dimensional image block are represented,Represent time span and the G component characterization values of each four-dimensional image block of time span identical of current four-dimensional image block
Average,Represent that the B component of time span and each four-dimensional image block of time span identical of current four-dimensional image block is special
The average of value indicative, ChStatic information matrix is represented, w (t) represents Normal DistributionNoise,Represent noise w
(t) variance.
Preferably, direction ladder of the static nature of the four-dimensional image block in step 12 also including the four-dimensional image block
Histogram feature is spent, i.e., the static nature of described four-dimensional image block includes the RGB feature of four-dimensional image block and the four-dimensional image
The histograms of oriented gradients feature of block, corresponding high-order dynamic system are：
xRGBH(t+1)=AhxRGBH(t)+v(t)
Wherein, xRGBH(t) the four-dimensional image block during, time evolution low-profile corresponding with current four-dimensional image block is represented
Static nature value, t represent the time, xRGBH(t)=(xR(t) xG(t) xB(t) xH(t))T, xR(t) represent and current four-dimensional image
The characteristic value of the R component of four-dimensional image block corresponding to block, during time evolution low-profile, xG(t) represent and current four-dimensional image block
The characteristic value of the G components of four-dimensional image block during corresponding, time evolution low-profile, xB(t) represent and current four-dimensional image block pair
The characteristic value of the B component of four-dimensional image block during answer, time evolution low-profile, xH(t) represent corresponding with current four-dimensional image block
, the characteristic value of the histograms of oriented gradients of four-dimensional image block during time evolution low-profile, AhRepresent multidate information matrix, v (t)
Represent Normal DistributionNoise,Noise v (t) variance is represented,Represent current four-dimensional image block
Static nature value,
The characteristic value of the R component of current four-dimensional image block is represented,The characteristic value of the G components of current four-dimensional image block is represented,The characteristic value of the B component of current four-dimensional image block is represented,Represent the direction gradient of current four-dimensional image block
The characteristic value of histogram,Represent each four-dimensional image of time span identical of time span and current four-dimensional image block
The average of the static nature of block,During expression
Between span and current four-dimensional image block each four-dimensional image block of time span identical R component characteristic value average,
Time span and the average of the G component characterization values of each four-dimensional image block of time span identical of current four-dimensional image block are represented,Represent the B component characteristic value of time span and each four-dimensional image block of time span identical of current four-dimensional image block
Average,Represent the direction ladder of time span and each four-dimensional image block of time span identical of current four-dimensional image block
Spend the average of histogram feature value, ChStatic information matrix is represented, w (t) represents Normal DistributionNoise,
Represent noise w (t) variance.
In order to lift the high efficiency and reliability that smog judges code book, it is (red that the present invention introduces RGB in four-dimensional image block
(R), green (G), blue (B) three passages) feature and histograms of oriented gradients (Histogram of Oriented Gradient,
HOG) feature, avoid and the problem of intensive sampling occur, and add the dimension of image block, improve image texture characteristic point
The reliability of analysis, amount of calculation are small.
Embodiment 2：
As shown in Fig. 2 a kind of smog judges that the generation system of code book includes：
Smog video acquiring module 21, for obtaining the video of smoke region；
Image block division module 22, for each two field picture of the video to be divided into some four-dimension using time slip-window
Image block, four dimensional features of the four-dimensional image block include：The length of four-dimensional image block, the width of four-dimensional image block, four-dimensional figure
As the time span of block and the static nature of four-dimensional image block, the static nature of the four-dimensional image block includes four-dimensional image block
RGB feature；
Characteristics extraction module 23, for extracting the RGB feature value of each four-dimensional image block；
High-order dynamic system establishes module 24, for the RGB feature value according to each four-dimensional image block, based on linear
Dynamical system framework, establish high-order dynamic system corresponding to each four-dimensional image block；
Code book determining module 25, for each high-order dynamic system to be clustered, determine that smog judges code book, it is described
Smog judges code book for the high-order dynamic system containing smog information.In the present embodiment, using K- central points (K-
Medoids) algorithm carries out clustering to each high-order dynamic system.
The present invention introduces RGB (red (R), green (G), blue (B) three passages) features and direction gradient in four-dimensional image block
Histogram (HistogramofOrientedGradient, HOG) feature, has taken into full account the textural characteristics of smog, therefore, leads to
The system for crossing the present embodiment offer, the smog that can establish high reliability judge code book.
Embodiment 3：
As shown in figure 3, a kind of detection method of smog includes：
Step 31：The linear dynamic system of video and smog video to be detected is obtained, the video to be detected includes each frame
Image to be detected；
Step 32：Extract the characteristic value of the static nature of image to be detected described in each frame, the static state of described image to be detected
Feature includes the RGB feature of image and the histograms of oriented gradients feature of image；
Step 33：According to the characteristic value of the static nature of image to be detected described in each frame, based on linear dynamic system frame
Frame, establish high-order dynamic system corresponding to image to be detected described in each frame；
Step 34：According to formula：RTPR-P=-QTQ, determine high-order dynamic system corresponding to each described image to be detected
The similarity measure values of any high-order dynamic system in code book are judged with smog, wherein, the smog judges code book for according to reality
Smog caused by applying the production method described in example 1 judges code book,Represent described to be checked
The multidate information matrix to be measured of high-order dynamic system corresponding to altimetric image,Represent high-order dynamic corresponding to described image to be detected
The static information matrix to be measured of system,Represent that smog judges the judgement multidate information square of any high-order dynamic system in code book
Battle array,Represent the judgement static information matrix of high-order dynamic system corresponding with the judgement multidate information matrix；
Step 35：The similarity measure values of the similar threshold value less than setting are judged whether, obtain the first judged result；
If the first judged result represents the similarity measure values that the similar threshold value less than setting be present, step 36 is performed；
If the first judged result represents that the similarity measure values of the similar threshold value less than setting are not present, step is performed
37；
Step 36：Then determine that described image to be detected has smog；
Step 37：Then determine that smog is not present in described image to be detected；
Specifically, step 34 specifically includes：
Step 341：According to formula：RTPR-P=-QTQ, the multidate information matrix to be measured, the static information square to be measured
Battle array, the judgement multidate information matrix and the judgement static information matrix, determine matrix
Step 342：According to formula：Similarity measure values are determined, wherein,Represent similarity measurement
Value, λiRepresenting matrixIth feature value.
In the present embodiment, video sequence pretreatment is carried out before step 31 is performed, in addition to video to be detected.Video
Sequence pretreatment is by sport foreground separation and color analysis, first to quiet under motion video image and same camera lens
State background is contrasted, and draws the track that moving target moves during camera is shot；Then color is carried out to moving target
Analysis and Gaussian smoothing filter, obtain preliminary smog suspected target region.Comprise the following steps that：
1) sport foreground separates
Sport foreground separation using independent component analysis (Independent ComponentAnalysis, ICA) technology and
Vision significance (Graph-BasedVisual Saliency, GBVS) based on figure carries out smoke foreground separation.Sport foreground
Separation first uses improved ICA moving objects identification model initial gross separation smoke foreground region in smog separation phase, then passes through
GBVS determines smoke foreground marking area.
2) color analysis
At the initial stage that fire occurs, because temperature is than relatively low, object often burns not enough fully, can produce linen cigarette
Mist.Smog itself is translucent, and for smog often than sparse, this can cause background image to fog during just generation；With
Smoke density increase, background image all cover from translucent to by smog, and background image loses original color characteristic, institute
So that smog can be judged whether according to color characteristic.
Under RGB color model, the gray value of tri- components of R, G, B of smog is fairly close, so if meet simultaneously
Following two conditions, then it is assumed that the pixel meets the color characteristic of smog：
Condition 1：Still equal, i.e. h three components of RGB color model increase and decrease simultaneously afterR± a=hG± a=hB± a, its
In, hRRepresent the gray value of R component, hGRepresent the gray value of G components, hBThe gray value of B component is represented, a represents constant；
Condition 2：It is the I component satisfaction of HSI models after HSI color models by RGB color model conversation：L1≤I≤L2 or
Person H1≤I≤H2, wherein, I represents that intensity or brightness, L1, L2, H1, H2 are component I empirical value.
After carrying out above-mentioned video sequence pretreatment, pretreated video to be detected is obtained, Smoke Detection can be reduced
Amount of calculation, improve detection speed and detection efficiency.
In the present embodiment, step 33 specifically includes：
Step 331：Establish the linear dynamic system framework (LinearDynamical System, LDS) of smog video：
Averaging model (Auto-Regressive Moving Average, ARMA) is moved based on first-order autoregression, simultaneously
The independent identically distributed white Gaussian noise with zero-mean is added, constructs the linear dynamic system framework of a smog video, it is fixed
Justice is as follows：
X (t+1)=Ax (t)+v (t) (1)
Wherein, t represents the time, and x (t) expressions are corresponding with the image of present frame, image during time evolution low-profile
Grey scale pixel value, and x (t) ∈ Rn, the image pixel gray level value and y (t) ∈ R of y (t) expression present framesd, n expression x (t) centre
Status number, d represents the sum of all pixels of current video image, and n≤d, A represent transition matrix and A ∈ Rn×n, C expression mapping matrixes
And C ∈ Rd×n, R expression real number fields, v (t) and w (t) represent caused noise in image generation process, v (t) expression obedience normal states
DistributionNoise, w (t) represent Normal DistributionNoise, wherein,Represent noise v (t) side
Difference,Noise w (t) variance is represented,Represent video in each two field picture pixel grey scale average and
Formula (1) illustrates the recurrence relation of adjacent time evolution low-profile process, and it is each time that formula (2), which illustrates y (t),
Evolution low-profile process x (t) linear expression, i.e. matrix A and C determine the linear dynamic system framework of smog video.
When handling multitude of video sequences, it is necessary to first solve A and C to estimate the linear dynamic system of smog video (LDS)
Framework, then carry out intensive sampling.However, be not that each sample image is effective in video to be measured, for example smog occurs early
Phase is either apart from distant and cause the smog that photographs than leaner, therefore the video photographed is more fuzzy.If will
This fuzzy video also serves as sample process, it will causes amount of calculation excessive.To overcome this limitation, while in order to determine image
The specific time that the accurate location and smog of middle smog occur, a 3-D view block, long a width of N × N are established, time span is
F, with Y ∈ RN×N×FTo represent, such as formula (3)：
Y=[ypatch(1),ypatch(2),…,ypatch(F)] (3)
In formula (3), Y represents 3-D view block collection, ypatch(t)∈RN×N(t=1,2 ..., F) represent each 3-D view
The gray value of block output, can be obtained by formula (2)：
In the linear dynamic system framework of smog video, matrix A and C solution procedure are as follows：
First construct a zero-mean matrix：
In formula (5),Represent the gray average of each 3-D view block.
To Y0Singular value decomposition is carried out, is obtained：
Y0=USVT (6)
In formula (6), U is d rank reality unitary matrice, and d represents the sum of all pixels of current video image, and S represents that d × F ranks are diagonal
Matrix, F represent image block time span, and the element in S on diagonal is Y0Singular value, VTIt is V transposition, V is the F rank reality tenth of the twelve Earthly Branches
Matrix.
It can be obtained by formula (4) and formula (5)：
Y0=CX (7)
Wherein, X=[x (1), x (2) ..., x (F)].
It can be obtained by formula (6) and formula (7)：
C=U (8)
Make X=[x (1) ..., x (F)], X1=[x (1) ..., x (F-1)], X2=[x (2), x (3) ..., x (F)], then：
Therefore, the linear dynamic system framework of smog video can be expressed as：MLDS=(A, C).
Step 332：Establish high order linear dynamical system (the Higher order-Linear Dynamical of smog video
System, h-LDS)
The high order linear dynamical system framework (h-LDS) of smog video can improve the reliability and drop of dynamic Feature Analysis
Low amount of calculation.Each two field picture of video is divided into the equal subsequence in some intervals by h-LDS using time slip-window, through undue
Block, extension obtain four-dimensional image block.Then a tetradic Y is constructedh∈RN×N×M×FTo represent four-dimensional image block, wherein N is
The length of side of image block, F are four-dimensional image block time spans, and M represents image static nature.For RGB color video, static nature
RGB feature and histograms of oriented gradients feature including four-dimensional image block.The structure for being illustrated in figure 4 four-dimensional spacetime image block is shown
It is intended to, wherein the characteristic value of the correspondence image block R component of passage 1, the characteristic value of the correspondence image block G components of passage 2, the correspondence of passage 3
The characteristic value of image block B component, the histograms of oriented gradients characteristic value of the correspondence image block R component of passage 4.
Four-dimensional image block Yh∈RN×N×M×FIt can be expressed as：
In formulaYh∈RN×N×M×F, t=1,2 ..., F.It is similar with LDS, with formula (1) and formula
(4) the high-order dynamic system of four-dimensional image block is established.Using high order tensor singular value decomposition (High Order
SingularValue Decomposition, HOSVD) method determine comprising image block multidate information and static information square
Battle array AhAnd Ch。
First YhIn the static nature value of each four-dimensional image block subtract YhIn all four-dimensional image blocks static nature value
AverageZero-mean matrix is obtained, formula (6) is copied, then tensor YhDecompose as follows：
S is d × F rank diagonal matrix in formula, UT (1), UT (2)..., UT (k)It is orthogonal matrix, is made up of orthogonal vectors, T is represented
It is the T moment, corresponding with time span F, ×i(i=1 ..., k) it is between tensor YhAnd matrix UT (k)Between i states.Therefore, square
Battle array Ah, ChIt is not unique, as long as there is suitable primary condition, it will there is numerous solution.
Make X=S ×1UT (1)…×k-1UT (k-1), Ch=UT (k), Ch∈RF×F, then formula (11) can turn to：
(12) formula can be turned to furtherWherein X (k)=X ×kAnd X(k)X k dimension expansions are represented, its
Form is consistent with the form of formula (7).
Order：
It can then be obtained with reference to formula (9)：
Ah∈RN×N.So transition matrix AhAnd ChAfter all determining, RGB feature is characterized
H-LDS models are represented by：
xRGB(t+1)=AhxRGB(t)+v(t) (15)
Wherein, xRGB(t) the four-dimensional image block during, time evolution low-profile corresponding with current four-dimensional image block is represented
Static nature value, t represent the time, xRGB(t)=(xR(t) xG(t) xB(t))T, xR(t) represent and current four-dimensional image block pair
The characteristic value of the R component of four-dimensional image block during answer, time evolution low-profile, xG(t) represent and current four-dimensional image block
The characteristic value of the G components of four-dimensional image block during corresponding, time evolution low-profile, xB(t) represent and current four-dimensional image
The characteristic value of the B component of four-dimensional image block corresponding to block, during time evolution low-profile, AhRepresent multidate information matrix, v
(t) Normal Distribution is representedNoise,The static nature value of current four-dimensional image block is represented,Represent R points of current four-dimensional image block
The characteristic value of amount,The characteristic value of the G components of current four-dimensional image block is represented,Represent current four-dimensional image block
B component characteristic value,Represent each four-dimensional image of time span identical of time span and current four-dimensional image block
The average of the static nature of block,Represent time span with
The average of the R component characteristic value of each four-dimensional image block of time span identical of current four-dimensional image block,Represent the time
The average of span and the G component characterization values of each four-dimensional image block of time span identical of current four-dimensional image block,Table
Show time span and the average of the B component characteristic value of each four-dimensional image block of time span identical of current four-dimensional image block, Ch
Static information matrix is represented, w (t) represents Normal DistributionNoise.
Smog is shown as brilliant white region in the visual image of histograms of oriented gradients (HOG) feature, and the present embodiment is
The ability that enhancing high-order dynamic system is handled multi-dimensional image data, while improve the robustness of detection method, in RGB feature
On the basis of HOG features are added to the static nature of four-dimensional image block.It is only doubtful to preliminary smog in order to reduce amount of calculation
The image zooming-out HOG features of target area, use hpatch(t) the HOG features of image block are represented.With U (t) ∈ Rd×LWith V (t) ∈ RD ×LRepresent respectivelyAnd hpatch(t) base, d represent the sum of all pixels of four-dimensional image block, and D represents direction gradient Nogata
Scheme the characteristic dimension of (HOG), L represents coefficient atDimension.AlthoughAnd hpatch(t) base is different, but uses oversubscription
Resolution sparse coding and the method for sparse dictionary study can obtain identical coefficient at, represent as follows：
In formula, hpatch(t)∈RDThe HOG features of image block are represented,Represent four-dimensional image block
RGB feature.
Therefore, representing the tetradic of four-dimensional image block static nature can be write as again：
Wherein F is the time span of four-dimensional image block,The Y that will newly obtainhSubstitute into formula (11), can obtain static nature include RGB feature and
The high order linear dynamical system of HOG features：
xRGBH(t+1)=AhxRGBH(t)+v(t) (18)
Wherein, xRGBH(t) the four-dimensional image block during, time evolution low-profile corresponding with current four-dimensional image block is represented
Static nature value, t represent the time, xRGBH(t)=(xR(t) xG(t) xB(t) xH(t))T, xR(t) represent and the current four-dimension
The characteristic value of the R component of four-dimensional image block corresponding to image block, during time evolution low-profile, xG(t) represent and current four
Tie up the characteristic value of the G components of the four-dimensional image block corresponding to image block, during time evolution low-profile, xB(t) represent and current
The characteristic value of the B component of four-dimensional image block corresponding to four-dimensional image block, during time evolution low-profile, xH(t) represent and work as
The characteristic value of the histograms of oriented gradients of four-dimensional image block corresponding to preceding four-dimensional image block, during time evolution low-profile, AhTable
Show multidate information matrix, v (t) represents Normal DistributionNoise,Represent current four-dimensional image block
Static nature value,
The characteristic value of the R component of current four-dimensional image block is represented,The characteristic value of the G components of current four-dimensional image block is represented,The characteristic value of the B component of current four-dimensional image block is represented,Represent the direction gradient of current four-dimensional image block
The characteristic value of histogram,Represent each four-dimensional image of time span identical of time span and current four-dimensional image block
The average of the static nature of block,During expression
Between span and current four-dimensional image block each four-dimensional image block of time span identical R component characteristic value average,
Time span and the average of the G component characterization values of each four-dimensional image block of time span identical of current four-dimensional image block are represented,Represent the B component characteristic value of time span and each four-dimensional image block of time span identical of current four-dimensional image block
Average,Represent the direction ladder of time span and each four-dimensional image block of time span identical of current four-dimensional image block
Spend the average of histogram feature value, ChStatic information matrix is represented, w (t) represents Normal DistributionNoise.
The present embodiment is pre-processing smog separation phase, is obtained just using ICA smoke foreground initial gross separation smoke models
Smoke foreground is walked, smoke foreground marking area is then obtained by GBVS extraction images multichannel, multiple dimensioned low-level image feature, most
Column hisgram contrast is entered to identify smog according to the combinations of features of color and histograms of oriented gradients afterwards, improves foreground target detection
The accuracy of separation of the stage to smoke foreground.
The stage is extracted in smoke characteristics, proposes that the smoke characteristics based on multidimensional characteristic analysis extract detection method.This method
Smog color and background difference is first passed around to pre-process to obtain smog candidate region, then in four-dimensional image block introduce RGB and
HOG features, the high-order decomposition to multi-dimensional image data is finally based on, the behavioral characteristics of smog video is analyzed, due to using slip
Time window, it may be determined that the specific time that the accurate location of smog and smog occur in picture, improve smoke characteristics extraction rank
The smoke characteristics stability extracted of section it is not high and to the judgment criterion of smog it is excessively simple the shortcomings that, improve dynamic texture spy
The reliability of analysis is levied, reduces amount of calculation.
Each embodiment is described by the way of progressive in this specification, what each embodiment stressed be and other
The difference of embodiment, between each embodiment identical similar portion mutually referring to.For system disclosed in embodiment
For, because it is corresponded to the method disclosed in Example, so description is fairly simple, related part is said referring to method part
It is bright.
Specific case used herein is set forth to the principle and embodiment of the present invention, and above example is said
It is bright to be only intended to help the method and its core concept for understanding the present invention；Meanwhile for those of ordinary skill in the art, foundation
The thought of the present invention, in specific embodiments and applications there will be changes.In summary, this specification content is not
It is interpreted as limitation of the present invention.
Claims (7)
1. a kind of smog judges the production method of code book, it is characterised in that the production method includes：
Obtain the video of smoke region；
Each two field picture of the video is divided into by some four-dimensional image blocks using time slip-window, the four of the four-dimensional image block
Dimensional feature includes：The length of four-dimensional image block, the width of four-dimensional image block, the time span and four-dimensional image block of four-dimensional image block
Static nature, the static nature of the four-dimensional image block includes the RGB feature of four-dimensional image block；
Extract the RGB feature value of each four-dimensional image block；
According to the RGB feature value of each four-dimensional image block, based on linear dynamic system framework, each four-dimensional figure is established
The high-order dynamic system as corresponding to block；
Each high-order dynamic system is clustered, determines that smog judges code book, the smog judges code book to contain smog
The high-order dynamic system of information.
2. production method according to claim 1, it is characterised in that the high-order dynamic system is：
xRGB(t+1)=AhxRGB(t)+v(t)
<mrow>
<msubsup>
<mi>y</mi>
<mrow>
<mi>p</mi>
<mi>a</mi>
<mi>t</mi>
<mi>c</mi>
<mi>h</mi>
<mo>-</mo>
<mi>R</mi>
<mi>G</mi>
<mi>B</mi>
</mrow>
<mi>h</mi>
</msubsup>
<mo>=</mo>
<msubsup>
<mover>
<mi>y</mi>
<mo>&OverBar;</mo>
</mover>
<mrow>
<mi>p</mi>
<mi>a</mi>
<mi>t</mi>
<mi>c</mi>
<mi>h</mi>
<mo>-</mo>
<mi>R</mi>
<mi>G</mi>
<mi>B</mi>
</mrow>
<mi>h</mi>
</msubsup>
<mo>+</mo>
<msub>
<mi>C</mi>
<mi>h</mi>
</msub>
<msub>
<mi>x</mi>
<mrow>
<mi>R</mi>
<mi>G</mi>
<mi>B</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>w</mi>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
</mrow>
Wherein, xRGB(t) the quiet of the four-dimensional image block during, time evolution low-profile corresponding with current four-dimensional image block is represented
State characteristic value, t represent time, xRGB(t)=(xR(t) xG(t) xB(t))T, xR(t) represent corresponding with current four-dimensional image block
, the characteristic value of the R component of four-dimensional image block during time evolution low-profile, xG(t) represent and current four-dimensional image block pair
The characteristic value of the G components of four-dimensional image block during answer, time evolution low-profile, xB(t) represent and current four-dimensional image block
The characteristic value of the B component of four-dimensional image block during corresponding, time evolution low-profile, AhRepresent multidate information matrix, v (t)
Represent Normal DistributionNoise,Noise v (t) variance is represented,Represent current four-dimensional
The static nature value of image block, Represent to work as
The characteristic value of the R component of preceding four-dimensional image block,The characteristic value of the G components of current four-dimensional image block is represented,
The characteristic value of the B component of current four-dimensional image block is represented,Represent the when span of time span and current four-dimensional image block
The average of the static nature of each four-dimensional image block of identical is spent, Represent time span and the R component characteristic value of each four-dimensional image block of time span identical of current four-dimensional image block
Average,Represent that the G components of time span and each four-dimensional image block of time span identical of current four-dimensional image block are special
The average of value indicative,Represent the B of time span and each four-dimensional image block of time span identical of current four-dimensional image block
The average of component characterization value, ChStatic information matrix is represented, w (t) represents Normal DistributionNoise,Table
Show noise w (t) variance.
3. production method according to claim 1, it is characterised in that the static nature of the four-dimensional image block also includes institute
State the histograms of oriented gradients feature of four-dimensional image block.
4. production method according to claim 3, it is characterised in that the high-order dynamic system is：
xRGBH(t+1)=AhxRGBH(t)+v(t)
<mrow>
<msubsup>
<mi>y</mi>
<mrow>
<mi>p</mi>
<mi>a</mi>
<mi>t</mi>
<mi>c</mi>
<mi>h</mi>
<mo>-</mo>
<mi>R</mi>
<mi>G</mi>
<mi>B</mi>
<mi>H</mi>
</mrow>
<mi>h</mi>
</msubsup>
<mo>=</mo>
<msubsup>
<mover>
<mi>y</mi>
<mo>&OverBar;</mo>
</mover>
<mrow>
<mi>p</mi>
<mi>a</mi>
<mi>t</mi>
<mi>c</mi>
<mi>h</mi>
<mo>-</mo>
<mi>R</mi>
<mi>G</mi>
<mi>B</mi>
<mi>H</mi>
</mrow>
<mi>h</mi>
</msubsup>
<mo>+</mo>
<msub>
<mi>C</mi>
<mi>h</mi>
</msub>
<msub>
<mi>x</mi>
<mrow>
<mi>R</mi>
<mi>G</mi>
<mi>B</mi>
<mi>H</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>w</mi>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
</mrow>
1
Wherein, xRGBH(t) the quiet of the four-dimensional image block during, time evolution low-profile corresponding with current four-dimensional image block is represented
State characteristic value, t represent time, xRGBH(t)=(xR(t) xG(t) xB(t) xH(t))T, xR(t) represent and current four-dimensional image block pair
The characteristic value of the R component of four-dimensional image block during answer, time evolution low-profile, xG(t) represent and current four-dimensional image block pair
The characteristic value of the G components of four-dimensional image block during answer, time evolution low-profile, xB(t) represent and current four-dimensional image block pair
The characteristic value of the B component of four-dimensional image block during answer, time evolution low-profile, xH(t) represent and current four-dimensional image block pair
The characteristic value of the histograms of oriented gradients of four-dimensional image block during answer, time evolution low-profile, AhMultidate information matrix is represented,
V (t) represents Normal DistributionNoise,Noise v (t) variance is represented,Represent current four-dimensional
The static nature value of image block, The characteristic value of the R component of current four-dimensional image block is represented,Represent the spy of the G components of current four-dimensional image block
Value indicative,The characteristic value of the B component of current four-dimensional image block is represented,Represent the direction of current four-dimensional image block
The characteristic value of histogram of gradients,Represent each four-dimension of time span identical of time span and current four-dimensional image block
The average of the static nature of image block, Represent
Time span and the average of the R component characteristic value of each four-dimensional image block of time span identical of current four-dimensional image block,Represent time span and the G component characterization values of each four-dimensional image block of time span identical of current four-dimensional image block
Average,Represent that the B component of time span and each four-dimensional image block of time span identical of current four-dimensional image block is special
The average of value indicative,Represent the side of time span and each four-dimensional image block of time span identical of current four-dimensional image block
To the average of histogram of gradients characteristic value, ChStatic information matrix is represented, w (t) represents Normal DistributionMake an uproar
Sound,Represent noise w (t) variance.
5. a kind of smog judges the generation system of code book, it is characterised in that the generation system includes：
Smog video acquiring module, for obtaining the video of smoke region；
Image block division module, for each two field picture of the video to be divided into some four-dimensional images using time slip-window
Block, four dimensional features of the four-dimensional image block include：The length of four-dimensional image block, the width of four-dimensional image block, four-dimensional image block
Time span and four-dimensional image block static nature, the static nature of the four-dimensional image block includes the RGB of four-dimensional image block
Feature；
Characteristics extraction module, for extracting the RGB feature value of each four-dimensional image block；
High-order dynamic system establishes module, for the RGB feature value according to each four-dimensional image block, based on linear dynamic system
System framework, establishes high-order dynamic system corresponding to each four-dimensional image block；
Code book determining module, for each high-order dynamic system to be clustered, determine that smog judges code book, the smog is sentenced
Short in size book is the high-order dynamic system containing smog information.
6. a kind of detection method of smog, it is characterised in that the detection method includes：
The linear dynamic system of video and smog video to be detected is obtained, the video to be detected includes each frame image to be detected；
The characteristic value of the static nature of image to be detected described in each frame is extracted, the static nature of described image to be detected includes image
RGB feature and image histograms of oriented gradients feature；
According to the characteristic value of the static nature of image to be detected described in each frame, based on linear dynamic system framework, establish each
High-order dynamic system corresponding to image to be detected described in frame；
According to formula：RTPR-P=-QTQ, determine that high-order dynamic system corresponding to each described image to be detected judges code with smog
The similarity measure values of any high-order dynamic system in book, wherein, the smog judges code book to appoint according to Claims 1 to 4
Smog caused by production method described in one judges code book, Represent described to be detected
The multidate information matrix to be measured of high-order dynamic system corresponding to image,Represent high-order dynamic system corresponding to described image to be detected
The static information matrix to be measured of system,Represent that smog judges the judgement multidate information matrix of any high-order dynamic system in code book,Represent the judgement static information matrix of high-order dynamic system corresponding with the judgement multidate information matrix；
The similarity measure values of the similar threshold value less than setting are judged whether, obtain the first judged result；
If the first judged result represents the similarity measure values that the similar threshold value less than setting be present, it is determined that the mapping to be checked
As smog be present；
If the first judged result represents that the similarity measure values of the similar threshold value less than setting are not present, it is determined that described to be detected
Smog is not present in image；
7. detection method according to claim 6, it is characterised in that described according to formula：RTPR-P=-QTQ, it is determined that often
High-order dynamic system corresponding to one described image to be detected judges the similarity measurements of any high-order dynamic system in code book with smog
Value, specifically include：
According to formula：RTPR-P=-QTQ, the multidate information matrix to be measured, the static information matrix to be measured, it is described judge it is dynamic
State information matrix and the judgement static information matrix, determine matrix P,
According to formula：Similarity measure values are determined, wherein,Represent similarity measure values, λiRepresenting matrixIth feature value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710742248.9A CN107463968A (en) | 2017-08-25 | 2017-08-25 | Smog judges the detection method of the production method of code book, generation system and smog |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710742248.9A CN107463968A (en) | 2017-08-25 | 2017-08-25 | Smog judges the detection method of the production method of code book, generation system and smog |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107463968A true CN107463968A (en) | 2017-12-12 |
Family
ID=60550596
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710742248.9A Pending CN107463968A (en) | 2017-08-25 | 2017-08-25 | Smog judges the detection method of the production method of code book, generation system and smog |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107463968A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111127433A (en) * | 2019-12-24 | 2020-05-08 | 深圳集智数字科技有限公司 | Method and device for detecting flame |
| CN111145222A (en) * | 2019-12-30 | 2020-05-12 | 浙江中创天成科技有限公司 | Fire detection method combining smoke movement trend and textural features |
| CN112132124A (en) * | 2020-11-30 | 2020-12-25 | 江苏久智环境科技服务有限公司 | Real-time VR video detection method for garden monitoring |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101339602B (en) * | 2008-07-15 | 2011-05-04 | 中国科学技术大学 | Video frequency fire hazard aerosol fog image recognition method based on light stream method |
| CN105469105A (en) * | 2015-11-13 | 2016-04-06 | 燕山大学 | Cigarette smoke detection method based on video monitoring |
-
2017
- 2017-08-25 CN CN201710742248.9A patent/CN107463968A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101339602B (en) * | 2008-07-15 | 2011-05-04 | 中国科学技术大学 | Video frequency fire hazard aerosol fog image recognition method based on light stream method |
| CN105469105A (en) * | 2015-11-13 | 2016-04-06 | 燕山大学 | Cigarette smoke detection method based on video monitoring |
Non-Patent Citations (3)
| Title |
|---|
| CHAO-HO CHEN 等: "The Smoke Detection for Early Fire-Alarming System Base on Video Processing", 《IEEE》 * |
| KOSMAS DIMITROPOULOS等: "Higher Order Linear Dynamical Systems for Smoke Detection in Video Surveillance Applications", 《IEEE》 * |
| 崔丽娜 等: "基于增量记忆视觉注意模型的复杂目标识别研究", 《微型机与应用》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111127433A (en) * | 2019-12-24 | 2020-05-08 | 深圳集智数字科技有限公司 | Method and device for detecting flame |
| CN111145222A (en) * | 2019-12-30 | 2020-05-12 | 浙江中创天成科技有限公司 | Fire detection method combining smoke movement trend and textural features |
| CN112132124A (en) * | 2020-11-30 | 2020-12-25 | 江苏久智环境科技服务有限公司 | Real-time VR video detection method for garden monitoring |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11615559B2 (en) | Methods and systems for human imperceptible computerized color transfer | |
| CN111488756B (en) | Face recognition-based living body detection method, electronic device, and storage medium | |
| CN108537743A (en) | A kind of face-image Enhancement Method based on generation confrontation network | |
| CN110689599B (en) | 3D visual saliency prediction method based on non-local enhancement generation countermeasure network | |
| CN110287849A (en) | A kind of lightweight depth network image object detection method suitable for raspberry pie | |
| CN111667400B (en) | Human face contour feature stylization generation method based on unsupervised learning | |
| CN105787930B (en) | The conspicuousness detection method and system for virtualization image based on sharpness | |
| CN107463968A (en) | Smog judges the detection method of the production method of code book, generation system and smog | |
| CN106096603A (en) | A kind of dynamic flame detection method merging multiple features and device | |
| CN109948566A (en) | A kind of anti-fraud detection method of double-current face based on weight fusion and feature selecting | |
| CN108470178B (en) | Depth map significance detection method combined with depth credibility evaluation factor | |
| Messai et al. | Adaboost neural network and cyclopean view for no-reference stereoscopic image quality assessment | |
| CN114067444A (en) | Face spoofing detection method and system based on meta-pseudo label and illumination invariant feature | |
| CN110853027A (en) | Three-dimensional synthetic image no-reference quality evaluation method based on local variation and global variation | |
| CN111681177A (en) | Video processing method and device, computer readable storage medium and electronic equipment | |
| CN111666852A (en) | Micro-expression double-flow network identification method based on convolutional neural network | |
| CN111784658A (en) | Quality analysis method and system for face image | |
| CN112734747B (en) | Target detection method and device, electronic equipment and storage medium | |
| CN112989958A (en) | Helmet wearing identification method based on YOLOv4 and significance detection | |
| CN106446764B (en) | Video object detection method based on improved fuzzy color aggregated vector | |
| Ying et al. | Region-aware RGB and near-infrared image fusion | |
| CN114973246A (en) | Crack detection method of cross mode neural network based on optical flow alignment | |
| Li et al. | Infrared and visible image fusion method based on principal component analysis network and multi-scale morphological gradient | |
| CN114449362A (en) | Video cover selecting method, device, equipment and storage medium | |
| CN108537771A (en) | MC-SILTP moving target detecting methods based on HSV |
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: 20171212 |
|
| RJ01 | Rejection of invention patent application after publication |