CN114283367A - Artificial intelligent open fire detection method and system for garden fire early warning - Google Patents
Artificial intelligent open fire detection method and system for garden fire early warning Download PDFInfo
- Publication number
- CN114283367A CN114283367A CN202111606186.1A CN202111606186A CN114283367A CN 114283367 A CN114283367 A CN 114283367A CN 202111606186 A CN202111606186 A CN 202111606186A CN 114283367 A CN114283367 A CN 114283367A
- Authority
- CN
- China
- Prior art keywords
- image
- feature information
- pixel
- pixel values
- image area
- 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
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/80—Management or planning
- Y02P90/84—Greenhouse gas [GHG] management systems
Abstract
The invention provides an artificial intelligent open fire detection method and system for garden fire early warning, and belongs to the technical field of artificial intelligent detection. The method comprises the following steps: acquiring a video image by a garden environment control monitoring model; acquiring a differential image between a current frame video image and a reference frame video image; counting the distribution condition of the pixel values of all the pixel points in the differential image; if the distribution condition meets a preset condition, determining a judgment threshold according to the pixel values of the pixel points in the differential image; determining a first image area from the difference image; determining a second image area in the current frame video image according to the position of the first image area in the difference image; and carrying out open fire identification on the second image area by using an artificial intelligence model. And monitoring the environment management state and predicting the combustible condition through a designed park environment-controlled monitoring model. The invention can improve the efficiency and benefit of artificial intelligent open fire detection, find the energy control and environmental protection loopholes of the garden and improve the comprehensive carbon neutralization management capability.
Description
Technical Field
The invention relates to the technical field of artificial intelligent detection, in particular to an artificial intelligent open fire detection method and system for park fire early warning.
Background
With the rapid development of various industries, various industrial parks appear, and parkerization becomes an important development trend. One of important contents of the park safety of fire early warning is an important guarantee link of enterprise safety production in the park.
At present, fire safety early warning mainly adopts two modes of manual inspection and sensor identification and alarm. The area of garden is great usually, and the conflagration hidden danger is more and comparatively dispersed, adopts the manual work to patrol and examine, and is higher to patrolling and examining personnel's self quality requirement, makes mistakes misjudgement easily when the people under fatigue state, leads to the inspection not in place or omits the potential safety hazard, and it is lower to patrol and examine efficiency simultaneously. The sensor identification alarm comprises a smoke sensor identification alarm, an image sensor identification alarm and the like. The smoke sensor identification alarm mode can trigger alarm only when smoke with certain concentration exists at the position of the smoke alarm, so that certain requirements are imposed on the installation position and the density of the smoke alarm, the installation cost is high, and even if the smoke sensor identification alarm mode is densely installed, the alarm still cannot be triggered in time under the condition that the smoke is small or the smoke is influenced by the wind direction. The image sensor identifies the alarm mode, because the image information amount is large, and the alarm can be realized in time only by real-time processing, the calculated amount is large, the requirement on a processor is high, and in addition, the collected image is greatly influenced by the change of external illumination, and the identification precision is influenced.
Disclosure of Invention
Therefore, the technical problem to be solved by the embodiments of the present invention is to overcome the defects that the processing capacity of a fire alarm mode for identifying a fire by an image sensor is large and the identification precision is affected by external illumination changes in the prior art, so as to provide an artificial intelligent open fire detection method and system for fire early warning in a park.
Therefore, the invention provides an artificial intelligent open fire detection method for garden fire early warning, which comprises the following steps:
acquiring a video image acquired by an image acquisition device installed in a park;
acquiring a differential image between a current frame video image and a reference frame video image; the reference frame video image is a video frame image acquired by the image acquisition device before;
counting the distribution condition of the pixel values of all the pixel points in the differential image;
if the distribution condition meets a preset condition, determining a judgment threshold value according to the pixel value of the pixel point in the differential image;
determining a first image area from the differential image, wherein the first image area comprises all pixel points of which the pixel values are greater than the judgment threshold value in the differential image;
determining a second image area in the current frame video image according to the position of the first image area in the differential image;
and inputting the information obtained according to the second image area into an artificial intelligent model for open fire identification.
Optionally, the preset condition is that a difference between a first pixel value and a second pixel value is greater than a preset threshold, the first pixel value is a mean value or a median value of third pixel values in pixel values of each pixel point in the difference image, the second pixel value is a mean value or a median value of fourth pixel values in pixel values of each pixel point in the difference image, the third pixel value is a maximum N pixel values in pixel values of each pixel point in the difference image, the fourth pixel value is a minimum K pixel values in pixel values of each pixel point in the difference image, and N, K is a preset positive integer or is determined according to a preset proportion of all pixel values in the difference image.
Optionally, the determining a judgment threshold according to the pixel value of the pixel point in the difference image includes:
and determining the judgment threshold according to the first pixel value and the second pixel value.
Optionally, after the statistics of the distribution of the pixel values of each pixel point in the differential image, the method further includes:
if the distribution condition does not meet the preset condition, updating the reference frame video image into the current frame video image;
and if the distribution condition meets the preset condition, the reference frame video image is continuously used as the reference frame video image.
Optionally, the inputting the information obtained according to the second image area into an artificial intelligence model for open fire recognition includes:
extracting a plurality of aspects of feature information from the second image region respectively, wherein the plurality of aspects of feature information comprise spectral feature information, shape feature information and texture feature information;
aiming at the characteristic information of the aspects, respectively operating with corresponding characteristic offset matrixes trained in advance;
cascading the calculated feature information of the multiple aspects to obtain cascaded feature information;
carrying out normalization processing on the cascaded feature information by using a probability normalization algorithm to obtain normalized feature information;
carrying out further feature extraction on the normalized feature information by utilizing differential operation, and removing redundant feature information to obtain deep-level feature information;
and processing the deep-level feature information by using a principal component analysis algorithm to obtain information for inputting the information to the artificial intelligent model to identify open fire.
Optionally, the extracting feature information of multiple aspects from the second image region respectively includes:
traversing the second image area by using a sliding window with a preset size according to the sequence of first row and second row, and calculating the entropy of the image area in the sliding window during each sliding; the entropy of the image area in the sliding window during the current sliding is the gray level histogram of the image area in the sliding window during the last sliding minus the gray level probability of the pixel points in the first row and plus the gray level probability of the pixel points in the last row in the image area in the sliding window during the current sliding;
acquiring a texture image of the second image area according to the entropy of the image area in the sliding window during each sliding;
and extracting the texture feature information from the texture image.
Optionally, the extracting feature information of multiple aspects from the second image region respectively further includes:
extracting feature information of the second image area by using a VGG neural network;
the method comprises the steps of constructing a characteristic pyramid by a top-down method, carrying out up-sampling on a deep characteristic diagram with strong semantic characteristics in a bilinear interpolation mode to enable the deep characteristic diagram to be consistent with the size of a next layer of characteristic diagram, carrying out characteristic extraction on the up-sampled deep characteristic diagram by utilizing cavity convolution, adding the extracted deep characteristic diagram with the next layer of characteristic diagram to obtain a new characteristic diagram, continuously adding each obtained new characteristic diagram with the next layer of characteristic diagram in the same mode, and finally obtaining a depth heterogeneous characteristic diagram fusing a shallow characteristic diagram and the deep characteristic diagram as characteristic information of one aspect.
Optionally, the artificial intelligence model includes a multilayer neural network, and the multilayer neural network f is synthesized according to the following formula:
wherein M is 1,2,3, …, M, gm(x) Is the m-th neural network, amIs the weight of the mth neural network.
Optionally, before inputting the information obtained according to the second image area to an artificial intelligence model for open fire identification, the method further includes:
respectively constructing a plurality of the neural networks;
aiming at a plurality of neural networks, respectively training by using preset training samples;
training the weight a in the artificial intelligence model by using the preset training samplem(ii) a Wherein the weight a is used in trainingmThe iterative update is performed according to the following formula:
wherein, am(t+1)、am(t) represents the weight a after the t +1 th iteration and the t times of iteration updating respectivelymAnd E (t) is the identification error in the artificial intelligence model at the time of the t iteration update, btIs a parameter value determined according to the minimum recognition error in the artificial intelligence model in the previous t iteration updates, clThe weighted value of the ith sample is the value manually marked during the construction of the training sample, ol、ylThe target output value and the actual output value of the ith sample.
The invention also provides an artificial intelligent open fire detection system for garden fire early warning, which comprises:
one or more processors;
storage means for storing one or more programs;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement any of the above-described artificial intelligence open fire detection methods for campus fire early warning.
The technical scheme of the embodiment of the invention has the following advantages:
according to the artificial intelligent open fire detection method and system for fire early warning in a park, image changes are detected through the difference images between the video frame images collected successively, then open fire detection is carried out on the changed area in the video frame images by using the artificial intelligent model, compared with the process of directly carrying out open fire detection on the collected video frames by using the artificial intelligent model, the processing amount of the artificial intelligent open fire detection is reduced, and the problem that the open fire detection is not timely due to interval detection can be avoided. In addition, because the artificial intelligence model only detects the change area in the video image, the accuracy of open fire detection can be improved, and the probability of false alarm is reduced. In addition, in the embodiment of the invention, the judgment threshold value for judging the change area is determined according to the pixel value distribution condition of the pixel points of the difference image between the successively collected video frame images, so that the influence of the change area judgment caused by the change of external illumination (such as lightning, lamplight, sunlight and the like) and other factors can be reduced, and the accuracy of the change area judgment is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of a specific example of an artificial intelligence open fire detection method for campus fire early warning according to embodiment 1 of the present invention;
fig. 2 is a flowchart of a specific example of a feature extraction method for open fire recognition in embodiment 1 of the present invention;
fig. 3 is a flowchart of another specific example of the feature extraction method for open flame recognition in embodiment 1 of the present invention;
fig. 4 is a schematic block diagram of a specific example of an artificial intelligence open fire detection system for campus fire early warning in embodiment 2 of the present invention.
Detailed Description
The technical solutions of the present invention will be described clearly and completely with reference to the accompanying drawings, and it should be understood that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In describing the present invention, it is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. The terms "comprises" and/or "comprising," when used in this specification, are intended to specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. The term "and/or" includes any and all combinations of one or more of the associated listed items. The terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," "outer," and the like are used in the orientation or positional relationship indicated in the drawings for convenience in describing the invention and for simplicity in description, and do not indicate or imply that the referenced device or element must have a particular orientation, be constructed and operated in a particular orientation, and are therefore not to be construed as limiting the invention. The terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. The terms "mounted," "connected," and "coupled" are to be construed broadly and may, for example, be fixedly coupled, detachably coupled, or integrally coupled; can be mechanically or electrically connected; the two elements can be directly connected, indirectly connected through an intermediate medium, or communicated with each other inside; either a wireless or a wired connection. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
In addition, the technical features involved in the different embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
Example 1
The embodiment provides an artificial intelligence open fire detection method for park fire early warning, which comprises the following steps as shown in fig. 1:
s1: acquiring a video image acquired by an image acquisition device installed in a park;
s2: acquiring a differential image between a current frame video image and a reference frame video image; the reference frame video image is a video frame image acquired by the image acquisition device before;
s3: counting the distribution condition of the pixel values of all the pixel points in the differential image;
s4: if the distribution condition meets a preset condition, determining a judgment threshold value according to the pixel value of the pixel point in the differential image;
and if the distribution condition meets the non-preset condition, further open fire identification is not carried out on the current frame video image.
S5: determining a first image area from the differential image, wherein the first image area comprises all pixel points of which the pixel values are greater than the judgment threshold value in the differential image;
s6: determining a second image area in the current frame video image according to the position of the first image area in the differential image;
s7: and inputting the information obtained according to the second image area into an artificial intelligent model for open fire identification.
The artificial intelligent open fire detection method for garden fire early warning provided by the embodiment of the invention can be realized by a garden environment-controlled monitoring model, namely the steps can be executed by the garden environment-controlled monitoring model.
In the embodiment of the invention, the image change is detected through the difference image between the video frame images collected successively, and then the open fire detection is carried out on the changed area in the video frame image by using the artificial intelligence model. In addition, because the artificial intelligence model only detects the change area in the video image, the accuracy of open fire detection can be improved, and the probability of false alarm is reduced. In addition, in the embodiment of the invention, the judgment threshold value for judging the change area is determined according to the pixel value distribution condition of the pixel points of the difference image between the successively collected video frame images, so that the influence of the change area judgment caused by the change of external illumination (such as lightning, lamplight, sunlight and the like) and other factors can be reduced, and the accuracy of the change area judgment is improved.
According to the embodiment of the invention, the environment management state is monitored through the designed park environment-control monitoring model, the combustible condition is predicted, the park energy control and environment protection loopholes can be found, and the carbon neutralization comprehensive management capability is improved.
Optionally, the preset condition is that a difference between a first pixel value and a second pixel value is greater than a preset threshold, the first pixel value is a mean value or a median value of third pixel values in pixel values of each pixel point in the difference image, the second pixel value is a mean value or a median value of fourth pixel values in pixel values of each pixel point in the difference image, the third pixel value is a maximum N pixel values in pixel values of each pixel point in the difference image, the fourth pixel value is a minimum K pixel values in pixel values of each pixel point in the difference image, and N, K is a preset positive integer or is determined according to a preset proportion of all pixel values in the difference image.
In the embodiment of the invention, whether the open fire identification needs to be further carried out is determined according to the pixel value condition of the difference image between the video frame images collected successively, and the judgment accuracy is high.
Optionally, the determining a judgment threshold according to the pixel value of the pixel point in the difference image includes:
and determining the judgment threshold according to the first pixel value and the second pixel value.
Specifically, an average value of the first pixel value and the second pixel value may be taken as the determination threshold.
In the embodiment of the invention, the image area needing further open fire identification is determined according to the pixel value condition of the difference image between the successively collected video frame images, and the judgment accuracy is high.
Optionally, after the statistics of the distribution of the pixel values of each pixel point in the differential image, the method further includes:
if the distribution condition does not meet the preset condition, updating the reference frame video image into the current frame video image;
and if the distribution condition meets the preset condition, the reference frame video image is continuously used as the reference frame video image.
In the embodiment of the invention, the reference frame video image is updated according to the pixel value condition of the difference image between the video frame images, so that the subsequent judgment and identification are more accurate.
Optionally, as shown in fig. 2, the step S7 of inputting the information obtained from the second image area into an artificial intelligence model for open fire recognition includes:
s71: extracting a plurality of aspects of feature information from the second image region respectively, wherein the plurality of aspects of feature information comprise spectral feature information, shape feature information and texture feature information;
s72: aiming at the characteristic information of the aspects, respectively operating with corresponding characteristic offset matrixes trained in advance;
s73: cascading the calculated feature information of the multiple aspects to obtain cascaded feature information;
s74: carrying out normalization processing on the cascaded feature information by using a probability normalization algorithm to obtain normalized feature information;
s75: carrying out further feature extraction on the normalized feature information by utilizing differential operation, and removing redundant feature information to obtain deep-level feature information;
s76: and processing the deep-level feature information by using a principal component analysis algorithm to obtain information for inputting the information to the artificial intelligent model to identify open fire.
In the embodiment of the invention, the open fire identification is carried out by extracting the characteristics of multiple aspects of the image area to be identified, so that the identification accuracy can be further improved, and the misjudgment and the missed judgment can be avoided.
Further optionally, the extracting the feature information of the aspects from the second image regions respectively includes:
traversing the second image area by using a sliding window with a preset size according to the sequence of first row and second row, and calculating the entropy of the image area in the sliding window during each sliding; the entropy of the image area in the sliding window during the current sliding is the gray level histogram of the image area in the sliding window during the last sliding minus the gray level probability of the pixel points in the first row and plus the gray level probability of the pixel points in the last row in the image area in the sliding window during the current sliding;
acquiring a texture image of the second image area according to the entropy of the image area in the sliding window during each sliding;
and extracting the texture feature information from the texture image.
In other alternative specific embodiments, a plurality of different classifiers may be used to extract feature information of a plurality of aspects from the second image region, respectively.
Optionally, the processing the deep-level feature information by using a principal component analysis algorithm to obtain information for inputting to the artificial intelligent model to perform naked flame recognition includes:
solving covariance of every two pieces of feature information in the deep-level feature information to construct a covariance matrix;
obtaining an eigenvalue and an eigenvector corresponding to the covariance matrix;
selecting eigenvectors corresponding to the N maximum eigenvalues;
and projecting the deep level feature information onto the selected N feature vectors to complete the processing of the deep level feature information, wherein N is an integer greater than 1.
Optionally, as shown in fig. 3, the step S71 of extracting feature information of multiple aspects from the second image region respectively further includes:
s711: extracting feature information of the second image area by using a VGG neural network;
s712: the method comprises the steps of constructing a characteristic pyramid by a top-down method, carrying out up-sampling on a deep characteristic diagram with strong semantic characteristics in a bilinear interpolation mode to enable the deep characteristic diagram to be consistent with the size of a next layer of characteristic diagram, carrying out characteristic extraction on the up-sampled deep characteristic diagram by utilizing cavity convolution, adding the extracted deep characteristic diagram with the next layer of characteristic diagram to obtain a new characteristic diagram, continuously adding each obtained new characteristic diagram with the next layer of characteristic diagram in the same mode, and finally obtaining a depth heterogeneous characteristic diagram fusing a shallow characteristic diagram and the deep characteristic diagram as characteristic information of one aspect.
Optionally, the artificial intelligence model includes a multilayer neural network, and the multilayer neural network f is synthesized according to the following formula:
wherein M is 1,2,3, …, M, gm(x) Is the m-th neural network, amIs the weight of the mth neural network.
Optionally, before inputting the information obtained according to the second image area to an artificial intelligence model for open fire identification, the method further includes:
respectively constructing a plurality of the neural networks;
aiming at a plurality of neural networks, respectively training by using preset training samples;
training the weight a in the artificial intelligence model by using the preset training samplem(ii) a Wherein the weight a is used in trainingmThe iterative update is performed according to the following formula:
wherein, am(t+1)、am(t) represents the weight a after the t +1 th iteration and the t times of iteration updating respectivelymAnd E (t) is the identification error in the artificial intelligence model at the time of the t iteration update, btIs a parameter value determined according to the minimum recognition error in the artificial intelligence model in the previous t iteration updates, clThe weighted value of the ith sample is the value manually marked during the construction of the training sample, ol、ylThe target output value and the actual output value of the ith sample.
Example 2
The present embodiment provides an artificial intelligence open fire detection system 40 for fire early warning in a park, as shown in fig. 4, including:
one or more processors 401;
a storage 402 for storing one or more programs;
the one or more programs, when executed by the one or more processors 401, cause the one or more processors 401 to implement any of the above-described artificial intelligence open fire detection methods for campus fire early warning.
The above procedure may be referred to as a campus environment monitoring model.
According to the artificial intelligent open fire detection system for fire early warning in a park, provided by the embodiment of the invention, image change is detected through the difference image between the successively acquired video frame images, and then open fire detection is carried out on the changed area in the video frame image by using the artificial intelligent model. In addition, because the artificial intelligence model only detects the change area in the video image, the accuracy of open fire detection can be improved, and the probability of false alarm is reduced. In addition, in the embodiment of the invention, the judgment threshold value for judging the change area is determined according to the pixel value distribution condition of the pixel points of the difference image between the successively collected video frame images, so that the influence of the change area judgment caused by the change of external illumination (such as lightning, lamplight, sunlight and the like) and other factors can be reduced, and the accuracy of the change area judgment is improved.
It should be understood that the above examples are only for clarity of illustration and are not intended to limit the embodiments. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. And obvious variations or modifications therefrom are within the scope of the invention.
Claims (10)
1. An artificial intelligent open fire detection method for park fire early warning is characterized by comprising the following steps:
acquiring a video image acquired by an image acquisition device installed in a park;
acquiring a differential image between a current frame video image and a reference frame video image; the reference frame video image is a video frame image acquired by the image acquisition device before;
counting the distribution condition of the pixel values of all the pixel points in the differential image;
if the distribution condition meets a preset condition, determining a judgment threshold value according to the pixel value of the pixel point in the differential image;
determining a first image area from the differential image, wherein the first image area comprises all pixel points of which the pixel values are greater than the judgment threshold value in the differential image;
determining a second image area in the current frame video image according to the position of the first image area in the differential image;
and inputting the information obtained according to the second image area into an artificial intelligent model for open fire identification.
2. The method according to claim 1, wherein the preset condition is that a difference between a first pixel value and a second pixel value is greater than a preset threshold, the first pixel value is a mean value or a median value of third pixel values among the pixel values of the respective pixels in the difference image, the second pixel value is a mean value or a median value of fourth pixel values among the pixel values of the respective pixels in the difference image, the third pixel value is a largest N pixel values among the pixel values of the respective pixels in the difference image, the fourth pixel value is a smallest K pixel values among the pixel values of the respective pixels in the difference image, and N, K is a preset positive integer or is determined according to a preset proportion of all the pixel values in the difference image.
3. The method of claim 2, wherein determining the decision threshold according to the pixel values of the pixels in the difference image comprises:
and determining the judgment threshold according to the first pixel value and the second pixel value.
4. The method according to claim 1, wherein after the counting the distribution of the pixel values of the pixels in the difference image, the method further comprises:
if the distribution condition does not meet the preset condition, updating the reference frame video image into the current frame video image;
and if the distribution condition meets the preset condition, the reference frame video image is continuously used as the reference frame video image.
5. The method of claim 1, wherein inputting information obtained from the second image region into an artificial intelligence model for open flame recognition comprises:
extracting a plurality of aspects of feature information from the second image region respectively, wherein the plurality of aspects of feature information comprise spectral feature information, shape feature information and texture feature information;
aiming at the characteristic information of the aspects, respectively operating with corresponding characteristic offset matrixes trained in advance;
cascading the calculated feature information of the multiple aspects to obtain cascaded feature information;
carrying out normalization processing on the cascaded feature information by using a probability normalization algorithm to obtain normalized feature information;
carrying out further feature extraction on the normalized feature information by utilizing differential operation, and removing redundant feature information to obtain deep-level feature information;
and processing the deep-level feature information by using a principal component analysis algorithm to obtain information for inputting the information to the artificial intelligent model to identify open fire.
6. The method according to claim 5, wherein the extracting the feature information of the aspects from the second image region respectively comprises:
traversing the second image area by using a sliding window with a preset size according to the sequence of first row and second row, and calculating the entropy of the image area in the sliding window during each sliding; the entropy of the image area in the sliding window during the current sliding is the gray level histogram of the image area in the sliding window during the last sliding minus the gray level probability of the pixel points in the first row and plus the gray level probability of the pixel points in the last row in the image area in the sliding window during the current sliding;
acquiring a texture image of the second image area according to the entropy of the image area in the sliding window during each sliding;
and extracting the texture feature information from the texture image.
7. The method of claim 5, wherein the extracting the feature information of the aspects from the second image regions respectively further comprises:
extracting feature information of the second image area by using a VGG neural network;
the method comprises the steps of constructing a characteristic pyramid by a top-down method, carrying out up-sampling on a deep characteristic diagram with strong semantic characteristics in a bilinear interpolation mode to enable the deep characteristic diagram to be consistent with the size of a next layer of characteristic diagram, carrying out characteristic extraction on the up-sampled deep characteristic diagram by utilizing cavity convolution, adding the extracted deep characteristic diagram with the next layer of characteristic diagram to obtain a new characteristic diagram, continuously adding each obtained new characteristic diagram with the next layer of characteristic diagram in the same mode, and finally obtaining a depth heterogeneous characteristic diagram fusing a shallow characteristic diagram and the deep characteristic diagram as characteristic information of one aspect.
8. The method of claim 1, wherein the artificial intelligence model comprises a multi-layer neural network, wherein the multi-layer neural network f is synthesized according to the following formula:
wherein M is 1,2,3, …, M, gm(x) Is the m-th neural network, amIs the weight of the mth neural network.
9. The method according to claim 8, wherein before inputting the information obtained from the second image area into an artificial intelligence model for open fire identification, the method further comprises:
respectively constructing a plurality of the neural networks;
aiming at a plurality of neural networks, respectively training by using preset training samples;
training the weight a in the artificial intelligence model by using the preset training samplem(ii) a Wherein the weight a is used in trainingmThe iterative update is performed according to the following formula:
wherein, am(t+1)、am(t) represents the weight a after the t +1 th iteration and the t times of iteration updating respectivelymAnd E (t) is the identification error in the artificial intelligence model at the time of the t iteration update, btIs a parameter value determined according to the minimum recognition error in the artificial intelligence model in the previous t iteration updates, clThe weighted value of the ith sample is the value manually marked during the construction of the training sample, ol、ylThe target output value and the actual output value of the ith sample.
10. An artificial intelligence naked light detection system for park fire early warning, which is characterized by comprising:
one or more processors;
storage means for storing one or more programs;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the artificial intelligence open fire detection method for campus fire early warning recited in any one of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111606186.1A CN114283367B (en) | 2021-12-26 | 2021-12-26 | Artificial intelligent open fire detection method and system for garden fire early warning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111606186.1A CN114283367B (en) | 2021-12-26 | 2021-12-26 | Artificial intelligent open fire detection method and system for garden fire early warning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114283367A true CN114283367A (en) | 2022-04-05 |
| CN114283367B CN114283367B (en) | 2022-09-06 |
Family
ID=80875600
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111606186.1A Active CN114283367B (en) | 2021-12-26 | 2021-12-26 | Artificial intelligent open fire detection method and system for garden fire early warning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114283367B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115862259A (en) * | 2023-02-24 | 2023-03-28 | 深圳市特安电子有限公司 | Fire alarm early warning system based on temperature monitoring |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050180600A1 (en) * | 2004-02-13 | 2005-08-18 | Quen-Zong Wu | Method for detecting motion pixels in image |
| CN102208018A (en) * | 2011-06-01 | 2011-10-05 | 西安工程大学 | Method for recognizing fire disaster of power transmission line based on video variance analysis |
| US20130236098A1 (en) * | 2012-03-09 | 2013-09-12 | Omron Corporation | Image processing device, image processing method, and image processing program |
| CN103338336A (en) * | 2013-07-03 | 2013-10-02 | 国家电网公司 | Video detection method for firework early warning of overhead power transmission line passage |
| CN105678272A (en) * | 2016-03-25 | 2016-06-15 | 符锌砂 | Complex environment target detection method based on image processing |
| CN105788142A (en) * | 2016-05-11 | 2016-07-20 | 中国计量大学 | Video image processing-based fire detection system and detection method |
| CN110135269A (en) * | 2019-04-18 | 2019-08-16 | 杭州电子科技大学 | A kind of fire image detection method based on blend color model and neural network |
| CN110322659A (en) * | 2019-06-21 | 2019-10-11 | 江西洪都航空工业集团有限责任公司 | A kind of smog detection method |
| CN110812753A (en) * | 2019-09-23 | 2020-02-21 | 重庆特斯联智慧科技股份有限公司 | Artificial intelligent fire extinguishing method with open fire point identification function and fire extinguisher equipment |
| CN111461076A (en) * | 2019-11-01 | 2020-07-28 | 武汉纺织大学 | Smoke detection method and smoke detection system combining frame difference method and neural network |
| CN113205659A (en) * | 2021-03-19 | 2021-08-03 | 武汉特斯联智能工程有限公司 | Fire disaster identification method and system based on artificial intelligence |
| CN113420600A (en) * | 2021-05-25 | 2021-09-21 | 杭州申弘智能科技有限公司 | Fire-fighting classification early warning algorithm based on visual image recognition |
-
2021
- 2021-12-26 CN CN202111606186.1A patent/CN114283367B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050180600A1 (en) * | 2004-02-13 | 2005-08-18 | Quen-Zong Wu | Method for detecting motion pixels in image |
| CN102208018A (en) * | 2011-06-01 | 2011-10-05 | 西安工程大学 | Method for recognizing fire disaster of power transmission line based on video variance analysis |
| US20130236098A1 (en) * | 2012-03-09 | 2013-09-12 | Omron Corporation | Image processing device, image processing method, and image processing program |
| CN103338336A (en) * | 2013-07-03 | 2013-10-02 | 国家电网公司 | Video detection method for firework early warning of overhead power transmission line passage |
| CN105678272A (en) * | 2016-03-25 | 2016-06-15 | 符锌砂 | Complex environment target detection method based on image processing |
| CN105788142A (en) * | 2016-05-11 | 2016-07-20 | 中国计量大学 | Video image processing-based fire detection system and detection method |
| CN110135269A (en) * | 2019-04-18 | 2019-08-16 | 杭州电子科技大学 | A kind of fire image detection method based on blend color model and neural network |
| CN110322659A (en) * | 2019-06-21 | 2019-10-11 | 江西洪都航空工业集团有限责任公司 | A kind of smog detection method |
| CN110812753A (en) * | 2019-09-23 | 2020-02-21 | 重庆特斯联智慧科技股份有限公司 | Artificial intelligent fire extinguishing method with open fire point identification function and fire extinguisher equipment |
| CN111461076A (en) * | 2019-11-01 | 2020-07-28 | 武汉纺织大学 | Smoke detection method and smoke detection system combining frame difference method and neural network |
| CN113205659A (en) * | 2021-03-19 | 2021-08-03 | 武汉特斯联智能工程有限公司 | Fire disaster identification method and system based on artificial intelligence |
| CN113420600A (en) * | 2021-05-25 | 2021-09-21 | 杭州申弘智能科技有限公司 | Fire-fighting classification early warning algorithm based on visual image recognition |
Non-Patent Citations (6)
| Title |
|---|
| XUEHUI WU等: ""An adaptive threshold deep learning method for fire and smoke detection"", 《2017 IEEE INTERNATIONAL CONFERENCE ON SYSTEMS, MAN, AND CYBERNETICS (SMC)》 * |
| YAKUN XIE等: ""Efficient Video Fire Detection Exploiting Motion-Flicker-Based Dynamic Features and Deep Static Features"", 《IEEE ACCESS》 * |
| 张斌: ""基于视频序列的早期林火烟雾检测技术研究"", 《中国优秀硕士学位论文全文数据库 农业科技辑》 * |
| 徐铭铭等: "基于时空特征的林火视频火焰识别研究", 《林业工程学报》 * |
| 林宏等: "基于视频的林火烟雾识别算法研究", 《安全与环境学报》 * |
| 蔡春兵等: ""基于深度学习的视频火焰识别方法"", 《信息技术与网络安全》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115862259A (en) * | 2023-02-24 | 2023-03-28 | 深圳市特安电子有限公司 | Fire alarm early warning system based on temperature monitoring |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114283367B (en) | 2022-09-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111209434B (en) | Substation equipment inspection system and method based on multi-source heterogeneous data fusion | |
| CN108446630B (en) | Intelligent monitoring method for airport runway, application server and computer storage medium | |
| CN112287816B (en) | Dangerous work area accident automatic detection and alarm method based on deep learning | |
| CN110717481B (en) | Method for realizing face detection by using cascaded convolutional neural network | |
| CN112132090A (en) | Smoke and fire automatic detection and early warning method based on YOLOV3 | |
| CN106128022B (en) | A kind of wisdom gold eyeball identification violent action alarm method | |
| CN110070078B (en) | Drunk driving detection method and system based on sensor and machine vision | |
| CN109484935A (en) | A kind of lift car monitoring method, apparatus and system | |
| CN106228709B (en) | A kind of wisdom gold eyeball identifies that one adds paper money alarm method and device | |
| CN111223263A (en) | Full-automatic comprehensive fire early warning response system | |
| CN116629465B (en) | Smart power grids video monitoring and risk prediction response system | |
| CN112163572A (en) | Method and device for identifying object | |
| CN114283367B (en) | Artificial intelligent open fire detection method and system for garden fire early warning | |
| CN108664875A (en) | Underground belt-conveying monitoring method based on image recognition | |
| CN113192038A (en) | Method for identifying and monitoring abnormal smoke and fire in existing flame environment based on deep learning | |
| CN113990017B (en) | Forest and grassland fire early warning system and method based on PNN neural network | |
| CN112643719A (en) | Tunnel security detection method and system based on inspection robot | |
| CN115691034A (en) | Intelligent household abnormal condition warning method, system and storage medium | |
| CN107704818A (en) | A kind of fire detection system based on video image | |
| CN117354986B (en) | Intelligent control method and system for multifunctional LED lamp beads | |
| CN109001210B (en) | System and method for detecting aging and cracking of sealing rubber strip of civil air defense door | |
| CN116761049B (en) | Household intelligent security monitoring method and system | |
| CN116758441B (en) | Unmanned aerial vehicle cluster intelligent scheduling management system | |
| CN115841731A (en) | Infrared-monitoring park fire early warning method | |
| CN115810161A (en) | Transformer substation fire identification method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |