CN110619293A - Flame detection method based on binocular vision - Google Patents
Flame detection method based on binocular vision Download PDFInfo
- Publication number
- CN110619293A CN110619293A CN201910840715.0A CN201910840715A CN110619293A CN 110619293 A CN110619293 A CN 110619293A CN 201910840715 A CN201910840715 A CN 201910840715A CN 110619293 A CN110619293 A CN 110619293A
- Authority
- CN
- China
- Prior art keywords
- feature vector
- visible light
- vector
- binocular vision
- detection method
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2132—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on discrimination criteria, e.g. discriminant analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- 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
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Fire-Detection Mechanisms (AREA)
Abstract
The invention discloses a flame detection method based on binocular vision, which comprises the following steps: calibrating the infrared camera and the visible light camera through a calibration plate to obtain a translation vector T and a rotation matrix R from the infrared camera to the visible light camera; setting a temperature alarm threshold, and traversing the infrared image through the temperature threshold to obtain a binary image; linking the binary image marks to obtain a target rectangular frame coordinate; converting and expanding the coordinates of the rectangular frame by a translation vector T and a rotation matrix R to obtain a visible light image target rectangular frame; taking a rectangular frame corresponding to the visible light image as a suspected target area; and extracting the color feature vector, the contour feature vector and the texture feature vector of the suspected target area to synthesize a one-dimensional column vector, and inputting the one-dimensional column vector into the SVM to obtain a judgment result.
Description
Technical Field
The invention relates to the field of binocular vision fire detection, and particularly provides a fire detection method utilizing an infrared camera and a visible light camera.
Background
The fire brings immeasurable loss to the human society, and the current technology mostly adopts binocular vision to detect the fire. Existing binocular vision detection methods generally include the following two types: 1. simultaneously carrying out video acquisition on a monitored scene by utilizing a black-and-white camera and a color camera, and respectively carrying out target identification; 2. detecting a high-temperature area of a scene by using an infrared camera, and starting a visible light camera to detect a fire after an abnormality is found; the first method for detecting the flame cannot effectively utilize the temperature information of the flame, and the second method for detecting the flame cannot utilize the coordinate information of the infrared high-temperature area, so that the false alarm rates of the two methods are high.
Therefore, how to comprehensively utilize the flame temperature data and the high-temperature coordinate information of the infrared image to judge the flame so as to reduce the false alarm rate is a problem to be solved in the field.
Disclosure of Invention
In view of this, the present invention aims to provide a flame detection method based on binocular vision, so as to solve the problem that the flame temperature data and the infrared image high-temperature coordinates cannot be effectively fused in the prior art.
The technical scheme provided by the invention is as follows: the flame detection method based on binocular vision comprises the following steps:
s1: calibrating the infrared camera and the visible light camera in a binocular manner, and obtaining a translation vector T and a rotation matrix R from the infrared camera to the visible light camera;
s2: acquiring an infrared image in a field of view by using an infrared camera, and presetting a temperature alarm threshold value;
s3: traversing the temperature value corresponding to each pixel in the infrared image, setting the pixel value of the pixel with the temperature value exceeding the temperature alarm threshold value to be 255 and setting the pixel value of the pixel with the temperature value smaller than the temperature alarm threshold value to be 0 to obtain a binary image of the infrared image, and then marking and communicating the binary image to obtain the coordinates (r1, c1) of a rectangular frame outside the high-temperature region (r2, c 2);
s4: converting the (R1, c1) (R2, c2) by using the translation vector T and the rotation matrix R to obtain the coordinates (y1, x1) (y2, x2) of the visible light image corresponding to the high-temperature region;
s5: expanding the rectangular frame determined by (y1, x1) (y2, x2) to obtain new target area coordinates (R1, C1) (R2, C2);
s6: selecting a rectangular frame determined by coordinates (R1, C1) (R2, C2) in the visible light image as a suspected target area, and then extracting a color feature vector, a contour feature vector and a texture feature vector of the suspected target area;
s7: and connecting the color feature vector, the contour feature vector and the texture feature vector end to form a one-dimensional column vector, and inputting the one-dimensional column vector into the SVM to obtain a judgment result.
Preferably, S1 includes the steps of:
s11: the calibration board is placed in a binocular vision field of the infrared camera and the visible light camera, so that the calibration board is ensured to be complete and clear in images collected by the two cameras, and the proportion of the calibration board in the infrared image is as large as possible;
s12: calibrating the two cameras respectively to obtain respective internal reference matrixes and distortion coefficient matrixes of the two cameras;
s13: and calculating a translation vector T and a rotation matrix R from the infrared camera to the visible light camera.
More preferably, in S11, the ratio of the calibration plate in the infrared image is 60% or more.
Further preferably, in S2, the value range of the temperature alarm threshold is 65 ℃ to 200 ℃.
More preferably, in S5, the rectangular frame is extended by 1/n of itself in both the length direction and the width direction, wherein n is 2-4.
Further preferably, in S6, the color feature vector is obtained by LDA dimensionality reduction of the R component of the target region.
Further preferably, in S6, the contour feature vector is obtained by the following method: and (3) operating with the target area by adopting a sobel operator as a template to obtain a contour matrix, and then performing LDA dimension reduction on the contour matrix to obtain a contour characteristic vector.
Further preferably, in S6, the texture feature vector is a 5-dimensional feature vector composed of energy, entropy, contrast, inverse variance, and correlation calculated from the gray level co-occurrence matrix.
According to the flame detection method based on binocular vision, firstly, the binocular vision is calibrated to obtain a translation vector T and a rotation matrix R from an infrared camera to a visible light camera, then, an infrared image is traversed and marked to obtain a suspected target area coordinate through temperature threshold communication, the coordinate is converted and expanded through the translation vector T and the rotation matrix R, and finally the coordinate of the suspected target area of the visible light image is locked.
Detailed Description
The invention will be further explained with reference to specific embodiments, without limiting the invention.
The invention provides a fire detection method based on binocular vision, which comprises the following steps:
s1: calibrating an infrared camera and a visible light camera in a binocular mode, and obtaining a translation vector T and a rotation matrix R from the infrared camera to the visible light camera, wherein the binocular calibration method specifically comprises the following steps:
s11: the calibration plate is placed in a binocular visual field of the infrared camera and the visible light camera, the calibration plate is ensured to be complete and clear in images collected by the two cameras, the proportion of the calibration plate in the infrared image is as large as possible, and preferably, the proportion of the calibration plate in the infrared image is more than 60%;
s12: calibrating the two cameras respectively to obtain respective internal reference matrixes and distortion coefficient matrixes of the two cameras, wherein the calibration of the two cameras can be realized by using a Matlab tool box;
s13: calculating a translation vector T and a rotation matrix R from the infrared camera to the visible light camera, wherein the translation vector T and the rotation matrix R can be realized by using a Matlab tool box;
s2: acquiring an infrared image in a field of view by using an infrared camera, and presetting a temperature alarm threshold, wherein the temperature alarm threshold is set according to the scene and the ambient temperature, and the value range is 65-200 ℃;
s3: traversing the temperature value corresponding to each pixel in the infrared image, setting the pixel value of the pixel with the temperature value exceeding the temperature alarm threshold value to be 255 and setting the pixel value of the pixel with the temperature value smaller than the temperature alarm threshold value to be 0 to obtain a binary image of the infrared image, and then marking and communicating the binary image to obtain the coordinates (r1, c1) of a rectangular frame outside the high-temperature region (r2, c 2);
s4: converting the (R1, c1) (R2, c2) by using the translation vector T and the rotation matrix R to obtain the coordinates (y1, x1) (y2, x2) of the visible light image corresponding to the high-temperature region;
s5: expanding the rectangular frame determined by (y1, x1) (y2, x2) to obtain new target area coordinates (R1, C1) (R2, C2), wherein the length direction and the width direction of the rectangular frame are respectively expanded by 1/n of the rectangular frame, and preferably, n is 2-4;
wherein, R1 ═ y 1- (y2-y1)/(2n), R2 ═ y2+ (y2-y1)/(2 n);
C1=x1–(x2-x1)/(2n),C2=x2+(x2-x1)/(2n);
s6: selecting a rectangular frame determined by coordinates (R1, C1) (R2, C2) in the visible light image as a suspected target area, and then extracting a color feature vector, a contour feature vector and a texture feature vector of the suspected target area;
preferably, the color feature vector is obtained by performing LDA dimension reduction on the R component of the target area;
the method for obtaining the contour feature vector comprises the following steps: adopting a sobel operator as a template, carrying out operation with a target area to obtain a contour matrix, and then carrying out LDA dimension reduction on the contour matrix to obtain a contour characteristic vector;
the texture feature vector is a 5-dimensional feature vector consisting of energy, entropy, contrast, inverse variance and correlation calculated by the gray level co-occurrence matrix;
s7: and connecting the color feature vector, the contour feature vector and the texture feature vector end to form a one-dimensional column vector, and inputting the one-dimensional column vector into the SVM to obtain a judgment result.
According to the flame detection method based on binocular vision, firstly, the binocular vision is calibrated to obtain a translation vector T and a rotation matrix R from an infrared camera to a visible light camera, then, an infrared image is traversed and marked to obtain a suspected target area coordinate, the coordinate is converted and expanded through the translation vector T and the rotation matrix R, and finally the coordinate of the suspected target area of the visible light image is locked.
The embodiments of the present invention have been written in a progressive manner with emphasis placed on the differences between the various embodiments, and similar elements may be found in relation to each other.
While the embodiments of the present invention have been described in detail, the present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention within the knowledge of those skilled in the art.
Claims (8)
1. The fire detection method based on binocular vision is characterized by comprising the following steps:
s1: calibrating the infrared camera and the visible light camera in a binocular manner, and obtaining a translation vector T and a rotation matrix R from the infrared camera to the visible light camera;
s2: acquiring an infrared image in a field of view by using an infrared camera, and presetting a temperature alarm threshold value;
s3: traversing the temperature value corresponding to each pixel in the infrared image, setting the pixel value of the pixel with the temperature value exceeding the temperature alarm threshold value to be 255 and setting the pixel value of the pixel with the temperature value smaller than the temperature alarm threshold value to be 0 to obtain a binary image of the infrared image, and then marking and communicating the binary image to obtain the coordinates (r1, c1) of a rectangular frame outside the high-temperature region (r2, c 2);
s4: converting the (R1, c1) (R2, c2) by using the translation vector T and the rotation matrix R to obtain the coordinates (y1, x1) (y2, x2) of the visible light image corresponding to the high-temperature region;
s5: expanding the rectangular frame determined by (y1, x1) (y2, x2) to obtain new target area coordinates (R1, C1) (R2, C2);
s6: selecting a rectangular frame determined by coordinates (R1, C1) (R2, C2) in the visible light image as a suspected target area, and then extracting a color feature vector, a contour feature vector and a texture feature vector of the suspected target area;
s7: and connecting the color feature vector, the contour feature vector and the texture feature vector end to form a one-dimensional column vector, and inputting the one-dimensional column vector into the SVM to obtain a judgment result.
2. The binocular vision based fire detection method of claim 1, wherein: s1 includes the steps of:
s11: the calibration board is placed in a binocular vision field of the infrared camera and the visible light camera, so that the calibration board is ensured to be complete and clear in images collected by the two cameras, and the proportion of the calibration board in the infrared image is as large as possible;
s12: calibrating the two cameras respectively to obtain respective internal reference matrixes and distortion coefficient matrixes of the two cameras;
s13: and calculating a translation vector T and a rotation matrix R from the infrared camera to the visible light camera.
3. The binocular vision based fire detection method of claim 2, wherein: in S11, the proportion of the calibration plate in the infrared image is 60% or more.
4. The binocular vision based fire detection method of claim 1, wherein: in S2, the value range of the temperature alarm threshold is 65-200 ℃.
5. The binocular vision based fire detection method of claim 1, wherein: in S5, the length direction and the width direction of the rectangular frame are both expanded by 1/n of the rectangular frame, wherein n is 2-4.
6. The binocular vision based fire detection method of claim 1, wherein: in S6, the color feature vector is obtained by performing LDA dimension reduction on the R component of the target region.
7. The binocular vision based fire detection method of claim 1, wherein: in S6, the contour feature vector is obtained as follows: and (3) operating with the target area by adopting a sobel operator as a template to obtain a contour matrix, and then performing LDA dimension reduction on the contour matrix to obtain a contour characteristic vector.
8. The binocular vision based fire detection method of claim 1, wherein: in S6, the texture feature vector is a 5-dimensional feature vector composed of energy, entropy, contrast, inverse variance, and correlation calculated from the gray level co-occurrence matrix.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910840715.0A CN110619293A (en) | 2019-09-06 | 2019-09-06 | Flame detection method based on binocular vision |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910840715.0A CN110619293A (en) | 2019-09-06 | 2019-09-06 | Flame detection method based on binocular vision |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110619293A true CN110619293A (en) | 2019-12-27 |
Family
ID=68922362
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910840715.0A Pending CN110619293A (en) | 2019-09-06 | 2019-09-06 | Flame detection method based on binocular vision |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110619293A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112651276A (en) * | 2020-09-04 | 2021-04-13 | 江苏濠汉信息技术有限公司 | Power transmission channel early warning system based on double-light fusion and early warning method thereof |
CN113205562A (en) * | 2021-05-31 | 2021-08-03 | 中国矿业大学(北京) | Mine thermal power disaster identification and positioning method based on binocular vision |
CN115494193A (en) * | 2022-11-16 | 2022-12-20 | 常州市建筑科学研究院集团股份有限公司 | Machine vision-based flame transverse propagation detection method and system for single body combustion test |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102110342A (en) * | 2009-12-24 | 2011-06-29 | 中国航天科工集团第三研究院第八三五八研究所 | Fire detection method and device for realizing same |
CN102567989A (en) * | 2011-11-30 | 2012-07-11 | 重庆大学 | Space positioning method based on binocular stereo vision |
US20120314066A1 (en) * | 2011-06-10 | 2012-12-13 | Lee Yeu Yong | Fire monitoring system and method using composite camera |
CN104933723A (en) * | 2015-07-21 | 2015-09-23 | 闽江学院 | Tongue image segmentation method based on sparse representation |
CN105488941A (en) * | 2016-01-15 | 2016-04-13 | 中林信达(北京)科技信息有限责任公司 | Double-spectrum forest fire disaster monitoring method and double-spectrum forest fire disaster monitoring device based on infrared-visible light image |
CN107253485A (en) * | 2017-05-16 | 2017-10-17 | 北京交通大学 | Foreign matter invades detection method and foreign matter intrusion detection means |
CN110135266A (en) * | 2019-04-17 | 2019-08-16 | 浙江理工大学 | A kind of dual camera electrical fire preventing control method and system based on deep learning |
-
2019
- 2019-09-06 CN CN201910840715.0A patent/CN110619293A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102110342A (en) * | 2009-12-24 | 2011-06-29 | 中国航天科工集团第三研究院第八三五八研究所 | Fire detection method and device for realizing same |
US20120314066A1 (en) * | 2011-06-10 | 2012-12-13 | Lee Yeu Yong | Fire monitoring system and method using composite camera |
CN102567989A (en) * | 2011-11-30 | 2012-07-11 | 重庆大学 | Space positioning method based on binocular stereo vision |
CN104933723A (en) * | 2015-07-21 | 2015-09-23 | 闽江学院 | Tongue image segmentation method based on sparse representation |
CN105488941A (en) * | 2016-01-15 | 2016-04-13 | 中林信达(北京)科技信息有限责任公司 | Double-spectrum forest fire disaster monitoring method and double-spectrum forest fire disaster monitoring device based on infrared-visible light image |
CN107253485A (en) * | 2017-05-16 | 2017-10-17 | 北京交通大学 | Foreign matter invades detection method and foreign matter intrusion detection means |
CN110135266A (en) * | 2019-04-17 | 2019-08-16 | 浙江理工大学 | A kind of dual camera electrical fire preventing control method and system based on deep learning |
Non-Patent Citations (3)
Title |
---|
蒋先刚: "《基于稀疏表达的火焰与烟雾探测方法研究》", 31 August 2017 * |
谢威主编: "《智慧科技与情报服务》", 30 November 2018, 北京邮电大学出版社 * |
齐力编: "《公共安全大数据技术与应用》", 31 December 2017 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112651276A (en) * | 2020-09-04 | 2021-04-13 | 江苏濠汉信息技术有限公司 | Power transmission channel early warning system based on double-light fusion and early warning method thereof |
CN113205562A (en) * | 2021-05-31 | 2021-08-03 | 中国矿业大学(北京) | Mine thermal power disaster identification and positioning method based on binocular vision |
CN113205562B (en) * | 2021-05-31 | 2023-09-15 | 中国矿业大学(北京) | Mine thermodynamic disaster identification and positioning method based on binocular vision |
CN115494193A (en) * | 2022-11-16 | 2022-12-20 | 常州市建筑科学研究院集团股份有限公司 | Machine vision-based flame transverse propagation detection method and system for single body combustion test |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2018076732A1 (en) | Method and apparatus for merging infrared image and visible light image | |
CN110619293A (en) | Flame detection method based on binocular vision | |
CN103400150B (en) | A kind of method and device that road edge identification is carried out based on mobile platform | |
US8724885B2 (en) | Integrated image processor | |
CN109447011B (en) | Real-time monitoring method for infrared leakage of steam pipeline | |
CN102811863A (en) | Automated Inspection Of A Printed Image | |
CN110879080A (en) | High-precision intelligent measuring instrument and measuring method for high-temperature forge piece | |
CN103281513B (en) | Pedestrian recognition method in the supervisory control system of a kind of zero lap territory | |
CN110751635B (en) | Oral cavity detection method based on interframe difference and HSV color space | |
JP3486229B2 (en) | Image change detection device | |
CN114581760B (en) | Equipment fault detection method and system for machine room inspection | |
CN112053392A (en) | Rapid registration and fusion method for infrared and visible light images | |
CN116797977A (en) | Method and device for identifying dynamic target of inspection robot and measuring temperature and storage medium | |
CN104813341B (en) | Image processing system and image processing method | |
CN110595397A (en) | Grate cooler working condition monitoring method based on image recognition | |
CN112634179B (en) | Camera shake prevention power transformation equipment image change detection method and system | |
CN111145234A (en) | Fire smoke detection method based on binocular vision | |
Chowdhury et al. | Robust human detection and localization in security applications | |
CN110580684A (en) | image enhancement method based on black-white-color binocular camera | |
EP2176829B1 (en) | Arrangement and method for processing image data | |
CN116343100B (en) | Target identification method and system based on self-supervision learning | |
CN116563391B (en) | Automatic laser structure calibration method based on machine vision | |
CN108254380A (en) | PCB circuit board template matching method based on Digital Image Processing | |
JP4682782B2 (en) | Image processing device | |
US9875549B2 (en) | Change detection in video data |
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: 20191227 |
|
RJ01 | Rejection of invention patent application after publication |