CN117853935A - Cable flame spread detection method and device based on visual analysis and service platform - Google Patents
Cable flame spread detection method and device based on visual analysis and service platform Download PDFInfo
- Publication number
- CN117853935A CN117853935A CN202410258748.5A CN202410258748A CN117853935A CN 117853935 A CN117853935 A CN 117853935A CN 202410258748 A CN202410258748 A CN 202410258748A CN 117853935 A CN117853935 A CN 117853935A
- Authority
- CN
- China
- Prior art keywords
- cable
- image
- visual analysis
- flame
- model
- 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 35
- 230000000007 visual effect Effects 0.000 title claims abstract description 32
- 238000004458 analytical method Methods 0.000 title claims abstract description 31
- 238000000034 method Methods 0.000 claims abstract description 36
- 230000008569 process Effects 0.000 claims abstract description 19
- 230000007480 spreading Effects 0.000 claims abstract description 11
- 238000012360 testing method Methods 0.000 claims description 12
- 238000003062 neural network model Methods 0.000 claims description 9
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000001914 filtration Methods 0.000 claims description 7
- 230000000873 masking effect Effects 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000000926 separation method Methods 0.000 claims description 2
- 238000002485 combustion reaction Methods 0.000 abstract description 5
- 230000007613 environmental effect Effects 0.000 abstract description 4
- 230000006870 function Effects 0.000 description 11
- 230000004913 activation Effects 0.000 description 6
- 238000011176 pooling Methods 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000003213 activating effect Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000008033 biological extinction Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000000779 smoke Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/36—Applying a local operator, i.e. means to operate on image points situated in the vicinity of a given point; Non-linear local filtering operations, e.g. median filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Nonlinear Science (AREA)
- Fire-Detection Mechanisms (AREA)
Abstract
The invention provides a cable flame spread detection method, a device and a service platform based on visual analysis. And transmitting the three-channel original image and the preprocessed covering image into a flame brightness and morphology mixed recognition model, so as to detect the flame occurrence position and the flame morphology characteristics in the original image and accurately obtain the cable flame spreading process parameters. The aim of the pretreatment is to reduce the interference of environmental factors on the objects of interest (i.e. the cables and their combustion conditions) and to improve the accuracy of the visual analysis.
Description
Technical Field
The invention belongs to the field of material analysis and detection, and particularly relates to a cable flame spread detection method, device and service platform based on visual analysis.
Background
With the remarkable improvement of electrification degree, electrical equipment and electronic products are widely used, the occurrence frequency of electrical fires is continuously improved, and the occurrence frequency of electrical fires is gradually the first cause of the fires. An electrical fire is generally defined as a non-faulty or faulty electrothermal conversion (e.g., arc, high temperature, spark, lightning, static electricity, etc.) of an electrical wire or cable, a power distribution device, an electrical instrument, etc., releasing a large amount of energy to ignite a body or a combustible object having combustion conditions in the vicinity, thereby initiating the fire. Wherein the cable and the short circuit and overload of the cable are one of the most leading causes of fire. Therefore, research and testing on related mechanisms of the ignition and spreading behaviors of the cable are carried out, and the method is an important and necessary work for guaranteeing the quality of the cable and preventing the fire disaster.
The ignition and propagation characteristics of cables are well known in the industry, and the propagation of cables along surfaces after ignition is often the primary cause of fire spread. The cable fire spreading process is more complex than the common solid, and therefore, the flame form, the fire spreading speed, the flame characteristic length and the extinction critical of the cable fire spreading process need to be researched and tested. Often, these tests need to be repeated and carried out in large quantities, and the conventional test process is identified by human eyes, which has the problems of high working strength, high risk degree and large deviation between the test level and the standard.
With the evolution of machine vision technology, the vision-based flame detection and measurement method has come to be widely applied to industries such as smoke detection, fire detection and the like. The classical visual flame detection task generally only needs to detect whether a fire source exists or not, but faces the more complex parameter test requirements in the cable fire spreading process, and the detection accuracy of a classical model is lower.
Disclosure of Invention
In order to meet the index requirements of a cable flame spread test, a cable flame spread detection method based on visual analysis is provided, a classical visual flame detection model is utilized and improved, and a detection process is optimized, so that the detection method achieves higher detection precision and robustness.
A cable flame spread detection method based on visual analysis, comprising:
(1) Aligning the test cable, and collecting video at a predetermined frame rate;
(2) Arranging two-dimensional image frames in time sequence, and preprocessing the two-dimensional image frames frame by frame; wherein the pretreatment comprises: establishing a light distribution color model of cable appearance
Wherein the method comprises the steps ofIs a variable of the light distribution color model, +.>Representing the color value of a pixel>Represents the average brightness level of the image, C represents +.>Set of->Represents the sample mean>Representing a covariance matrix;
is provided withRepresenting positive sample model parameters +.>Representing negative sample model parameters; for any image->One pixel of +.>Calculating the average brightness level of the image>If->The pixel is marked 1, otherwise 0, whereby +/for each image>Calculating to obtain a mask image->;
(3) Constructing a flame brightness and morphology mixed recognition model, and detecting the flame occurrence position and flame morphology characteristics in an original image by using the input original image and a covering image so as to obtain cable flame spreading process parameters; the model is a neural network model and comprises 6 hidden layers and an output layer;
wherein the first hidden layer is defined as:
wherein the first hidden layer is transportedThe outlet is divided into two partsAnd->;/>Obtained from the hue channel H and saturation channel S of the original image, comprising +.>Total 16×2=32 independent convolution kernels; />Obtained from the luminance channel V and the mask image channel M of the original image, comprising +.>A total of 8*2 = 16 independent convolution kernels.
The preprocessing further comprises: and carrying out preliminary filtering on natural noise of each frame of image data by adopting a median filtering algorithm.
The preprocessing enables marking of each pixel of the image, distinguishing between a cable portion and a non-cable portion.
The pixel labeled 1 represents a positive sample and the pixel labeled 0 represents a negative sample.
The positive samples are cables and the negative samples are non-cables.
Before inputting the model, the original image is separated into brightness V, hue H, and saturation S.
The separation method comprises the following steps:
where R, G, B is the color parameter of the original image.
The neural network model is provided with four input channels, namely a brightness V channel, a tone H channel and a saturation S channel of an original image, and a masking image channel.
A device for detecting flame spread of a cable, which uses the method for detecting flame spread of a cable based on visual analysis.
A service platform for realizing cable flame spread detection, and the cable flame spread detection method based on visual analysis is implemented.
The invention has the following technical effects:
1. the invention is characterized in that a video/image acquisition system for cable flame spread is provided, video data of a test cable are captured and converted into two-dimensional images frame by frame for preprocessing. The aim of the pretreatment is to reduce the interference of environmental factors on the objects of interest (i.e. the cables and their combustion conditions) and to improve the accuracy of the visual analysis.
2. The invention is characterized in that in the preprocessing stage, a light distribution color model of the appearance of the cable is established, the distribution model is used for measuring the color values of the pixels of the image under different environmental lights, the pixels are marked according to the light distribution color model, a covering image is generated, and the covering image and the original image are transmitted into a visual analysis model, so that the analysis result is improved.
3. The invention is characterized in that the invention provides a flame brightness and morphology mixed recognition model, which utilizes an input three-channel original image and a masking image to detect the flame occurrence position and flame morphology characteristics in the original image so as to accurately obtain the cable flame spreading process parameters.
4. The invention is characterized in that the neural network model is used to introduce a hidden image channel and further introduce a separation-restoration structure, and the characteristics are respectively extracted by separating the input original image data into two parts in logic, so that the flame form and the flame brightness are respectively encoded, and the recognition rate of characteristic recognition is improved; the combination of the separated logic portions in the fifth and sixth hidden layers is restored and features are further encoded by the fully connected layers, enabling improved flame identification.
Detailed Description
Step 1, capturing video data of a test cable by using a video/image acquisition system of cable flame propagation, and converting the video data into a two-dimensional covering image frame by frame so as to facilitate visual analysis.
First, a high definition camera is used to align the test cable device and capture video at a frame rate.
Next, two-dimensional image frames are arranged in time order and preprocessed frame by frame. The aim of the pretreatment is to reduce the interference of environmental factors on the objects of interest (i.e. the cables and their combustion conditions) and to improve the accuracy of the visual analysis.
Preprocessing is one of the common processes of visual analysis for removing signal content (called noise) in a video or image signal that is not related to an object of interest (called a signal) to increase the signal-to-noise ratio, also called filtering. For continuous signals, frequency domain filters are typically used. For discrete data, a convolution template similar to a frequency domain filter, such as a gaussian convolution, is typically employed.
The conventional pretreatment process mainly removes natural noise naturally generated due to electronic components, interference of ambient light, and the like. In this case, since automatic detection is required for multiple parameters of flame spread of the cable, the position of the cable in the video data needs to be located on the basis of removing the above natural noise, so as to provide a reference for the subsequent visual analysis process.
The improved pretreatment procedure is as follows.
And P1, carrying out preliminary filtering on natural noise of each frame of image data by adopting a median filtering algorithm.
P2, each pixel of the image is further marked, i.e. cable portions and non-cable portions.
The marking process comprises the following specific operation steps:
and P2.1, establishing a light distribution color model of the appearance of the cable, and measuring a distribution model of pixel color values of the image under different ambient lights.
Because the illumination is generated when the cable burns, the color distribution of the cable in the image is influenced, and in order to ensure that the method can stably work in the whole process of the burning propagation test, the establishment of a light distribution color model of the cable is required.
The light distribution color model of the cable is defined as:
in the above-mentioned method, the step of,is a variable of the light distribution color model, +.>Representing the color value of a pixel>Representing the average brightness level of the image, wherein the calculating method is the average value of pixel color values between 5% and 25% of the highest color value in the image;represents the probability parameter to be estimated, here denoted by C +.>Is a set of (3). />Represents the sample mean>Representing the covariance matrix. By introducing the average brightness level variable of the image, the influence of illumination generated during cable combustion on the color distribution of the image can be compensated, and the stability of visual analysis is improved.
Training samples are prepared, including positive and negative samples. The positive samples are cable images and the negative samples are non-cable images. And adopts maximum likelihood estimation method to respectively calculate according to positive and negative samples、/>A light distribution color model of the cable is obtained for marking whether a pixel in the image is a cable.
And P2.2, marking the pixels according to a light distribution color model.
Is provided withRepresenting positive sample model parameters +.>Representing negative sample model parameters. One pixel in one image +.>Calculating the average brightness level of the image>And compare, if->The pixel is marked as positive sample 1 (cable) and otherwise as negative sample 0 (non-cable).
Through the marking process, for each imageCalculating to obtain a mask image->The size of the image is equal to that of an original image, the value of each pixel is 1 or 0, and the pixel corresponding to the position corresponding to the original image is a positive sample or a negative sample. The mask image will be passed into the visual analysis model along with the original image, thereby improving the analysis results.
And 2, constructing a flame brightness and morphology mixed recognition model, and detecting the flame occurrence position and the flame morphology characteristics in the original image by using the input original image and the mask image so as to obtain the cable fire spreading process parameters.
The original image obtained with the camera comprises three channels RGB, the pixel value of each channel being a scalar, which is first separated into brightness (V), hue (H) and saturation (S).
Designing a neural network classification model, which inputs two-dimensional images of four channels, namely three channels of brightness (V), hue (H) and saturation (S), and masking the image channels obtained in the step 1。
The input passes through a plurality of hidden layers of the neural network model, and a required identification result is output. The specific process is as follows.
First, by the first hidden layer, it is defined as:
the left side is the output of the first hidden layer and is divided into two partsAnd->。/>Obtained from hue channel H and saturation channel S, comprising +.>A total of 16×2=32 independent convolution kernels, each modeling the hue feature saturation feature of the image, which mainly encodes the flame morphology. />Obtained from the luminance channel V and the mask image channel M, comprisingA total of 8*2 = 16 independent convolution kernels for modeling the brightness characteristics of the flame, this part mainly encoding the flame's position.
On the right side of the frame, the frame is provided with a plurality of grooves,、/>、/>、/>linear coefficients representing the convolution kernel, +.>、/>Is linear intercept>Is an activation function of the neural network, +.>Refer to image coordinates +.>Refers to coordinates within the convolution kernel.
Activation function of neural networkThe definition is as follows:
in the method, in the process of the invention,and for activating the scale parameters of the function, the learning convergence rate is adjusted. Compared with classical sigmoid functions, reLU functions and the like, the piecewise functions have better classification effect on the input joint samples of all channels.
Next, by the second hidden layer, it is defined as:
the left side in the above equation is the output of the second hidden layer, the right max represents the maximum pooling, and the local maximum is taken in the pooling window of 4*4.
Next, by the third hidden layer, it is defined as:
the left side is the output of the third hidden layer, corresponding toComprises->Total 16×2=32 independent convolution kernels. />Comprises->A total of 8*2 = 16 independent convolution kernels. Right, add>、/>Linear coefficients representing the convolution kernel, +.>、/>Is linear intercept>Is the activation function defined previously.
Next, by the fourth hidden layer, it is defined as:
the left side in the above equation is the output of the fourth hidden layer, the right side max represents the maximum pooling, and the local maximum is taken in the pooling window of 4*4.
Next, by the fifth hidden layer, it is defined as:
left, output of the fifth hidden layerAnd->Corresponding->Comprises->Total 16×2=32 independent convolution kernels. />Comprises->A total of 8*2 = 16 independent convolution kernels. Right, add>、/>Linear coefficients representing the convolution kernel, +.>、/>Is linear intercept>Is the activation function defined previously.
Next, by the sixth hidden layer, it is defined as:
left, is the output of the sixth hidden layerBy a set of linear coefficients->The two logical parts of the fifth hidden layer are connected together and the pixels at the corresponding positions share coefficients. />For linear intercept>Is the activation function defined previously.
All six hidden layers of the neural network model are defined above. Compared with the classical convolution network for image recognition, the neural network model used in the method introduces a masked image channel and further introduces a separation-restoration structure, and features are respectively extracted by separating input original image data into two parts on logic, so that the flame morphology and flame brightness are respectively encoded, and the recognition rate of feature recognition is improved; the combination of the separated logic portions in the fifth and sixth hidden layers is restored and features are further encoded by the fully connected layers, enabling improved flame identification.
The output layer of the neural network model is defined as:
from linear coefficientsAnd forming a full connection layer with the sixth hidden layer. />For linear intercept>Is the activation function defined previously. Output->Corresponding to flames of different morphologies.
By usingQTraining data of flame-like morphology, training the model according to a cost function
And (5) iteratively optimizing and solving an optimal solution of the linear coefficient and the linear intercept.Representing training sample marker values.
According to the trained model, the position and the form of the flame in the image can be detected frame by frame, and detection indexes such as the flame spreading speed, the flame characteristic length and the like can be further calculated.
The invention provides a cable flame spread detection method based on visual analysis, which utilizes and improves a classical visual flame detection model, optimizes the detection process, realizes the detection of more cable flame spread indexes, and achieves higher detection precision and robustness compared with the existing visual detection method.
TABLE 1
It is to be understood that the above embodiments are merely examples presented under the inventive concept and do not provide a less limiting effect on the inventive concept.
Claims (10)
1. A cable flame spread detection method based on visual analysis is characterized in that:
(1) Aligning the test cable, and collecting video at a predetermined frame rate;
(2) Arranging two-dimensional image frames in time sequence, and preprocessing the two-dimensional image frames frame by frame; wherein the pretreatment comprises: establishing a light distribution color model of cable appearance
;
Wherein the method comprises the steps ofIs a variable of the light distribution color model, +.>Representing the color value of a pixel>Represents the average brightness level of the image, C represents +.>Set of->Represents the sample mean>Representing a covariance matrix;
is provided withRepresenting positive sample model parameters +.>Representing negative sample model parameters; for any image->One pixel of +.>Calculating the average brightness level of the image>If->The pixel is marked 1, otherwise 0, whereby +/for each image>Calculating to obtain a mask image->;
(3) Constructing a flame brightness and morphology mixed recognition model, and detecting the flame occurrence position and flame morphology characteristics in an original image by using the input original image and a covering image so as to obtain cable flame spreading process parameters; the model is a neural network model and comprises 6 hidden layers and an output layer;
wherein the first hidden layer is defined as:
;
wherein the output of the first hidden layer is divided into two partsAnd->;/>Obtained from the hue channel H and saturation channel S of the original image, comprising +.>Total 16×2=32 independent convolution kernels; />Obtained from the luminance channel V and the mask image channel M of the original image, comprising +.>A total of 8*2 = 16 independent convolution kernels.
2. A method of cable flame spread detection based on visual analysis as claimed in claim 1, wherein: the preprocessing further comprises: and carrying out preliminary filtering on natural noise of each frame of image data by adopting a median filtering algorithm.
3. A method of cable flame spread detection based on visual analysis as claimed in claim 1, wherein: the preprocessing enables marking of each pixel of the image, distinguishing between a cable portion and a non-cable portion.
4. A method of cable flame spread detection based on visual analysis as claimed in claim 1, wherein: the pixel labeled 1 represents a positive sample and the pixel labeled 0 represents a negative sample.
5. The method for detecting flame spread of a cable based on visual analysis according to claim 4, wherein: the positive samples are cables and the negative samples are non-cables.
6. A method of cable flame spread detection based on visual analysis as claimed in claim 1, wherein: before inputting the model, the original image is separated into brightness V, hue H, and saturation S.
7. The visual analysis-based cable flame spread detection method of claim 6, wherein: the separation method comprises the following steps:
;
where R, G, B is the color parameter of the original image.
8. A method of cable flame spread detection based on visual analysis as claimed in claim 1, wherein: the neural network model is provided with four input channels, namely a brightness V channel, a tone H channel and a saturation S channel of an original image, and a masking image channel.
9. The utility model provides a device that cable flame spread detected which characterized in that: use of a visual analysis based cable flame spread detection method according to any of claims 1-8.
10. The utility model provides a realize cable flame and spread service platform who detects which characterized in that: implementing a visual analysis-based cable flame spread detection method according to any one of claims 1-8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410258748.5A CN117853935B (en) | 2024-03-07 | 2024-03-07 | Cable flame spread detection method and device based on visual analysis and service platform |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410258748.5A CN117853935B (en) | 2024-03-07 | 2024-03-07 | Cable flame spread detection method and device based on visual analysis and service platform |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117853935A true CN117853935A (en) | 2024-04-09 |
CN117853935B CN117853935B (en) | 2024-06-11 |
Family
ID=90546917
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410258748.5A Active CN117853935B (en) | 2024-03-07 | 2024-03-07 | Cable flame spread detection method and device based on visual analysis and service platform |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117853935B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20090054522A (en) * | 2007-11-27 | 2009-06-01 | 계명대학교 산학협력단 | Fire detection system and method basedon visual data |
CN101840571A (en) * | 2010-03-30 | 2010-09-22 | 杭州电子科技大学 | Flame detection method based on video image |
US20120148148A1 (en) * | 2010-11-16 | 2012-06-14 | Industry Academic Cooperation Foundation Keimyung University | Method for detecting fire-flame using fuzzy finite automata |
CN102509078A (en) * | 2011-10-28 | 2012-06-20 | 北京安控科技股份有限公司 | Fire detection device based on video analysis |
CN107944359A (en) * | 2017-11-14 | 2018-04-20 | 中电数通科技有限公司 | Flame detecting method based on video |
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 |
WO2022121129A1 (en) * | 2020-12-12 | 2022-06-16 | 南方电网调峰调频发电有限公司 | Fire recognition method and apparatus, and computer device and storage medium |
CN115909196A (en) * | 2022-11-03 | 2023-04-04 | 安徽大学 | Video flame detection method and system |
-
2024
- 2024-03-07 CN CN202410258748.5A patent/CN117853935B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20090054522A (en) * | 2007-11-27 | 2009-06-01 | 계명대학교 산학협력단 | Fire detection system and method basedon visual data |
CN101840571A (en) * | 2010-03-30 | 2010-09-22 | 杭州电子科技大学 | Flame detection method based on video image |
US20120148148A1 (en) * | 2010-11-16 | 2012-06-14 | Industry Academic Cooperation Foundation Keimyung University | Method for detecting fire-flame using fuzzy finite automata |
CN102509078A (en) * | 2011-10-28 | 2012-06-20 | 北京安控科技股份有限公司 | Fire detection device based on video analysis |
CN107944359A (en) * | 2017-11-14 | 2018-04-20 | 中电数通科技有限公司 | Flame detecting method based on video |
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 |
WO2022121129A1 (en) * | 2020-12-12 | 2022-06-16 | 南方电网调峰调频发电有限公司 | Fire recognition method and apparatus, and computer device and storage medium |
US20240005757A1 (en) * | 2020-12-12 | 2024-01-04 | Csg Power Generation Co., Ltd. | Method and device for smoke or fire recognition, computer apparatus, and storage medium |
CN115909196A (en) * | 2022-11-03 | 2023-04-04 | 安徽大学 | Video flame detection method and system |
Non-Patent Citations (3)
Title |
---|
DODDAPANENI TANMAYI ET AL.: "Detecting Fire in Color Images using Convolutional Neural Network Architectures", 《2023 IEEE 8TH INTERNATIONAL CONFERENCE FOR CONVERGENCE IN TECHNOLOGY》, 23 May 2023 (2023-05-23), pages 1 - 6 * |
PAVEL CHMELAR ET AL.: "Efficiency of HSV over RGB Gaussian Mixture Model for Fire Detection", 《2014 24TH INTERNATIONAL CONFERENCE RADIOELEKTRONIKA》, 16 June 2014 (2014-06-16), pages 1 - 4 * |
蒋先刚 等: "基于时空块协方差融合特征的火焰识别方法", 《计算机工程与应用》, vol. 52, no. 17, 31 December 2016 (2016-12-31), pages 208 - 214 * |
Also Published As
Publication number | Publication date |
---|---|
CN117853935B (en) | 2024-06-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Mahmoud et al. | Forest fire detection and identification using image processing and SVM | |
Ahmad et al. | Color classifier for symptomatic soybean seeds using image processing | |
Premal et al. | Image processing based forest fire detection using YCbCr colour model | |
Gasparini et al. | Color correction for digital photographs | |
CN107944359A (en) | Flame detecting method based on video | |
CN109637068A (en) | Intelligent pyrotechnics identifying system | |
CN104184925A (en) | Video scene change detection method | |
JP2006524378A (en) | Method for qualitative determination of a material having at least one recognition characteristic | |
CN107657619A (en) | A kind of low-light (level) Forest fire image dividing method | |
CN110812753B (en) | Artificial intelligent fire extinguishing method with open fire point identification function and fire extinguisher equipment | |
CN111160216A (en) | Multi-feature multi-model living human face recognition method | |
Gouinaud et al. | Color correction in the framework of color logarithmic image processing | |
CN105894015A (en) | Gate state analyzing method and system | |
CN109089992A (en) | A kind of classification method and system of the Estimation of The Fish Freshness based on machine vision | |
WO2014200155A1 (en) | Apparatus for separating overlapping peak in spectrum and x-ray fluorescence analysis apparatus using same | |
Li et al. | Identifying photorealistic computer graphics using second-order difference statistics | |
CN117853935B (en) | Cable flame spread detection method and device based on visual analysis and service platform | |
Mohammed et al. | Image segmentation for skin detection | |
Suhasini et al. | PCB defect detection using image subtraction algorithm | |
Ji-neng et al. | An early flame identification method based on edge gradient feature | |
CN111179293A (en) | Bionic contour detection method based on color and gray level feature fusion | |
CN116052041A (en) | Indicating lamp state identification method based on depth network | |
Wang et al. | Low-light forest flame image segmentation based on color features | |
CN112560706B (en) | Method and device for identifying water body target of multi-source satellite image | |
CN113504240A (en) | Power equipment fault detection method based on layered coding and visual perception |
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 |