CN108648409A - A kind of smog detection method and device - Google Patents
A kind of smog detection method and device Download PDFInfo
- Publication number
- CN108648409A CN108648409A CN201810396544.2A CN201810396544A CN108648409A CN 108648409 A CN108648409 A CN 108648409A CN 201810396544 A CN201810396544 A CN 201810396544A CN 108648409 A CN108648409 A CN 108648409A
- Authority
- CN
- China
- Prior art keywords
- smog
- pixel
- image
- gradient
- video image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 56
- 238000000034 method Methods 0.000 claims abstract description 43
- 230000005540 biological transmission Effects 0.000 claims abstract description 31
- 238000012360 testing method Methods 0.000 claims abstract description 13
- 230000002146 bilateral effect Effects 0.000 claims abstract description 11
- 238000001914 filtration Methods 0.000 claims abstract description 11
- 239000003595 mist Substances 0.000 claims abstract description 10
- 230000009467 reduction Effects 0.000 claims abstract description 6
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 claims description 6
- 230000002708 enhancing effect Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000009738 saturating Methods 0.000 claims 1
- 238000012545 processing Methods 0.000 abstract description 7
- 230000033001 locomotion Effects 0.000 description 8
- 238000003384 imaging method Methods 0.000 description 7
- 239000000779 smoke Substances 0.000 description 7
- 239000002245 particle Substances 0.000 description 4
- 235000019504 cigarettes Nutrition 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000036425 denaturation Effects 0.000 description 1
- 238000004925 denaturation Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000035484 reaction time Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B17/00—Fire alarms; Alarms responsive to explosion
- G08B17/12—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions
- G08B17/125—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions by using a video camera to detect fire or smoke
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to technical field of image processing more particularly to a kind of smog detection methods, including：Enhance collected video image using dark channel prior method and bilateral filtering method；According to enhanced video image and atmospherical scattering model, the transmission function of video image is obtained；The transmission function obtained using the detection of ViBe methods, obtains smog candidate region；According to ViBe method testing results, Gradient Reconstruction, the background area of reduction smog candidate region covering are carried out to enhanced video image；For each pixel of Gradient Reconstruction back scene area, the similarity degree of the pixel and respective pixel in the background area of enhanced video image is sought, when similarity degree is more than preset similar threshold value, judges the pixel for smog pixel.The invention further relates to a kind of mist detecting devices.Smog detection method and device detection provided by the present invention is fast and accuracy rate is high, is not only able to distinguish smog and other moving targets, and can distinguish smog and shade.
Description
Technical field
The present invention relates to technical field of image processing more particularly to a kind of smog detection methods and device.
Background technology
Fire is the one of the major reasons for threatening social development and personnel safety.Traditional fire detection means are by right
The particle of fire burning is differentiated.However the place openr in space, fire particle reach detector examination criteria
Time can greatly increase, and thus be extremely difficult to the purpose of fire alarm.With the Pattern recognition and image processing based on image
Development.Fire detection based on video image has been the research direction of a mainstream.When fire occurs, often first generate
Smog, so the Smoke Detection based on video image is the main means of fire alarm.Currently, many cigarettes based on video image
Mist detection method usually can not accurately detect location of smoke, be vulnerable to the interference of the environmental factors such as shade, or the efficiency of detection is low,
Reaction time is slow, can not achieve timely and effective fire alarm.
Invention content
(1) technical problems to be solved
It is low the technical problem to be solved by the present invention is to solve the smog detection method accuracy rate based on image in the prior art,
Detection speed is slow, it is difficult to the problem of obtaining ideal testing result.
(2) technical solution
In order to solve the above technical problem, the present invention provides a kind of smog detection methods, including：
S1, enhance collected video image using dark channel prior method and bilateral filtering method；
S2, according to enhanced video image and atmospherical scattering model, obtain the transmission function of video image；
S3, the transmission function obtained using the detection of ViBe methods, obtain smog candidate region；
S4, according to ViBe method testing results, gradient weight is carried out to the smog candidate region in enhanced video image
It builds, the background area of reduction smog candidate region covering；
S5, for each pixel of Gradient Reconstruction back scene area, ask the pixel to carry out enhanced video with step S1
The similarity degree of respective pixel in the background area of image judges institute when the similarity degree is more than preset similar threshold value
It is smog pixel to state pixel, is otherwise non-smog pixel.
Preferably, when carrying out Gradient Reconstruction to the smog candidate region in video image in the step S4, using as follows
Formula obtain video image in the horizontal direction with the gradient image of vertical direction：
Gh(m, n)=p (m+1, n)-p (m, n)；
Gv(m, n)=p (m, n+1)-p (m, n)；
Wherein, Gh(m, n) indicates the gradient image of horizontal direction, Gv(m, n) indicates the gradient image of vertical direction, (m, n)
Indicate that the point in video image, p (m, n) indicate the pixel value of point (m, n)；
By following formula restore video image in the horizontal direction with the gradient of vertical direction：
Gh-estimate(x, y)=k (p (x+1, y)-p (x, y))；
Gv-estimate(x, y)=k (p (x, y+1)-p (x, y))；
Wherein, Gh-estimate(x, y) indicates that horizontal direction moves up the gradient estimation of moving-target, Gv-estimate(x, y) indicates perpendicular
Histogram moves up the gradient estimation of target, and (x, y) indicates that the point in video image, p (x, y) indicate the pixel value of point (x, y),
K indicates the multiplying factor in gradient.
Preferably, the ranging from 1.3-1.5 of the multiplying factor k in the gradient.
Preferably, in the step S4, according to Gh-estimate(x, y) and Gv-estimate(x, y) obtains Graded factor G'(x,
Y), by solving Poisson's equation from Graded factor G'(x, y) in obtain image I'(x, y after Gradient Reconstruction), I'(x, y) and G'
The relationship of (x, y) meets following formula：
Wherein, ▽2Indicate Laplce's factor, I'(x, y) indicate that the image after Gradient Reconstruction, (x, y) indicate video figure
Point as in；Graded factor G'(x, y) expression formula be G'(x, y)=[Gh-estimate(x,y),Gv-estimate(x,y)]。
Preferably, when solving Poisson's equation in the step S4, the approximation of Poisson's equation is solved using quick Poisson solution
Solution, with sine transform replace Laplce's factor, I'(x, y) and G'(x, y) relationship meet following formula：
SinI'(x, y)=div (G'(x, y))=div ([Gh-estimate(x,y),Gv-estimate(x,y)])；
Wherein, sin indicates sine transform Laplce's factor, I'(x, y) indicate the image after Gradient Reconstruction, (x, y) table
Show the point in video image.
Preferably, it when solving Poisson's equation in the step S4, is solved using Dirichlet boundary conditions, in all directions
Gray value is filled with zero.
Preferably, the step S5 includes：
With I'(x, y) indicate the image after rebuilding, in the background area after Gradient Reconstruction each pixel (x,
Y), select a size for the block m1 of 3 × 3 pixels, center is pixel (x, y)；In the image for not carrying out Gradient Reconstruction,
Block m2, the similitude between calculation block m1 and block m2 are chosen in identical position in the picture frame of arest neighbors：
Wherein, r (x, y) indicates the Gradient Reconstruction image of pixel (x, y) and does not carry out the Background regional image of Gradient Reconstruction
The similarity degree of picture, M × N indicate that the size of video image, E (m1) indicate that the mean value of block m1, E (m2) indicate the mean value of block m2；
Setting similar threshold value threhold judges that the pixel (x, y) is if similarity degree r (x, y) ＞ threhold
Smog pixel.
Preferably, in the step S5, the picture frame of arest neighbors meets the following conditions：
At position (x, y), the pixel in block m2 is background area pixels, and the time difference between the frame and present frame is most
It is small.
Preferably, in the step S5, similar threshold value threhold is 0.9.
The present invention also provides a kind of mist detecting devices, including：
Image enhancement module, for enhancing collected video figure using dark channel prior method and bilateral filtering method
Picture；
Transmission function acquisition module, for according to enhanced video image and atmospherical scattering model, obtaining video image
Transmission function；
Moving object detection module, the transmission function for being obtained using the detection of ViBe methods, obtains smog candidate region；
Gradient Reconstruction module is used for according to ViBe method testing results, candidate to the smog in enhanced video image
Region carries out Gradient Reconstruction, the background area of reduction smog candidate region covering；
Smog determination module is sought the pixel and is only increased for each pixel for Gradient Reconstruction back scene area
The similarity degree of respective pixel in strong background area, when the similarity degree is more than preset similar threshold value, described in judgement
Pixel is smog pixel.
(3) advantageous effect
The above-mentioned technical proposal of the present invention has the following advantages that：The present invention provides a kind of cigarettes based on Computer Vision
Mist detection method enhances collected video image first with dark channel prior method and bilateral filtering method, removes environment
The interference of factor obtains clear-cut video image.Then it according to enhanced video image and atmospherical scattering model, obtains
The transmission function of video image, what transmission function reflected is the transmissivity between target context and imaging sensor, and is transmitted
Function only depends on the depth information of scene.According to the motion feature of smog, the movement of moving target is considered as target context and is arrived
The variation of transmissivity between imaging sensor, the transmission function obtained using the detection of ViBe methods, detects the target of movement,
Include smog and other moving targets.According to the translucent characteristics of smog, Gradient Reconstruction is carried out to video image, restores smoke
The background area covered by smog in mist candidate region.Background area after finally asking Gradient Reconstruction to restore and only enhanced
The similitude of background area, if similitude is higher, then it is assumed that the region is smoke region.This method avoid directly utilize
Result is inaccurate caused by Vibe methods detect original video, and the factors such as variation of illumination are to motion detection result in elimination video
Influence, and the semi-transparency property that smog is utilized distinguishes smog and other moving targets, is a kind of highly effective and handles
The object detection method of speed quickly.
The present invention also provides a kind of mist detecting devices based on Computer Vision, including：Image enhancement module, thoroughly
Penetrate function acquisition module, moving object detection module, Gradient Reconstruction module and smog determination module.The device is not only able to a cigarette
Mist is distinguished apparently with other moving targets, and smog and shade can be distinguished.
Description of the drawings
Fig. 1 is smog detection method block diagram in the embodiment of the present invention；
Fig. 2 is one of smog detection method testing result figure in the embodiment of the present invention；
Fig. 3 is two of smog detection method testing result figure in the embodiment of the present invention；
Fig. 4 is three of smog detection method testing result figure in the embodiment of the present invention；
Fig. 5 is four of smog detection method testing result figure in the embodiment of the present invention；
Fig. 6 is mist detecting device structural schematic diagram in the embodiment of the present invention.
In figure：100：Image enhancement module；200：Transmission function acquisition module；300：Moving object detection module；400：
Gradient Reconstruction module；500：Smog determination module.
Specific implementation mode
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
The every other embodiment that member is obtained without making creative work, shall fall within the protection scope of the present invention.
As shown in Figure 1, a kind of smog detection method provided in an embodiment of the present invention, including：
S1, enhance collected video image using dark channel prior method and bilateral filtering method.
Collected video image is filtered by dark channel prior method and bilateral filtering method, can enhance and regard
Frequency image so that the contour of object in video image is apparent, excludes the factors such as ambient light and is interfered caused by video image, just
In subsequent processing work.Since sky areas is partially bright partially white, dark channel prior method can be according to the distribution of gray value, accurately
Estimate sky areas.Dark channel prior method and bilateral filtering method are state of the art, and details are not described herein.
S2, according to enhanced video image and atmospherical scattering model, obtain the transmission function of video image.
In computer vision and computer picture, atmospherical scattering model is widely used described in following equations：
I (x, y)=A ρ (x, y) e-βd(x,y)+A(1-e-βd(x,y))；
Wherein, I (x, y) indicates that the pixel value of pixel (x, y), A indicate that atmosphere light, β indicate atmospheric scattering coefficient, ρ (x, y)
Indicate that the scene albedo at pixel (x, y), d (x, y) indicate the depth of field at pixel (x, y), the wherein transmission at pixel (x, y)
Rate t (x, y)=1-e-βd(x,y), unobstructed natural image J (x, y)=A ρ (x, y) seen by person.
First item J (x, y) (1-t (x, y)) on the right of equation is called direct attenuation term.Since the scattering of atmospheric particles is made
With the part in target surface reflected light is lost because of scattering, and unscattered part directly reaches imaging sensor, is reached
Light intensity exponentially decay with the increase of propagation distance.Section 2 At (x, y) is then atmosphere light ingredient, this is because air
Particle causes air to show the characteristic of light source the scattering of natural light.
Atmospherical scattering model illustrates the imaging mechanism that smog is penetrated in object.It is according to the meaning analysis of model it is found that anti-
The atmospheric transmissivity transmission function t (x, y) reflected between target context and imaging sensor only depends on the depth information (scape of scene
It is deep) it is influenced without being shone by atmosphere light.
Enhancing treated video image will be filtered in step S1 as J (x, y), imaging sensor is collected to be regarded
Frequency image is I (x, y), you can finds out transmission function t (x, y).
S3, the transmission function obtained using the detection of ViBe methods, obtain smog candidate region.
ViBe is the object detection method of a kind of highly effective and speed quickly, but because illumination in video image
The factors such as variation are difficult directly then to obtain ideal motion detection result to the processing of collected original video.
According to atmospherical scattering model, transmission function only depends on the depth of field, and the movement of moving target can also be considered as background
Variation of the region to transmissivity between imaging sensor, it is possible to by carrying out Vibe motion detections to transmission function, obtain
Ideal result.
Present invention utilizes the kinetic characteristic of smog, other moving objects are influenced in original scene in smog and scene
The basic reason of transmission function.Moving object detection is carried out using ViBe methods to transmission function, detects the foreground moved
Target, i.e. detection obtain smog candidate region, have both included smog in smog candidate region, and have also included other moving objects.Afterwards
Smog and other moving objects are distinguished in continuous processing.
S4, according to ViBe method testing results, gradient weight is carried out to the smog candidate region in enhanced video image
It builds, the background area of reduction smog candidate region covering.
The target that can be needed, including smog and some other targets are detected using ViBe methods, since smog has
There is the characteristic of translucence, if Gradient Reconstruction is accurately estimated in horizontal and vertical direction, smog can be moved from other
It is separated in target.
Video image is divided into horizontal direction and vertical direction in Gradient Reconstruction.Gradient Reconstruction needs three width images, waits for weight
The gradient image of the estimation of video image, horizontal direction and the vertical direction built.It, can profit when carrying out Gradient Reconstruction to video image
With following formula obtain video image in the horizontal direction with the gradient image of vertical direction：
Gh(m, n)=p (m+1, n)-p (m, n)；
Gv(m, n)=p (m, n+1)-p (m, n)；
Wherein, Gh(m, n) indicates the gradient image of horizontal direction, Gv(m, n) indicates the gradient image of vertical direction, (m, n)
Indicate that the point in video image, p (m, n) indicate the pixel value of point (m, n).
It can be regarded by the foundation of following formula by grad enhancement by the background area of smog covering in smog candidate region
Frequency image in the horizontal direction with the gradient map of vertical direction：
Gh-estimate(x, y)=k (p (x+1, y)-p (x, y))；
Gv-estimate(x, y)=k (p (x, y+1)-p (x, y))；
Wherein, Gh-estimate(x, y) indicates that horizontal direction moves up the gradient estimation of moving-target, Gv-estimate(x, y) indicates perpendicular
Histogram moves up the gradient estimation of target, and (x, y) indicates that the point of enhanced video image J (x, y), p (x, y) indicate point
The pixel value of (x, y), k indicate the multiplying factor in gradient.
Preferably, the value range of k is 1.3-1.5.It is further preferred that in the present embodiment, the multiplying factor k in gradient
=1.4 so that the gradient of mobile target is enhanced, but other parts do not change.
Gradient Reconstruction is carried out to the result after ViBe moving object detections.According to Gh-estimate(x, y) and Gv-estimate(x,
Y) Graded factor G'(x, y are obtained), on the directions 2D, improved Graded factor G'(x, y) expression formula be G'(x, y)=
[Gh-estimate(x,y),Gv-estimate(x, y)], Graded factor G'(x, y) and uncorrelated, I'(x, y) represent from G'(x, y) in obtain
The Gradient Reconstruction obtained, I'(x, y) by going to determine based on the method for Laplce and lock out operation, Poisson's equation is frequently used
In computer vision, it to solve closer to practical situation in the problems in practical problem.By solving Poisson side
Journey is from Graded factor G'(x, y) in obtain image I'(x, y after Gradient Reconstruction), I'(x, y) and G'(x, y) relationship meet such as
Lower formula：
Wherein, ▽2Indicate Laplce's factor, I'(x, y) indicate that the image after Gradient Reconstruction, (x, y) indicate video figure
Point as in.It can obtain being similar to linear solution of equation, Laplce's factor ▽ using above-mentioned formula2With div linear operations,
When solving Poisson's equation, in order to improve efficiency, accelerate the processing speed of video image, Poisson side is solved using quick Poisson solution
The approximate solution of journey, with sine transform replace Laplce's factor, I'(x, y) and G'(x, y) relationship meet following formula：
SinI'(x, y)=div (G'(x, y))=div ([Gh-estimate(x,y),Gv-estimate(x,y)])；
Wherein, sin indicates sine transform Laplce's factor, I'(x, y) indicate the image after Gradient Reconstruction, (x, y) table
Show the point in video image.Replacing Laplce's factor using sine transform enables to the complexity of operation to be increased to O (n
(log(n)))。
It is further preferred that when solving Poisson's equation, solved instead of Newman boundary condition using Dirichlet boundary conditions
Reconstruction image I'(x, y), gray value is filled with zero in all directions, it can be to avoid the movement of gray level.
S5, for each pixel of Gradient Reconstruction back scene area, ask the pixel to carry out enhanced video with step S1
The similarity degree of respective pixel in the background area of image judges institute when the similarity degree is more than preset similar threshold value
It is smog pixel to state pixel, is otherwise non-smog pixel.
Present invention utilizes the semi-transparency properties of smog to distinguish smog and other moving targets.By after reconstruction image with
The background area that step S1 carries out enhanced video image is compared, and is considered smog part if similarity is very high.
Preferably, step S5 includes：
With I'(x, y) indicate the image after rebuilding, in the background area after Gradient Reconstruction each pixel (x,
Y), select a size for the block m1 of 3 × 3 pixels, center is pixel (x, y)；In the image for not carrying out Gradient Reconstruction,
Block m2, the similitude between calculation block m1 and block m2 are chosen in identical position in the picture frame of arest neighbors：
Wherein, r (x, y) indicates pixel (x, y) Gradient Reconstruction image and does not carry out the background area image of Gradient Reconstruction
Similarity degree, M × N indicates that the size of whole video image, E (m1) indicate that the mean value of block m1, E (m2) indicate that block m2's is equal
Value, if by formula it is found that m1=m2, r value are 1.Similar threshold value threhold is set, if similarity degree r (x, y) ＞
Threhold then judges that the pixel (x, y) is smog pixel.Preferably, similar threshold value threhold is in the present embodiment
0.9。
Wherein, the picture frame of arest neighbors meets the following conditions：First is the frame at position (x, y), the pixel in block m2
It is background area pixels, second is that the frame and present frame are nearest, i.e., the time difference between the frame and present frame is minimum.
The present invention realizes above-mentioned smog detection method using MATLAB programmings, and has passed through under four kinds of different scenes
Test, as shown in Figures 2 to 5, Fig. 2 are the scene that the first only has smog, and Fig. 3 has smog and other moving targets for second
Scene, Fig. 4 be the third to have the scene of smog and shade, Fig. 5 be the 4th kind of scene for having smog, shade and moving target,
The smoke region detected is marked in Fig. 2 to Fig. 5, as shown in gray patches part in figure.
Smoke Detection result from Fig. 2 to Fig. 5 can be seen that smog detection method provided in this embodiment and be not only able to handle
Smog is distinguished apparently with other moving targets, and smog and shade can be distinguished.
As shown in fig. 6, a kind of mist detecting device is additionally provided in the present embodiment, including：Image enhancement module 100, thoroughly
Penetrate function acquisition module 200, moving object detection module 300, Gradient Reconstruction module 400 and smog determination module 500.Wherein：
Image enhancement module 100 is used to enhance collected video figure using dark channel prior method and bilateral filtering method
Picture；
Transmission function acquisition module 200 is used to, according to enhanced video image and atmospherical scattering model, obtain video figure
The transmission function of picture；
The transmission function that moving object detection module 300 is used to obtain using the detection of ViBe methods, obtains smog candidate regions
Domain；
Gradient Reconstruction module 400 is used to, according to ViBe method testing results, wait the smog in enhanced video image
Favored area carries out Gradient Reconstruction, the background area of reduction smog candidate region covering；
Smog determination module 500 is used for each pixel for Gradient Reconstruction back scene area, seeks the pixel and only carries out
The similarity degree of respective pixel in the background area of enhancing judges institute when the similarity degree is more than preset similar threshold value
It is smog pixel to state pixel.
In conclusion the present invention provides a kind of smog detection method and device, video image can be based on and carry out smog inspection
It surveys.This method and device are based on atmospherical scattering model, and transmission function is calculated using dark channel prior information and bilateral filtering, and
Obtain the gray value of sky.Smog is detected using transmission function and ViBe methods, the illumination dependent on transmission function is not
The kinetic characteristic of denaturation and smog detects transmission function using ViBe, can obtain smog candidate region.In order in smog candidate
Accurate location of smoke is found in region, and image restoration is carried out to image using Poisson's equation to smog candidate region, is then compared
Compared with the image after recovery and the similitude between former background, and then determine the accurate location of smog.It is not only able to smog and its
His moving target is distinguished apparently, and smog and shade can be distinguished.
Finally it should be noted that：The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, it will be understood by those of ordinary skill in the art that：It still may be used
With technical scheme described in the above embodiments is modified or equivalent replacement of some of the technical features；
And these modifications or replacements, various embodiments of the present invention technical solution that it does not separate the essence of the corresponding technical solution spirit and
Range.
Claims (10)
1. a kind of smog detection method, which is characterized in that including：
S1, enhance collected video image using dark channel prior method and bilateral filtering method；
S2, according to enhanced video image and atmospherical scattering model, obtain the transmission function of video image；
S3, the transmission function obtained using the detection of ViBe methods, obtain smog candidate region；
S4, the smog candidate region progress Gradient Reconstruction in enhanced video image is gone back according to ViBe method testing results
The background area of former smog candidate region covering；
S5, for each pixel of Gradient Reconstruction back scene area, ask the pixel to carry out enhanced video image with step S1
Background area in respective pixel similarity degree, when the similarity degree be more than preset similar threshold value when, judge the picture
Element is smog pixel, is otherwise non-smog pixel.
2. smog detection method according to claim 1, which is characterized in that
When carrying out Gradient Reconstruction to the smog candidate region in video image in the step S4, video is obtained using following formula
Image in the horizontal direction with the gradient image of vertical direction：
Gh(m, n)=p (m+1, n)-p (m, n)；
Gv(m, n)=p (m, n+1)-p (m, n)；
Wherein, Gh(m, n) indicates the gradient image of horizontal direction, Gv(m, n) indicates that the gradient image of vertical direction, (m, n) indicate
Point in video image, p (m, n) indicate the pixel value of point (m, n)；
By following formula restore video image in the horizontal direction with the gradient of vertical direction：
Gh-estimate(x, y)=k (p (x+1, y)-p (x, y))；
Gv-estimate(x, y)=k (p (x, y+1)-p (x, y))；
Wherein, Gh-estimate(x, y) indicates that horizontal direction moves up the gradient estimation of moving-target, Gv-estimate(x, y) indicates vertical side
The gradient estimation of target is moved up, (x, y) indicates that the point in video image, p (x, y) indicate the pixel value of point (x, y), k tables
Show the multiplying factor in gradient.
3. smog detection method according to claim 2, which is characterized in that the range of the multiplying factor k in the gradient
For 1.3-1.5.
4. smog detection method according to claim 2, which is characterized in that in the step S4, according to Gh-estimate(x,
And G y)v-estimate(x, y) obtains Graded factor G'(x, y), by solving Poisson's equation from Graded factor G'(x, y) in obtain
Image I'(x, y after Gradient Reconstruction), I'(x, y) and G'(x, y) relationship meet following formula：
▽2I'(x, y)=div (G'(x, y))=div ([Gh-estimate(x,y),Gv-estimate(x,y)])；
Wherein, ▽2Indicate Laplce's factor, I'(x, y) indicate that the image after Gradient Reconstruction, (x, y) indicate in video image
Point；Graded factor G'(x, y) expression formula be G'(x, y)=[Gh-estimate(x,y),Gv-estimate(x,y)]。
5. smog detection method according to claim 4, it is characterised in that：When solving Poisson's equation in the step S4,
The approximate solution that Poisson's equation is solved using quick Poisson solution replaces Laplce's factor, I'(x, y with sine transform) and G'
The relationship of (x, y) meets following formula：
SinI'(x, y)=div (G'(x, y))=div ([Gh-estimate(x,y),Gv-estimate(x,y)])；
Wherein, sin indicates sine transform Laplce's factor, I'(x, y) indicate that the image after Gradient Reconstruction, (x, y) expression regard
Point in frequency image.
6. smog detection method according to claim 4 or 5, it is characterised in that：Poisson's equation is solved in the step S4
When, it is solved using Dirichlet boundary conditions, gray value is filled with zero in all directions.
7. smog detection method according to claim 1, which is characterized in that the step S5 includes：
With I'(x, y) indicate rebuild after image, for I'(x after Gradient Reconstruction, y) background area in each pixel
(x, y) selects a size for the block m1 of 3 × 3 pixels, and center is pixel (x, y)；In the image for not carrying out Gradient Reconstruction
In, block m2, the similitude between calculation block m1 and block m2 are chosen in identical position in the picture frame of arest neighbors：
Wherein, r (x, y) indicates pixel (x, y) Gradient Reconstruction image and does not carry out the phase of the background area image of Gradient Reconstruction
Like degree, M × N indicates that the size of video image, E (m1) indicate that the mean value of block m1, E (m2) indicate the mean value of block m2；
Similar threshold value threhold is set, if similarity degree r (x, y) ＞ threhold, judges that the pixel (x, y) is smog
Pixel.
8. smog detection method according to claim 7, which is characterized in that in the step S5, the picture frame of arest neighbors
Meet the following conditions：
At position (x, y), the pixel in block m2 is background area pixels, and the time difference between the frame and present frame is minimum.
9. according to claim 7 or 8 any one of them smog detection methods, which is characterized in that in the step S5, similar threshold
Value threhold is 0.9.
10. a kind of mist detecting device, which is characterized in that including：
Image enhancement module, for enhancing collected video image using dark channel prior method and bilateral filtering method；
Transmission function acquisition module, for according to enhanced video image and atmospherical scattering model, obtaining the saturating of video image
Penetrate function；
Moving object detection module, the transmission function for being obtained using the detection of ViBe methods, obtains smog candidate region；
Gradient Reconstruction module is used for according to ViBe method testing results, to the smog candidate region in enhanced video image
Carry out Gradient Reconstruction, the background area of reduction smog candidate region covering；
Smog determination module is sought the pixel and is only enhanced for each pixel for Gradient Reconstruction back scene area
The similarity degree of respective pixel in background area judges the pixel when the similarity degree is more than preset similar threshold value
For smog pixel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810396544.2A CN108648409B (en) | 2018-04-28 | 2018-04-28 | Smoke detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810396544.2A CN108648409B (en) | 2018-04-28 | 2018-04-28 | Smoke detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108648409A true CN108648409A (en) | 2018-10-12 |
| CN108648409B CN108648409B (en) | 2020-07-24 |
Family
ID=63748100
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810396544.2A Active CN108648409B (en) | 2018-04-28 | 2018-04-28 | Smoke detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108648409B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109522819A (en) * | 2018-10-29 | 2019-03-26 | 西安交通大学 | A kind of fire image recognition methods based on deep learning |
| CN109978876A (en) * | 2019-04-03 | 2019-07-05 | 北京环境特性研究所 | A kind of smog recognition methods and device based on quick bilateral filtering |
| CN116977327A (en) * | 2023-09-14 | 2023-10-31 | 山东拓新电气有限公司 | Smoke detection method and system for roller-driven belt conveyor |
Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003281644A (en) * | 2002-03-26 | 2003-10-03 | Nohmi Bosai Ltd | Extinction type smoke sensor |
| US20070165966A1 (en) * | 2005-07-15 | 2007-07-19 | Yissum Research Development Co. | Closed form method and system for matting a foreground object in an image having a background |
| CN101944267A (en) * | 2010-09-08 | 2011-01-12 | 大连古野软件有限公司 | Smoke and fire detection device based on videos |
| CN102013009A (en) * | 2010-11-15 | 2011-04-13 | 无锡中星微电子有限公司 | Smoke image recognition method and device |
| JP2011215806A (en) * | 2010-03-31 | 2011-10-27 | Nohmi Bosai Ltd | Smoke detection device |
| US20120038644A1 (en) * | 2009-02-17 | 2012-02-16 | Uws Ventures Ltd. | Rendering caustics in computer graphics with photon mapping by relaxation |
| CN102609710A (en) * | 2012-02-22 | 2012-07-25 | 武汉大学 | Smoke and fire object segmentation method aiming at smog covering scene in fire disaster image video |
| CN102930514A (en) * | 2012-09-27 | 2013-02-13 | 西安电子科技大学 | Rapid image defogging method based on atmospheric physical scattering model |
| CN103489166A (en) * | 2013-10-12 | 2014-01-01 | 大连理工大学 | Bilateral filter-based single image defogging method |
| WO2014179482A1 (en) * | 2013-04-30 | 2014-11-06 | The Regents Of The University Of California | Fire urgency estimator in geosynchronous orbit (fuego) |
| CN104700365A (en) * | 2015-02-02 | 2015-06-10 | 电子科技大学 | Image contrast enhancing method |
| CN104751591A (en) * | 2015-04-01 | 2015-07-01 | 大连希尔德安全技术有限公司 | Early fire smoke detection method based on semi transparent feature |
| CN104807611A (en) * | 2015-05-04 | 2015-07-29 | 中国科学技术大学 | Flue gas velocity field and turbulence field experimental measurement device and method based on video |
| JP2015197788A (en) * | 2014-04-01 | 2015-11-09 | 能美防災株式会社 | Laminar flow smoke detection device and laminar flow smoke detection method |
| DE102014219838A1 (en) * | 2014-09-30 | 2016-03-31 | Robert Bosch Gmbh | Smoke detection apparatus and method for detecting smoke of a fire and computer program |
| CN107045723A (en) * | 2017-03-10 | 2017-08-15 | 北京环境特性研究所 | Smog recognition methods based on transmissivity dynamic detection |
| CN107146209A (en) * | 2017-05-02 | 2017-09-08 | 四川大学 | A kind of single image to the fog method based on gradient field |
| CN107527329A (en) * | 2017-09-19 | 2017-12-29 | 重庆邮电大学 | A kind of image defogging method based on dark channel prior and Markov random field |
| CN107749067A (en) * | 2017-09-13 | 2018-03-02 | 华侨大学 | Fire hazard smoke detecting method based on kinetic characteristic and convolutional neural networks |
| KR20180059612A (en) * | 2016-11-25 | 2018-06-05 | 포테닛 주식회사 | Image apparatus for fire detection |
-
2018
- 2018-04-28 CN CN201810396544.2A patent/CN108648409B/en active Active
Patent Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003281644A (en) * | 2002-03-26 | 2003-10-03 | Nohmi Bosai Ltd | Extinction type smoke sensor |
| US20070165966A1 (en) * | 2005-07-15 | 2007-07-19 | Yissum Research Development Co. | Closed form method and system for matting a foreground object in an image having a background |
| US20120038644A1 (en) * | 2009-02-17 | 2012-02-16 | Uws Ventures Ltd. | Rendering caustics in computer graphics with photon mapping by relaxation |
| JP2011215806A (en) * | 2010-03-31 | 2011-10-27 | Nohmi Bosai Ltd | Smoke detection device |
| CN101944267A (en) * | 2010-09-08 | 2011-01-12 | 大连古野软件有限公司 | Smoke and fire detection device based on videos |
| CN102013009A (en) * | 2010-11-15 | 2011-04-13 | 无锡中星微电子有限公司 | Smoke image recognition method and device |
| CN102609710A (en) * | 2012-02-22 | 2012-07-25 | 武汉大学 | Smoke and fire object segmentation method aiming at smog covering scene in fire disaster image video |
| CN102930514A (en) * | 2012-09-27 | 2013-02-13 | 西安电子科技大学 | Rapid image defogging method based on atmospheric physical scattering model |
| WO2014179482A1 (en) * | 2013-04-30 | 2014-11-06 | The Regents Of The University Of California | Fire urgency estimator in geosynchronous orbit (fuego) |
| CN103489166A (en) * | 2013-10-12 | 2014-01-01 | 大连理工大学 | Bilateral filter-based single image defogging method |
| JP2015197788A (en) * | 2014-04-01 | 2015-11-09 | 能美防災株式会社 | Laminar flow smoke detection device and laminar flow smoke detection method |
| DE102014219838A1 (en) * | 2014-09-30 | 2016-03-31 | Robert Bosch Gmbh | Smoke detection apparatus and method for detecting smoke of a fire and computer program |
| CN104700365A (en) * | 2015-02-02 | 2015-06-10 | 电子科技大学 | Image contrast enhancing method |
| CN104751591A (en) * | 2015-04-01 | 2015-07-01 | 大连希尔德安全技术有限公司 | Early fire smoke detection method based on semi transparent feature |
| CN104807611A (en) * | 2015-05-04 | 2015-07-29 | 中国科学技术大学 | Flue gas velocity field and turbulence field experimental measurement device and method based on video |
| KR20180059612A (en) * | 2016-11-25 | 2018-06-05 | 포테닛 주식회사 | Image apparatus for fire detection |
| CN107045723A (en) * | 2017-03-10 | 2017-08-15 | 北京环境特性研究所 | Smog recognition methods based on transmissivity dynamic detection |
| CN107146209A (en) * | 2017-05-02 | 2017-09-08 | 四川大学 | A kind of single image to the fog method based on gradient field |
| CN107749067A (en) * | 2017-09-13 | 2018-03-02 | 华侨大学 | Fire hazard smoke detecting method based on kinetic characteristic and convolutional neural networks |
| CN107527329A (en) * | 2017-09-19 | 2017-12-29 | 重庆邮电大学 | A kind of image defogging method based on dark channel prior and Markov random field |
Non-Patent Citations (4)
| Title |
|---|
| QIANG ZHANG等: "Dissipation Function and ViBe Based Smoke Detection in Video", 《2017 2ND INTERNATIONAL CONFERENCE ON MULTIMEDIA AND IMAGE PROCESSING》 * |
| 张汉营等: ""基于暗通道和小波的单幅图像烟雾检测算法"", 《计算机系统应用》 * |
| 杨勇: "基于求解泊松方程和梯度的图像修复的研究", 《计算机技术与发展》 * |
| 王利娟: ""视频监控中火灾烟雾探测方法研究"", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109522819A (en) * | 2018-10-29 | 2019-03-26 | 西安交通大学 | A kind of fire image recognition methods based on deep learning |
| CN109978876A (en) * | 2019-04-03 | 2019-07-05 | 北京环境特性研究所 | A kind of smog recognition methods and device based on quick bilateral filtering |
| CN116977327A (en) * | 2023-09-14 | 2023-10-31 | 山东拓新电气有限公司 | Smoke detection method and system for roller-driven belt conveyor |
| CN116977327B (en) * | 2023-09-14 | 2023-12-15 | 山东拓新电气有限公司 | Smoke detection method and system for roller-driven belt conveyor |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108648409B (en) | 2020-07-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107025652B (en) | A kind of flame detecting method based on kinetic characteristic and color space time information | |
| CA2779795C (en) | Inclusion detection in polished gemstones | |
| Zhang et al. | Matched filtering algorithm for pavement cracking detection | |
| US6704447B2 (en) | Method and apparatus for using illumination from a display for computer vision based user interfaces and biometric authentication | |
| US6728593B2 (en) | System for analysis of fabric surface | |
| CN108648409A (en) | A kind of smog detection method and device | |
| CN106600625A (en) | Image processing method and device for detecting small-sized living thing | |
| CN110110675A (en) | A kind of wavelet field of combination of edge information divides shape infrared cirrus detection method | |
| CN109084350A (en) | A kind of kitchen ventilator and oil smoke concentration detection method having filtering functions vision-based detection module | |
| Sun et al. | Fusion of infrared and visible images for remote detection of low-altitude slow-speed small targets | |
| CN108760590A (en) | A kind of kitchen fume Concentration Testing based on image procossing and interference elimination method | |
| CN107045723A (en) | Smog recognition methods based on transmissivity dynamic detection | |
| CN111239142A (en) | Paste appearance defect detection device and method | |
| Kropp et al. | Drywall state detection in image data for automatic indoor progress monitoring | |
| US20190392601A1 (en) | Image Processing System for Inspecting Object Distance and Dimensions Using a Hand-Held Camera with a Collimated Laser | |
| Žuži et al. | Impact of dehazing on underwater marker detection for augmented reality | |
| CN106611165B (en) | A kind of automotive window detection method and device based on correlation filtering and color-match | |
| CN114049490A (en) | Intelligent occupational health early warning method and system | |
| CN110211112A (en) | A kind of casting defect inspection method based on filtering selective search | |
| CN109409285A (en) | Remote sensing video object detection method based on overlapping slice | |
| Batchelor | Interactive image analysis as a prototyping tool for industrial inspection | |
| CN107980094A (en) | A kind of vision detection system and method | |
| Husain et al. | VRHAZE: The simulation of synthetic haze based on visibility range for dehazing method in single image | |
| Dong et al. | A window detection algorithm for remote laser gas leakage detection system | |
| Cao | Detection of fruit surface defects based on machine vision |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |