CN114973104A - Dynamic flame detection algorithm and system based on video image - Google Patents
Dynamic flame detection algorithm and system based on video image Download PDFInfo
- Publication number
- CN114973104A CN114973104A CN202210705670.8A CN202210705670A CN114973104A CN 114973104 A CN114973104 A CN 114973104A CN 202210705670 A CN202210705670 A CN 202210705670A CN 114973104 A CN114973104 A CN 114973104A
- Authority
- CN
- China
- Prior art keywords
- flame
- image
- video
- images
- suspected
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 29
- 238000012545 processing Methods 0.000 claims abstract description 24
- 238000000034 method Methods 0.000 claims abstract description 9
- 238000012544 monitoring process Methods 0.000 claims abstract description 8
- 238000011410 subtraction method Methods 0.000 claims abstract description 5
- 238000012706 support-vector machine Methods 0.000 claims abstract description 5
- 238000000605 extraction Methods 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 2
- 230000000877 morphologic effect Effects 0.000 claims description 2
- 238000012360 testing method Methods 0.000 claims description 2
- 230000005540 biological transmission Effects 0.000 claims 1
- 238000009529 body temperature measurement Methods 0.000 claims 1
- 230000004927 fusion Effects 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Images
Classifications
-
- 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
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J5/00—Radiation pyrometry, e.g. infrared or optical thermometry
- G01J5/0014—Radiation pyrometry, e.g. infrared or optical thermometry for sensing the radiation from gases, flames
- G01J5/0018—Flames, plasma or welding
-
- 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
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- 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
- G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
-
- 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
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/54—Extraction of image or video features relating to texture
-
- 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
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- 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
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B17/00—Fire alarms; Alarms responsive to explosion
- G08B17/12—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions
- G08B17/125—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions by using a video camera to detect fire or smoke
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Emergency Management (AREA)
- Computational Linguistics (AREA)
- Plasma & Fusion (AREA)
- Business, Economics & Management (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Image Analysis (AREA)
- Fire-Detection Mechanisms (AREA)
Abstract
The invention provides a dynamic flame detection algorithm and a dynamic flame detection system based on video images, and provides a video flame detection algorithm combining multiple fusion features and a support vector machine. The method comprises the following steps: acquiring video images of a monitoring place, performing image processing on the acquired continuous N frames of video images, performing image processing on input images and outputting the images; detecting a suspected flame area by adopting a background subtraction method and a flame color model, and extracting the suspected flame area; acquiring characteristics of the suspected area, such as dynamic, geometric, texture and the like; and (4) the trained SVM is utilized to complete recognition by fusing multiple characteristic quantities, and whether a fire disaster occurs or not can be determined by combining the temperature change of the infrared temperature sensor. The algorithm has good detection effect and short consumed time for the common phenomena of low detection rate and high false alarm rate in the conventional fire detection method.
Description
Technical Field
The invention relates to the technical field of flame detection, in particular to image type fire detection, and specifically relates to a dynamic flame detection algorithm and system based on video images.
Background
Fire serves as a very harmful safety event, and poses a great threat to the economy, the environment and the life safety of people. The economic loss caused by the fire in China reaches billions of yuan every year, the number of casualties is higher than thousands of people, and the importance of preventing and treating the fire is increasingly obvious. The fire condition can be found and controlled as early as possible, the further expansion of the fire can be prevented, and unnecessary casualties and property loss can be reduced to a certain extent.
In recent years, with the rapid development of image processing technology, more and more researchers are beginning to study flame detection based on video analysis. However, the existing video image processing technology generally has the problem that the effectiveness of the flame characteristic value is low for a single color characteristic and a dynamic characteristic.
In the existing flame identification classifier, the flame identification processing method is complex, the software operation amount is overlarge, the condition of misjudgment often occurs in the flame detection identification process, and the positioning precision of the flame is low.
Disclosure of Invention
In view of the above, the present invention is directed to a dynamic flame detection algorithm and system based on video images. The invention adopts the flame identification classifier to identify the flame by means of video image processing, and can quickly and effectively detect the flame and give an alarm at the initial stage of the fire. In order to achieve the purpose, the invention adopts the following technical means:
a dynamic flame detection algorithm and system based on video images comprises the following units:
the system comprises a video image acquisition unit, a video image acquisition unit and a video image processing unit, wherein the video image acquisition unit is used for acquiring a video image of a monitoring area, all cameras in all visible light ranges at present are provided with video sensors for detecting RGB color formats, and the acquired image information is an RGB color image;
the image processing unit is used for carrying out image processing on the acquired continuous N frames of video images, carrying out image processing on the input images and outputting the images;
the image information extraction unit is used for determining a suspected flame area, extracting the area change rate, the color moment, the circularity and the texture characteristic value of the processed video image and locking a suspected target; extracting and recording the change trend of the numerical value;
and the identification alarm unit is used for identifying and classifying the video images, locking the suspected target, judging whether flame exists in the suspected area or not, and judging whether the suspected area meets the alarm condition or not by combining the temperature change of the infrared temperature sensor.
A dynamic flame detection algorithm and system based on video images comprises the following steps:
acquiring a video image through a monitoring camera;
converting the image into a YCbCr color space model and carrying out gray processing;
carrying out filtering denoising and gray level binarization on a currently acquired black-and-white video, a video in an RGB color format and a monitoring video in a YCbCr color space by using an image processing technology, and processing video images of continuous N frames by using an edge operator for preliminarily determining a suspected flame region;
carrying out suspected flame area detection by adopting a background subtraction method and a flame color model;
extracting a characteristic value of a suspected flame area, extracting color moment, circularity, texture characteristic and area change of the flame area according to color characteristic, morphological characteristic and dynamic characteristic of the flame, and using the extracted characteristic value as characteristic input quantity of a flame identification classifier;
inputting the characteristic values into a support vector machine to identify flames according to the obtained characteristic values, and locking a suspected target by combining the temperature change of an infrared temperature sensor;
and judging whether flame exists in the suspected area or not and whether the suspected area meets the alarm condition or not.
Drawings
FIG. 1 is a flow chart of a method for image pre-processing according to the present invention;
FIG. 2 is a flow chart of a flame characteristic quantity extraction method provided by the present invention;
FIG. 3 is a schematic diagram of a process for extracting color moments based on a color space according to the present invention;
FIG. 4 is a schematic view of a process for changing the flame area according to the present invention;
FIG. 5 is a schematic flow chart of SVM model establishment based on parameter optimization according to the present invention;
FIG. 6 is a schematic diagram of a dynamic flame detection algorithm and system based on video images according to the present invention.
Detailed Description
In order to better understand the technical content of the present invention, specific embodiments are described below with reference to the accompanying drawings.
FIG. 1 is a flow chart of a method of image pre-processing, comprising the steps of:
acquiring a video image, wherein the acquired image information is an RGB color image and a black-and-white video image, and setting a frame rate of an image sequence in the video, wherein the frame rate of the image sequence in the embodiment is set to 25 fps;
preprocessing an image, and processing the acquired continuous N frames of video images, wherein N is a natural number greater than or equal to 2;
the image conversion is used for converting the acquired RGB video image into a YCbCr image, and the conversion formula is as follows:
the filtering and denoising is used for denoising the acquired image, and the median filtering of the acquired image is defined as:
wherein f (x, y) is an image to be processed, g (x, y) is an image after median filtering, S is a selected two-dimensional window, and the value of S is set to be a window of 3 x 3;
the gray level binarization is used for carrying out binarization processing on the acquired image, and the maximum inter-class variance method (Otsu) divides the image into A, B parts by using a threshold value T to acquire a flame gray level image, and the calculation is as follows:
σ 2 =ω 0 (μ 0 -μ) 2 +ω 1 (μ 1 -μ) 2
wherein, ω is 0 Is the proportion of the target image pixel to the whole image, omega 1 Is the proportion of background image pixels to the whole image, mu 0 Is the mean gray value, mu, of the target image 1 The average gray value of the background image is used, and mu is the average gray value of the whole image;
edge detection is used to preliminarily determine areas of suspected flame.
Fig. 2 is a flow chart of flame characteristic quantity extraction, which includes the following steps:
detecting a suspected flame area by adopting a background subtraction method and a flame color model;
the background subtraction method comprises the steps of establishing a background statistical model to enable the background model at each moment to better approximate to a real environment background, and then solving the difference between a current image and a background image to extract a motion foreground;
further, a suspected flame area is determined based on a comprehensive flame model of RGB and YCbCr color spaces, and the identification conditions comprise:
Rule1:R>G>B
Rule2:R>R T
Rule3:S>(255-R)×S T /R T
If(Rule1)And(Rule2)And(Rule3)And(Rule4)=True
Then fire pixel
Else not fire pixel
in the formula, R, G, B represents the values of three channels of red, green and blue colors, S, respectively, of one pixel T Represents a threshold value on the R channel, and the value range is [55,65 ]],R T A threshold value representing the color saturation, which is in the range of [115,135 ]]And tau is an empirical constant, and is taken as 40 according to empirical values obtained by a series of flame image tests.
The circularity of the object can be calculated by the perimeter of the object outline and the area of the region where the object is located:
C=P 2 /4πS
wherein S is the area of the region where the object is located, and P is the perimeter of the region where the object is located. The more complex the object contour, the larger the value of circularity. The edge complexity of objects such as car lamps, street lamps, incandescent lamps and the like similar to the flame color is not high, the shape is close to a circle, the circularity is approximately 1, the structure of the flame edge is relatively complex, and the circularity is far greater than 1, so that the circularity is used as the criterion for flame identification.
The texture features are analyzed using a gray level co-occurrence matrix (GLCM).
Fig. 3 is a schematic diagram of a process of extracting color moments based on a color space, which is used for performing learning training input, where the color moment feature vectors are represented as follows:
R=[R Mean, R Var ,G Mean, G Var ,B Mean, B Var l
Y=[Y Mean, Y Var, Cb Mean, Cb Var ,Cr Mean, Cr Var l
FIG. 4 is a schematic representation of the flow chart of the change in flame area, with the area growth rate defined as:
wherein, A (n +1), A (n) is the area of the suspected area of the flame of two adjacent frames, eps is a minimum value to prevent the condition that the denominator is 0, and through analysis of a large amount of experimental data, the area change rate of the flame usually changes in the interval of [0.1-0.4 ].
Fig. 5 is a flowchart illustrating SVM model building based on parameter optimization, which specifically includes the steps of:
acquiring a video image;
taking all flame images in the video as positive samples of a training set;
taking the image without flame in the video as a negative sample of the training set;
carrying out normalization processing on the sample data;
ensuring the classification performance of the SVM, selecting a radial basis kernel function, selecting a kernel function parameter of 0.2 and a penalty factor of 10;
and carrying out model detection.
FIG. 6 is a schematic diagram of a flame detection alarm system based on video image processing according to the present invention, which detects all flame target areas and inputs the obtained feature values into a support vector machine for classification according to the above steps;
after all the flame target areas are detected, whether a fire disaster occurs can be determined by combining the temperature change of the infrared temperature sensor.
The above description is only an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can make various changes in the technical scope of the present invention disclosed in the present invention according to the actual situation and shall be covered by the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (9)
1. A dynamic flame detection algorithm and a system based on video images are used for identifying suspicious flames in the video images and monitoring the occurrence of the flames in real-time feedback video images. The device is characterized by comprising a video image acquisition unit, an image processing unit, an image information extraction unit and an identification alarm unit.
The system comprises a video image acquisition unit, a video image acquisition unit and a video image processing unit, wherein the video image acquisition unit is used for acquiring a video image of a monitoring area, all cameras in all visible light ranges at present are provided with video sensors for detecting RGB color formats, and the acquired image information is an RGB color image;
the image processing unit is used for carrying out image processing on the acquired continuous N frames of video images, carrying out image processing on the input images and outputting the images;
determining a target area, extracting a suspected flame area by adopting a background subtraction method and suspected flame area detection of a flame color model,
the image information extraction unit is used for extracting the characteristic values in continuous frames of the processed video images and recording the change trend of the numerical values;
and the identification alarm unit is used for identifying and classifying the video images, locking the suspected target and judging whether flame exists in the suspected area or not and whether the suspected area meets the alarm condition or not.
2. The video image dynamic flame detection algorithm and system of claim 1, comprising:
selecting the continuous N frames of images, wherein N is a natural number which is more than or equal to 2;
before preprocessing the image, simultaneously combining the acquirable black and white image to carry out auxiliary detection;
measuring temperature change by combining an infrared temperature sensor;
the video image acquisition unit is composed of a field monitoring camera and a wireless transmission infrared temperature measurement camera and provides a video image of a monitoring place.
3. The video image dynamic flame detection algorithm and system of claim 1, wherein the pre-processing process comprises:
the image conversion is used for converting the acquired RGB video image into a YCbCr image;
filtering and denoising, which is used for denoising the acquired image;
carrying out grey level binarization for carrying out binarization processing on the acquired image;
and detecting edges, wherein the edges are used for preliminarily judging the suspected flame area.
4. The video image dynamic flame detection algorithm and system according to claim 1, wherein the video image to be detected is processed based on N consecutive frames of detected flame target images, and the processed flame target images are inputted into the image information extraction unit to obtain the feature values of the detected flame target in the N consecutive frames of images.
5. A flame feature extraction system, comprising:
one or more image features;
the image processing program according to claims 3 to 5, wherein the image information extraction unit classifies the image of the object of the image processing by extracting feature amounts from the images of the object of the image processing of the consecutive N frames.
6. The image information extraction unit according to claim 5, wherein the area change rate, color matrix, circularity, texture feature, and the like are extracted from the color feature, morphological feature, and dynamic feature of the flame as the feature input amount of the flame recognition classifier.
7. The identification and judgment unit according to claim 1, wherein the eigenvalue obtained according to claim 6 is input into a support vector machine to perform flame identification, lock a suspected target, and judge whether flame exists in a suspected area or not, and whether the suspected area meets an alarm condition or not.
8. The support vector machine classifier model of claim 7,
acquiring a video image;
taking all flame images in the video as positive samples of a training set;
taking the image without flame in the video as a negative sample of the training set;
and (3) extracting characteristic values by taking a training set and a test set formed by the positive samples and the negative samples as objects, marking the flame video images as the positive samples, and marking the flameless video images as the negative samples to obtain a flame classification detection model.
9. The video image dynamic flame detection algorithm and system according to claim 1, wherein the video image input is inputted into the flame classification detection model based on the images of the flame targets detected in the consecutive N frames after the above claims, and the target detected in the consecutive N frames of images is determined to be a real flame and an alarm is given.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210705670.8A CN114973104A (en) | 2022-06-21 | 2022-06-21 | Dynamic flame detection algorithm and system based on video image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210705670.8A CN114973104A (en) | 2022-06-21 | 2022-06-21 | Dynamic flame detection algorithm and system based on video image |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114973104A true CN114973104A (en) | 2022-08-30 |
Family
ID=82964724
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210705670.8A Pending CN114973104A (en) | 2022-06-21 | 2022-06-21 | Dynamic flame detection algorithm and system based on video image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114973104A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115376268A (en) * | 2022-10-21 | 2022-11-22 | 山东太平天下智慧科技有限公司 | Monitoring alarm fire-fighting linkage system based on image recognition |
-
2022
- 2022-06-21 CN CN202210705670.8A patent/CN114973104A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115376268A (en) * | 2022-10-21 | 2022-11-22 | 山东太平天下智慧科技有限公司 | Monitoring alarm fire-fighting linkage system based on image recognition |
CN115376268B (en) * | 2022-10-21 | 2023-02-28 | 山东太平天下智慧科技有限公司 | Monitoring alarm fire-fighting linkage system based on image recognition |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110516609B (en) | Fire disaster video detection and early warning method based on image multi-feature fusion | |
CN109165577B (en) | Early forest fire detection method based on video image | |
US10692216B2 (en) | Colony contrast gathering | |
CN108537215B (en) | Flame detection method based on image target detection | |
CN110135269B (en) | Fire image detection method based on mixed color model and neural network | |
CN111666834A (en) | Forest fire automatic monitoring and recognizing system and method based on image recognition technology | |
WO2021000524A1 (en) | Hole protection cap detection method and apparatus, computer device and storage medium | |
CN110427825B (en) | Video flame identification method based on fusion of key frame and fast support vector machine | |
CN109948566B (en) | Double-flow face anti-fraud detection method based on weight fusion and feature selection | |
CN112149543B (en) | Building dust recognition system and method based on computer vision | |
CN108563979B (en) | Method for judging rice blast disease conditions based on aerial farmland images | |
CN109918971B (en) | Method and device for detecting number of people in monitoring video | |
CN103761529A (en) | Open fire detection method and system based on multicolor models and rectangular features | |
CN110033040B (en) | Flame identification method, system, medium and equipment | |
CN113192038B (en) | Method for recognizing and monitoring abnormal smoke and fire in existing flame environment based on deep learning | |
CN112396011A (en) | Face recognition system based on video image heart rate detection and living body detection | |
CN111127810A (en) | Automatic alarming method and system for open fire of machine room | |
CN110057820B (en) | Method, system and storage medium for on-line detection of chlorine-hydrogen ratio of hydrogen chloride synthesis furnace | |
CN114973104A (en) | Dynamic flame detection algorithm and system based on video image | |
CN113744326B (en) | Fire detection method based on seed region growth rule in YCRCB color space | |
CN107704818A (en) | A kind of fire detection system based on video image | |
CN107038690A (en) | A kind of motion shadow removal method based on multi-feature fusion | |
CN113221763A (en) | Flame identification method based on video image brightness | |
CN112364884A (en) | Method for detecting moving object | |
CN113221603A (en) | Method and device for detecting shielding of monitoring equipment by foreign matters |
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 |