CN111598840A - Smoke blackness detection method and system and storage medium - Google Patents
Smoke blackness detection method and system and storage medium Download PDFInfo
- Publication number
- CN111598840A CN111598840A CN202010326812.0A CN202010326812A CN111598840A CN 111598840 A CN111598840 A CN 111598840A CN 202010326812 A CN202010326812 A CN 202010326812A CN 111598840 A CN111598840 A CN 111598840A
- Authority
- CN
- China
- Prior art keywords
- smoke
- image
- brightness
- area
- region
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000000779 smoke Substances 0.000 title claims abstract description 134
- 238000001514 detection method Methods 0.000 title claims description 34
- 238000000034 method Methods 0.000 claims abstract description 32
- 238000013145 classification model Methods 0.000 claims abstract description 28
- 230000011218 segmentation Effects 0.000 claims abstract description 19
- 238000012545 processing Methods 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims description 24
- 239000013598 vector Substances 0.000 claims description 22
- 238000012706 support-vector machine Methods 0.000 claims description 19
- 238000013507 mapping Methods 0.000 claims description 14
- 239000003086 colorant Substances 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- UGFAIRIUMAVXCW-UHFFFAOYSA-N Carbon monoxide Chemical compound [O+]#[C-] UGFAIRIUMAVXCW-UHFFFAOYSA-N 0.000 abstract description 5
- 239000003546 flue gas Substances 0.000 abstract description 5
- 230000006870 function Effects 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 6
- 238000012544 monitoring process Methods 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000013500 data storage Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000002912 waste gas Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G06T5/94—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Abstract
The application discloses a method, a system and a storage medium for detecting smoke blackness, wherein the method comprises the following steps: acquiring a first image, wherein the first image comprises an image of a sky area and a smoke area; graying the first image to obtain a second image; performing binary segmentation on the second image to obtain a first sky area and a first smoke area; calculating the brightness of a first sky area in the first image, and compensating the brightness of pixel points of a first smoke area in the first image based on the brightness of the first sky area to obtain the brightness data of the compensated first smoke area; carrying out brightness distribution probability statistics on the compensated brightness data of the first smoke region to obtain statistical data; and inputting the statistical data into a classification model to obtain a first prediction result of the blackness of the smoke. This application compensates regional luminance of flue gas based on regional luminance of sky, can reduce the environment to the interference that detects, improves the degree of accuracy that detects. The method and the device can be widely applied to the field of image processing.
Description
Technical Field
The application relates to the field of image processing, in particular to a smoke blackness detection method, a smoke blackness detection system and a storage medium.
Background
With the development of social economy, people put higher demands on environmental protection. Various systems for detecting the environment have been developed. For plants producing waste gas, the blackness of the flue gas often represents the pollution degree, and in the past, technicians visually inspect the blackness of the flue gas by setting a monitoring station.
The traditional method consumes manpower, so that the blackness of the smoke is judged by adopting an image detection mode. However, the existing image technology is interfered by many factors such as weather and the like when detecting the smoke blackness, and the accuracy is limited.
Disclosure of Invention
To solve at least one of the above technical problems, the present application aims to: the method, the system and the storage medium for detecting the smoke blackness are provided, so that the accuracy of detecting the smoke blackness by the image is improved.
In a first aspect, an embodiment of the present application provides:
a smoke blackness detection method comprises the following steps:
acquiring a first image, wherein the first image is an image comprising a sky area and a smoke area;
graying the first image to obtain a second image;
performing binary segmentation on the second image to obtain a first sky area and a first smoke area;
calculating the brightness of a first sky area in the first image, and compensating the brightness of pixel points of a first smoke area in the first image based on the brightness of the first sky area to obtain the brightness data of the compensated first smoke area;
carrying out brightness distribution probability statistics on the compensated brightness data of the first smoke region to obtain statistical data;
and inputting the statistical data into a classification model to obtain a first prediction result of the blackness of the smoke.
Further, the calculating the brightness of the first sky area in the first image, and compensating the brightness of the pixel point of the first smoke area in the first image based on the brightness of the first sky area to obtain the brightness data of the compensated first smoke area includes:
selecting at least two sub-regions from the first antenna region;
calculating the average brightness of the two sub-areas as the brightness of a first sky area in the first image;
performing inverse mapping operation on the brightness of a first sky area in the first image to obtain first brightness;
performing inverse mapping operation on the brightness of all pixel points in the first smoke region in the first image and subtracting the first brightness to obtain second brightness of all pixel points in the first smoke region;
and performing inverse mapping operation on the second brightness of all the pixel points in the first smoke region to obtain third brightness of all the pixel points in the first smoke region, and taking the third brightness of all the pixel points in the first smoke region as brightness data of the compensated first smoke region.
Further, the graying the first image to obtain a second image includes:
and weighting the RGB colors of the pixels of the first image, and taking the weighted result as the gray scale to obtain a second image.
Further, the binary segmentation is performed on the second image to obtain a first sky area and a first smoke area, and the method includes:
performing binary segmentation on the second image by using a maximum inter-class variance method;
wherein, the dividing threshold T of the maximum inter-class variance method is the dividing threshold which enables the K value to be maximum; the above-mentionedσ2Showing the difference between the two parts after the segmentation,the distribution of the internal gray value of each part after division is shown.
Furthermore, the statistical data are 256-dimensional vectors, and the vectors represent the probability of the pixel points with the brightness of 0-255 appearing in the first smoke region.
Further, the classification model is obtained by training through the following method:
acquiring a training set, wherein the training set comprises a plurality of third images and marking information, and the third images comprise sky areas and smoke areas;
graying the third image to obtain a fourth image;
performing binary segmentation on the fourth image to obtain a second sky area and a second smoke area;
calculating the brightness of a second sky area in the third image, and compensating the brightness of pixel points of a second smoke area in the third image based on the brightness of the second sky area to obtain the brightness data of the compensated second smoke area;
carrying out brightness distribution probability statistics on the compensated brightness data of the second smoke region to obtain statistical data;
inputting the statistical data into a classification model to obtain a second prediction result of the smoke blackness;
and adjusting parameters of the classification model according to the error between the second prediction result and the labeling information, and performing iterative training until the classification model meets the training condition.
Further, the classification model is a support vector machine, the input of the support vector machine is a 256-dimensional vector, the output of the support vector machine is a 6-dimensional vector, and the kernel function of the support vector machine is a gaussian radial basis kernel function.
Further, the method also comprises the following steps:
counting first prediction results of a plurality of first images acquired according to a set frequency in a preset period;
and determining the grade of the smoke blackness according to the number of the first prediction result belonging to each grade.
In a second aspect, embodiments of the present application provide:
a smoke blackness detection system, comprising:
the camera is arranged in the black smoke detection area and used for shooting a first image;
and the processing module is connected with the camera through a wireless network and is used for executing the smoke blackness detection method.
In a third aspect, embodiments of the present application provide:
a storage medium storing a program which, when executed by a processor, performs the smoke blackness detection method.
The beneficial effects of the embodiment of the application are that: according to the embodiment of the application, the image comprising the sky area and the smoke area is obtained, binary segmentation is carried out after graying is carried out on the image, the sky area and the smoke area are obtained, then the luminance of the smoke area in the image is compensated based on the luminance of the sky area in the image, luminance distribution probability statistics is carried out on the luminance data of the smoke area after compensation, statistical data are obtained, the statistical data are predicted through a classification model, and the prediction result of the smoke blackness is obtained.
Drawings
Fig. 1 is a flowchart of a smoke blackness detection method according to an embodiment of the present application;
FIG. 2 is a flowchart of a method for training a classification model according to an embodiment of the present application;
fig. 3 is a block diagram of a smoke blackness detection system according to an embodiment of the present application.
Detailed Description
The present application is further described with reference to the following figures and specific examples. The described embodiments should not be considered as limiting the present application, and all other embodiments obtained by a person skilled in the art without making any inventive step are within the scope of protection of the present application.
In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of the present application only and is not intended to be limiting of the application.
First, the terms appearing in the present application are explained:
a support vector machine: the (SVM) is a generalized linear classifier (generalized linear classifier) that binary classifies data according to a supervised learning (superiattice) method, and a decision boundary of the SVM is a maximum-margin hyperplane (maximum-margin hyperplane) that solves for a learning sample.
And (3) binarization segmentation: the method is an operation of segmenting the image according to the numerical value of the pixel point of the image and the segmentation threshold value.
Referring to fig. 1, the present embodiment discloses a smoke blackness detection method, which is applied to a computer device, where the computer device may be an independent computer, a distributed server, or the like, and the method of the present embodiment includes the following steps:
The first image may be captured by a camera on site in real time, or the first image may be an image stored in a database. The shooting angle of the field camera can be adjusted, so that the first image obtained by shooting only comprises a sky area and a smoke area.
And 120, graying the first image to obtain a second image.
Wherein the steps are as follows: and weighting the RGB colors of the pixels of the first image, and taking the weighted result as the gray scale to obtain a second image.
In this embodiment, the purpose of graying is to convert the RGB color of each pixel into grayscale, and the grayscale of the pixel can be calculated by the following formula when converting, where Y is 0.299R +0.587G +0.114B, and in the formula, R is a red value, G is a green value, and B is a blue value. Wherein, the values of RGB in the first image are positive numbers of 0-255. And the value obtained after the final calculation of the integer Y is also 0-255.
After the processing of the step, the value range of each pixel point in the second image is a positive number of 0-255.
And step 130, performing binary segmentation on the second image to obtain a first sky area and a first smoke area.
In this embodiment, the second image is divided into two values by using a maximum inter-class variance method.
Specifically, the division threshold T of the maximum inter-class variance method is a division threshold that maximizes the K value; the above-mentionedσ2Showing the difference between the two parts after the segmentation,expressing the distribution of internal gray value of each part after divisionThe method is described.
In this embodiment, two mask maps (masks) are obtained after binary segmentation in general. The size of the mask pattern is the same as the original size, the pixel points of the mask pattern are represented by 1 and 0, and the portions represented by the mask pattern are obtained by comparing the mask pattern with the original size.
For example, in the case of binary division, the division threshold is 100, all pixels equal to or smaller than 100 are set to 1, and other pixels are set to 0, thereby obtaining a mask map of a region having a gradation equal to or smaller than the division value. Similarly, all the pixels larger than 100 are set to be 1, and other pixels are set to be 0, so that a mask image of an area with the gray scale larger than the division value is obtained.
Of course, the image may alternatively be segmented, such as by neural network segmentation.
Specifically, in this embodiment, the step specifically includes:
S140A, selecting at least two sub-regions from the first antenna region. In this embodiment, the size of the sub-region may be set, for example, a 20 × 20 pixel region is selected. The selection of the sub-regions may be random or one on each side of the smoke image region.
And S140B, calculating the average brightness of the two sub-areas as the brightness of the first sky area in the first image.
Of course, in some embodiments, the brightness of the entire sky region may also be averaged. In this embodiment, the gradation may be taken as the luminance, and therefore, the luminance of the sky region in the first image is equal to the gradation of the sky region in the second image. Of course, when calculating the brightness, a formula different from the gray scale calculation formula may be used, and the basic principle is to weight the RGB values.
S140C, performing inverse mapping operation on the brightness of the first sky area in the first image to obtain a first brightness. In the present embodiment, the inverse mapping operation is to subtract the preset maximum value from the value, for example, the maximum value of the luminance is 255, and the value obtained by performing the inverse mapping operation on the luminance with the value of 200 is (255-.
S140D, performing inverse mapping operation on the brightness of all pixel points in the first smoke region in the first image and subtracting the first brightness to obtain second brightness of all pixel points in the first smoke region.
In this step, the inverse mapping operation is performed on all the pixel points in the first smoke region. After the inverse mapping operation, since the color of the smoke is darker relative to the color of the sky, the subtraction of the luminance of the sky area after the inverse mapping is performed is positive.
S140E, performing inverse mapping operation on the second brightness of all the pixels in the first smoke region to obtain third brightness of all the pixels in the first smoke region, and taking the third brightness of all the pixels in the first smoke region as the brightness data of the compensated first smoke region.
In this step, the brightness data should be understood as a set of compensated brightness values of each pixel point in the smoke region.
And 150, carrying out brightness distribution probability statistics on the compensated brightness data of the first smoke region to obtain statistical data.
In this embodiment, the value range of each pixel in the luminance data finally obtained in step 140 is an integer of 0 to 255, so the statistical data is a 256-dimensional vector, and 256 dimensions in the vector respectively represent the number of pixels with values of 0 to 255. The vector may be represented by P ═ (l)1、l2、……l256) And (4) showing. Wherein lnRepresenting the probability of the occurrence of the pixel point with the brightness value of n-1. Since the vector represents a probability distribution, the length of the vector is independent of the size of the black smoke region.
And 160, inputting the statistical data into a classification model to obtain a first prediction result of the blackness of the smoke.
In the present embodiment, the classification model employs a Support Vector Machine (SVM) as the classification model. The statistical data obtained in step 150, i.e. the 256-dimensional vector, is input into a support vector machine trained with a large amount of data, so as to obtain a prediction result. The prediction result is a preset classification level. In the present embodiment, the grades are classified into 0-5 grades, and a total of 6 grades.
Therefore, in the present embodiment, the input of the support vector machine is a 256-dimensional vector, and the output is a 6-dimensional vector, and the kernel function of the support vector machine in the present embodiment is a gaussian radial basis kernel function.
The gaussian radial basis kernel function can be expressed as: k (x, x)i)=exp(-‖x-xi‖2/2σ2)。xiIs the kernel function center, σ is the width parameter of the function, and x is the input value.
It should be understood that the main reason for selecting the support vector machine in the present embodiment is that the support vector machine has a shorter training time and a lower cost compared to the deep neural network. Thus, in practice, neural networks such as CNNs or RNNs may also be selected as classification models.
Based on the description of the above embodiment, it can be known that the influence of the ambient brightness on the recognition of the smoke blackness can be reduced by compensating the brightness of the smoke region based on the brightness of the sky region in the same photo, so that the detection accuracy is increased.
In some embodiments, the classification model is trained by:
acquiring a training set, wherein the training set comprises a plurality of third images and marking information, and the third images comprise sky areas and smoke areas;
graying the third image to obtain a fourth image;
performing binary segmentation on the fourth image to obtain a second sky area and a second smoke area;
calculating the brightness of a second sky area in the third image, and compensating the brightness of pixel points of a second smoke area in the third image based on the brightness of the second sky area to obtain the brightness data of the compensated second smoke area;
carrying out brightness distribution probability statistics on the compensated brightness data of the second smoke region to obtain statistical data;
inputting the statistical data into a classification model to obtain a second prediction result of the smoke blackness;
and adjusting parameters of the classification model according to the error between the second prediction result and the labeling information, and performing iterative training until the classification model meets the training condition.
In this embodiment, the process of model training is the same as the prediction process, except that the third image is an artificially labeled image. In the training process, the third image is subjected to prediction processing to obtain a prediction result, parameters of the classification model are corrected based on an error between the prediction result and the labeled value, so that the final model has good fitting degree with training data, and the classification model is trained through a plurality of third images until the fitting degree of the model reaches a preset value.
Of course, in some embodiments, the classification model may also be trained based on statistical data and corresponding labeled values obtained from the actual image processing as training data.
Referring to fig. 2, the present embodiment provides a method for training a classification model, including the following steps:
(1) and a cloud platform database is built in the remote monitoring module and used for storing the obtained flue gas image, the corresponding vector P and the corresponding blackness level.
(2) Vector P (l) for all smoke images of database1,l2,…,l256) Standardization is carried out, i.e. |i’=(li-lmin)/(lmax-lmin) The normalized vector P' is obtained. Wherein lminIs min (l)n) I.e. the minimum value of l in the vector, lmaxIs max (l)n) I.e. the maximum value of l in the vector.
(3) The Gaussian radial basis kernel is chosen as the kernel of the support vector machine, i.e., K (x, x)i)=exp(-‖x-xi‖2/2σ2) The normalized vector is used as an input parameter, the corresponding blackness level of the image is used as an output vector, and the blackness of the smoke has 6 levels, namely 0 level, 1 level, 2 level, 3 level, 4 level and 5 level.
(4) And training by adopting images in a database based on the established support vector machine, obtaining the optimal classification parameters of the support vector machine by using a cross validation method until the blackness classification accuracy reaches more than 99.9%, and stopping training to obtain a smoke blackness support vector machine classification fitting model, namely a classification model.
(5) The database can continuously collect and store images obtained by field detection, and retrain the images to obtain a new fitting model so as to improve the grade classification precision of the smoke blackness. Namely, the images and the corresponding prediction results are used as training samples to continue training.
In some embodiments, further comprising the steps of:
counting first prediction results of a plurality of first images acquired according to a set frequency in a preset period, wherein the first images are images shot in real time. And the first image is shot at the detection site according to the set frequency.
And determining the grade of the smoke blackness according to the number of the first prediction result belonging to each grade.
Specifically, taking a preset week of 30 minutes and a set frequency of every 15 seconds as an example, when 5-level blackness appears in 30 minutes, the smoke is counted by 5 levels; when the blackness of the smoke of more than 4 grades appears within 30 minutes and exceeds 2 minutes, the blackness of the smoke is calculated according to 4 grades; when the blackness of the smoke of more than 3 grades is over 2 minutes within 30 minutes, the blackness of the smoke is calculated according to 3 grades; when the smoke blackness of more than 2 grades appears within 30 minutes and exceeds 2 minutes, the smoke blackness is calculated according to 2 grades; when the smoke blackness of more than 1 grade appears in 30 minutes and exceeds 2 minutes, the smoke blackness is calculated according to 1 grade; and when the smoke blackness of less than 1 grade appears within 30 minutes and exceeds 28 minutes, the smoke blackness is counted as less than 1 grade.
Wherein the number of minutes present is 15 seconds.
By adopting the statistical mode, the condition of the black smoke can be analyzed in a certain time span, the analysis frequency is not too high, and the requirement of the analysis on the network and hardware can be reduced by adopting the mode.
Referring to fig. 3, the embodiment discloses a smoke blackness detection system, which includes an on-site detection module and a remote monitoring module.
The on-site detection module comprises a camera module, a data storage and analysis module and a remote transmission module, and works as follows:
(1) and (4) photographing the smoke discharged from the chimney by using a camera of the camera sub-module and storing the photographed smoke, wherein the photographing frequency is once every 1 second.
(2) The data storage and analysis submodule is responsible for storing and analyzing the read data and automatically giving out the blackness level of the smoke.
(3) The remote transmission sub-module is responsible for transmitting the shot photos and the detection results to the remote monitoring module in real time, and the transmission mode adopts a 5G wireless transmission mode.
The remote monitoring module comprises an intelligent analysis submodule, a manual analysis submodule and a report generation submodule, and works as follows:
(1) the intelligent analysis submodule stores the pictures and the detection data transmitted on site, and processes and analyzes the pictures to obtain the smoke blackness grade of the pictures.
(2) And the manual analysis submodule is used for a remote compound expert to judge the shot picture and recheck the detection result in time.
(3) And the report generation module is responsible for automatically generating reports for the detection data and results, calculating the average value of every 15 seconds and continuously recording 60 readings.
The embodiment discloses a flue gas blackness detecting system, includes:
the camera is arranged in the black smoke detection area and used for shooting a first image;
and the processing module is connected with the camera through a wireless network and is used for executing the smoke blackness detection method described in the embodiment.
The embodiment discloses a storage medium, which stores a program, and the program is executed by a processor to carry out the smoke blackness detection method.
The step numbers in the above method embodiments are set for convenience of illustration only, the order between the steps is not limited at all, and the execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.
While the present application has been described with reference to the preferred embodiments, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. The smoke blackness detection method is characterized by comprising the following steps of:
acquiring a first image, wherein the first image is an image comprising a sky area and a smoke area;
graying the first image to obtain a second image;
performing binary segmentation on the second image to obtain a first sky area and a first smoke area;
calculating the brightness of a first sky area in the first image, and compensating the brightness of pixel points of a first smoke area in the first image based on the brightness of the first sky area to obtain the brightness data of the compensated first smoke area;
carrying out brightness distribution probability statistics on the compensated brightness data of the first smoke region to obtain statistical data;
and inputting the statistical data into a classification model to obtain a first prediction result of the blackness of the smoke.
2. The method according to claim 1, wherein the calculating the brightness of the first sky region in the first image, and compensating the brightness of the pixel point of the first smoke region in the first image based on the brightness of the first sky region to obtain the brightness data of the compensated first smoke region includes:
selecting at least two sub-regions from the first antenna region;
calculating the average brightness of the two sub-areas as the brightness of a first sky area in the first image;
performing inverse mapping operation on the brightness of a first sky area in the first image to obtain first brightness;
performing inverse mapping operation on the brightness of all pixel points in the first smoke region in the first image and subtracting the first brightness to obtain second brightness of all pixel points in the first smoke region;
and performing inverse mapping operation on the second brightness of all the pixel points in the first smoke region to obtain third brightness of all the pixel points in the first smoke region, and taking the third brightness of all the pixel points in the first smoke region as brightness data of the compensated first smoke region.
3. The method for detecting smoke blackness according to claim 1, wherein the graying the first image to obtain a second image comprises:
and weighting the RGB colors of the pixels of the first image, and taking the weighted result as the gray scale to obtain a second image.
4. The method for detecting smoke blackness according to claim 1, wherein the performing binary segmentation on the second image to obtain a first sky region and a first smoke region comprises:
performing binary segmentation on the second image by using a maximum inter-class variance method;
wherein, the dividing threshold T of the maximum inter-class variance method is the dividing threshold which enables the K value to be maximum; the above-mentionedσ2Showing the difference between the two parts after the segmentation,the distribution of the internal gray value of each part after division is shown.
5. The smoke blackness detection method according to claim 1, wherein the statistical data is a 256-dimensional vector representing a probability that a pixel with a brightness of 0-255 appears in the first smoke region.
6. The smoke blackness detection method according to claim 1, wherein the classification model is trained by the following method:
acquiring a training set, wherein the training set comprises a plurality of third images and marking information, and the third images comprise sky areas and smoke areas;
graying the third image to obtain a fourth image;
performing binary segmentation on the fourth image to obtain a second sky area and a second smoke area;
calculating the brightness of a second sky area in the third image, and compensating the brightness of pixel points of a second smoke area in the third image based on the brightness of the second sky area to obtain the brightness data of the compensated second smoke area;
carrying out brightness distribution probability statistics on the compensated brightness data of the second smoke region to obtain statistical data;
inputting the statistical data into a classification model to obtain a second prediction result of the smoke blackness;
and adjusting parameters of the classification model according to the error between the second prediction result and the labeling information, and performing iterative training until the classification model meets the training condition.
7. The smoke blackness detection method according to claim 1, wherein the classification model is a support vector machine, an input of the support vector machine is a 256-dimensional vector, an output of the support vector machine is a 6-dimensional vector, and a kernel function of the support vector machine is a gaussian radial basis kernel function.
8. The smoke blackness detection method according to any one of claims 1 to 7, further comprising the steps of:
counting first prediction results of a plurality of first images acquired according to a set frequency in a preset period; the first image is shot at a detection site according to the set frequency;
and determining the grade of the smoke blackness according to the number of the first prediction result belonging to each grade.
9. A smoke blackness detection system, comprising:
the camera is arranged in the black smoke detection area and used for shooting a first image;
a processing module connected with the camera via a wireless network, the processing module being configured to perform the method according to any one of claims 1-8.
10. A storage medium storing a program, characterized in that the program, when executed by a processor, implements the method according to any one of claims 1-8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010326812.0A CN111598840A (en) | 2020-04-23 | 2020-04-23 | Smoke blackness detection method and system and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010326812.0A CN111598840A (en) | 2020-04-23 | 2020-04-23 | Smoke blackness detection method and system and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111598840A true CN111598840A (en) | 2020-08-28 |
Family
ID=72191945
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010326812.0A Pending CN111598840A (en) | 2020-04-23 | 2020-04-23 | Smoke blackness detection method and system and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111598840A (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080137906A1 (en) * | 2006-12-12 | 2008-06-12 | Industrial Technology Research Institute | Smoke Detecting Method And Device |
CN102156099A (en) * | 2011-01-17 | 2011-08-17 | 深圳市朗驰欣创科技有限公司 | Method and system for detecting atmospheric pollutants |
CN102456142A (en) * | 2010-11-02 | 2012-05-16 | 上海宝信软件股份有限公司 | Analysis method for smoke blackness based on computer vision |
CN104899895A (en) * | 2015-05-19 | 2015-09-09 | 三峡大学 | Detection method of trace complexity of mobile targets of fire video in channel of power transmission line |
CN109241983A (en) * | 2018-09-13 | 2019-01-18 | 电子科技大学 | A kind of cigarette image-recognizing method of image procossing in conjunction with neural network |
-
2020
- 2020-04-23 CN CN202010326812.0A patent/CN111598840A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080137906A1 (en) * | 2006-12-12 | 2008-06-12 | Industrial Technology Research Institute | Smoke Detecting Method And Device |
CN102456142A (en) * | 2010-11-02 | 2012-05-16 | 上海宝信软件股份有限公司 | Analysis method for smoke blackness based on computer vision |
CN102156099A (en) * | 2011-01-17 | 2011-08-17 | 深圳市朗驰欣创科技有限公司 | Method and system for detecting atmospheric pollutants |
CN104899895A (en) * | 2015-05-19 | 2015-09-09 | 三峡大学 | Detection method of trace complexity of mobile targets of fire video in channel of power transmission line |
CN109241983A (en) * | 2018-09-13 | 2019-01-18 | 电子科技大学 | A kind of cigarette image-recognizing method of image procossing in conjunction with neural network |
Non-Patent Citations (1)
Title |
---|
包信宗: "烟气林格曼黑度远程视频监测系统的应用研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108765423B (en) | Convolutional neural network training method and device | |
CN107403175A (en) | Visual tracking method and Visual Tracking System under a kind of movement background | |
CN111292264A (en) | Image high dynamic range reconstruction method based on deep learning | |
CN112241699A (en) | Object defect category identification method and device, computer equipment and storage medium | |
CN108830197A (en) | Image processing method, device, computer equipment and storage medium | |
CN111260645B (en) | Tampered image detection method and system based on block classification deep learning | |
CN113869162A (en) | Violation identification method and system based on artificial intelligence | |
CN111126115A (en) | Violence sorting behavior identification method and device | |
CN113435407B (en) | Small target identification method and device for power transmission system | |
CN116993738B (en) | Video quality evaluation method and system based on deep learning | |
AU2017443986B2 (en) | Color adaptation using adversarial training networks | |
CN111274964B (en) | Detection method for analyzing water surface pollutants based on visual saliency of unmanned aerial vehicle | |
CN113421223A (en) | Industrial product surface defect detection method based on deep learning and Gaussian mixture | |
CN109658405B (en) | Image data quality control method and system in crop live-action observation | |
CN116228780A (en) | Silicon wafer defect detection method and system based on computer vision | |
CN111598840A (en) | Smoke blackness detection method and system and storage medium | |
CN111402185A (en) | Image detection method and device | |
CN112750113B (en) | Glass bottle defect detection method and device based on deep learning and linear detection | |
CN114648738A (en) | Image identification system and method based on Internet of things and edge calculation | |
Xiao et al. | Blind Quality Metric via Measurement of Contrast, Texture, and Colour in Night-Time Scenario. | |
CN110852314A (en) | Article detection network method based on camera projection model | |
CN111242047A (en) | Image processing method and apparatus, electronic device, and computer-readable storage medium | |
CN116563770B (en) | Method, device, equipment and medium for detecting vehicle color | |
CN117237383B (en) | Intelligent door and window control method and system based on indoor environment | |
Baifeng et al. | Research on nameplate image recognition algorithm based on R-CNN and SSD deep learning detection methods |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200828 |
|
RJ01 | Rejection of invention patent application after publication |