A flame combustion position detection method, device and storage medium

Abstract

The present invention discloses a flame burning position detection method, which uses a camera to photograph the burning position of a grate and marks a rectangular area on the photographed image; measures the distance Z between the camera and the grate area. w , and the left and right deviation from the bottom center of the grate area X w and upper and lower deviations Y w At the same time, the inclination angle Q of the grate is obtained; by comparing the size of the projection rectangle and the actual selected rectangular area, the camera distance Z is optimized and adjusted w , attitude angle Q r , Q p , Q y , left and right deviation X w and upper and lower deviations Y w , so that the projected quadrilateral A c B c C c D c Move the whole image until it matches the actual selected rectangle; based on the optimized Q r , Q p , Q y , X w , Y w , Z w , update the external parameter matrix [R, T]; use the projection formula to convert the edge contour points of the flame and the grate contact obtained on the image into the actual position on the grate, and draw these points on the grate. The present invention can not only overcome the problem of biased burning in the actual combustion process, but also accurately reflect the grate combustion distribution.

Description

Technical Field

[0001] The present invention relates to the field of flame sampling and flame position detection, and in particular to a flame combustion position detection method, device and storage medium. Background Art

[0002] At present, the flame combustion situation of the incinerator is mainly perceived by installing a camera on the observation window, observing the flame through video, and sensing the flame combustion situation by observing the flame brightness and the distribution of the flame in the two-dimensional picture. However, this technical means often brings some undesirable effects, such as: because the installation position and installation angle of the camera are not clear, it only relies on the video information captured by the camera, and it is impossible to accurately judge the combustion position of the flame, resulting in the inability to accurately analyze and judge the two important parameters of the fire line and the biased burning. Summary of the invention

[0003] Purpose of the invention: The purpose of the present invention is to provide a flame combustion position detection method, equipment and storage medium that can not only overcome the problems of biased burning and slippage in the actual combustion process, but also accurately reflect the grate combustion distribution.

[0004] Technical solution: In order to achieve the above-mentioned invention object, a flame burning position detection method of the present invention comprises the following steps:

[0005] Use a camera to photograph the burning position of the grate, and mark a rectangular area on the photographed image as the grate area;

[0006] Measure the distance Z between the camera and the grate area w , and the left and right deviation from the bottom center of the grate area X w and upper and lower deviations Y w , at the same time, obtain the inclination angle Q of the grate;

[0007] Establish a new coordinate system with the center of the rectangular area as the origin;

[0008] Based on the above coordinate system and related measurements, the coordinates of the four vertices of the grate area are derived;

[0009] Using the photographic geometry projection formula, these vertex coordinates are projected into the camera image to obtain four projection points A c , B c , C c , D c ;

[0010] By comparing the size of the projected rectangle and the actual selected rectangular area, the camera distance Z is optimized and adjusted w ;

[0011] Optimize the camera's attitude angle Q r , Qp , Q y , adjust the left and right deviation X w and upper and lower deviations Y w , so that the projected quadrilateral A c B c C c D c Move the whole image until it matches the actual selected rectangle;

[0012] Based on the optimized Q r , Q p , Q y , X w , Y w , Z w , update the external parameter matrix [R,T];

[0013] The HSV color model is used to extract the position of the bottom edge of the flame;

[0014] Using the projection formula, the edge contour points where the flame and the grate are in contact obtained on the image are converted into actual positions on the grate, and these points are drawn on the grate.

[0015] Furthermore, the method further comprises the step of: calibrating the intrinsic parameters of the camera in an offline state to obtain the camera intrinsic parameter matrix A.

[0016] Furthermore, the intrinsic parameter matrix

[0017]

[0018] Among them, f x and f y is the focal length parameter, (c x , c y ) are the coordinates of the optical center.

[0019] Furthermore, the photographic geometric projection formula is [u,v,1] T =A*[R,T]*[X,Y,Z] T , where [u,v,1] T is a three-dimensional vector representing a point in the coordinate system. u and v are image coordinates, and 1 is used to convert the formula into homogeneous coordinate form. [R, T] is composed of the rotation matrix R and the translation vector T, which describes the transformation from the world coordinate system to the camera coordinate system. [X, Y, Z] T It is a three-dimensional vector representing a point in the world coordinate system, and X, Y, and Z are the coordinates of the point in the world coordinate system.

[0020] Furthermore, the optimization algorithm is a gradient descent method, a Levenberg-Marquardt algorithm, a Newton method or a quasi-Newton method.

[0021] Furthermore, the size of the projected rectangle and the actually selected rectangular area is compared by calculating the area difference, aspect ratio difference or position difference of edge pixels between the projected rectangle and the actually selected rectangular area.

[0022] Furthermore, using the HSV color model to extract the position of the lowermost edge of the flame also includes: preprocessing the image and performing edge detection.

[0023] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, a method for detecting a flame combustion position is implemented.

[0024] A non-transitory computer-readable storage medium stores a computer program, which implements a flame combustion position detection method when executed by a processor.

[0025] Beneficial effect: The present invention introduces the camera's installation position and installation angle information to switch the angle of viewing the fire video from the original two-dimensional plane to a 2.5D BEV perspective. After the camera is calibrated, the fire line of the flame in contact with the grate in the video is extracted in real time, and the extracted fire line information is projected onto the grate plane. The combustion distribution of the flame on the grate is calculated in real time, and the fire line and distribution information of the flame is parameterized and provided to the ACC (automatic combustion control system). This not only overcomes the problems of biased burning, material slippage and combustion control in actual combustion, but also can accurately and standardly reflect the fuel consumption and combustion distribution of the flame on the grate. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a schematic diagram of the combustion grate and the fire viewing camera structure of this embodiment;

[0027] Figure 2 is the burning position of the flame on the grate from the BEV perspective of this embodiment;

[0028] Figure 3 is the burning position of the flame on the grate in a two-dimensional perspective;

[0029] Figure 4 It is a schematic diagram of the physical structure of an electronic device provided in this embodiment. DETAILED DESCRIPTION

[0030] A flame burning position detection method comprises the following steps:

[0031] Calibrate the intrinsic parameters of the camera in offline state to obtain the camera intrinsic parameter matrix A and the intrinsic parameter matrix

[0032]

[0033] Among them, f x and f y is the focal length parameter, (c x , c y ) are the coordinates of the optical center. The focal length describes the strength of the camera perspective transformation and is generally on the main diagonal of the K matrix. The optical center is usually located at the center of the image plane, representing the center position of the camera imaging. In the K matrix, it is generally represented as (c x , c y ). Taking an industrial camera as an example, the intrinsic parameter matrix obtained after calibration is as follows:

[0034]

[0035] Use a camera to photograph the burning position of the grate, and mark a rectangular area on the photographed image, which is recorded as the grate area. The length of the rectangular area is recorded as W, and the width of the rectangular area is recorded as H.

[0036] like Figure 1 As shown, the distance Z between the camera and the grate area is measured w , and the left and right deviation from the bottom center of the grate area X w and upper and lower deviations Y w , at the same time, obtain the inclination angle Q of the grate;

[0037] The grate is usually installed at an angle of about 20 to 30°. The camera attitude angle Q is estimated relative to the standard horizontal plane. r , Q p , Q y , here Q r , Q p , Q y , Q, X w , Y w , Z w It can be measured when the equipment is installed, and does not require high measurement accuracy. If the corresponding data is not retained during installation or the measurement is inconvenient, a rough estimate can be made manually. r Indicates the roll angle, the rotation angle around the object's own X axis. p Indicates the pitch angle, the rotation angle around the object's own Y axis. y Represents the yaw angle, the rotation angle around the object's own Z axis.

[0038] According to the parameters of the implementation site, Q r , Q p , Q y , Q, X w , Y w , Z w It is generally related to the size of the grate furnace. This embodiment takes a certain grate furnace as an example. wAbout 0 meters, Y w About 2 meters, grate distance Z w About 15 meters, roll angle Q when installed horizontally r About 0°, pitch angle Q when installed horizontally p It is about 20-30°, and in this embodiment, it is 30°. When the camera is facing the grate, the yaw angle Q y About 0°.

[0039] A new coordinate system is established with the center of the rectangular area as the origin, in which the x-axis is parallel to the width direction of the grate area, and the z-axis is parallel to the length direction of the grate area. According to the right-hand rule, in the right-hand coordinate system, the y-axis is perpendicular to the xz plane.

[0040] Based on the above coordinate system and related measurement values, the coordinates of the four vertices of the grate area are derived. The coordinates of the four vertices are A W (x a ,y a 、z a ), B W (x b ,y b 、z b ), C W (x c ,y c 、z c )、D W (x d ,y d 、z d ).

[0041] This example takes a 5m wide grate as an example: the area needs to be as large as possible, so x is set here. a =x c =-2, x b =x d =2,y a =y b =3,y c =y d =-3, z a =z b =z c =z d =0.

[0042] Substitute the above parameters into the photographic geometry projection formula [u,v,1] T =A*[R,T]*[X,Y,Z] T , where [u,v,1] Tis a three-dimensional vector representing a point in the coordinate system. u and v are image coordinates, and 1 is used to convert the formula into homogeneous coordinate form. [R, T] is composed of the rotation matrix R and the translation vector T, which describes the transformation from the world coordinate system to the camera coordinate system. [X, Y, Z] T is a three-dimensional vector representing a point in the world coordinate system, and X, Y, and Z are the coordinates of the point in the world coordinate system. Projecting these vertex coordinates into the camera image, we get four projected points A c , B c , C c , D c The specific coordinate values ​​of R are based on the Euler angles of the right-handed coordinate system ZXY (0°, (90-Q)°, 0°) = (0°, 60°, 0°). The matrix R corresponding to a 0° rotation around the X axis in the right-handed coordinate system x ,

[0043]

[0044] Rotate 60° around the Y axis in the right-hand system, the corresponding matrix R y ,

[0045]

[0046] Rotate 0° around the Z axis in the right-hand system, the corresponding matrix R z ,

[0047]

[0048] R a =R x *R y *R z ;

[0049] T=[0,-sin(Q)*Z w ,0] T =[0,-7.5,0] T ;

[0050] [R,T]=R a [0:1,T];

[0051] Where R a [0:1] means taking R a The first two columns of and T are concatenated into a 3*3 matrix [R,T], and the formula [u,v,1] is substituted T =A*[R,T]*[X,Y,Z] T , u and v are calculated to be 4 rows and 1 column vectors, respectively A c , B c、 C c、 D cThe pixel coordinates in the image.

[0052] By comparing the size of the projected rectangle and the actual selected rectangular area, the camera distance Z is optimized and adjusted w , so that the two match;

[0053] Select an optimization algorithm, such as gradient descent, Levenberg-Marquardt algorithm, or other nonlinear optimization algorithm, to minimize the objective function and use the optimization algorithm to optimize the camera's attitude angle Q r , Q p , Q y , adjust the left and right deviation X w and upper and lower deviations Y w , so that the projected quadrilateral A c B c C c D c Move the whole image until it matches the actual selected rectangle; Qr represents the roll angle, which is adjusted to make the projected rectangle consistent with the actual space rectangle in the horizontal direction; Qp represents the pitch angle, which is adjusted to make the upper and lower sides of the projected quadrilateral match; Qy represents the yaw angle, which is adjusted to make the left and right sides of the projected quadrilateral match. Use the optimized camera attitude angle and deviation value to reproject and check the projected quadrilateral A c B c C c D c Check whether it matches the actual selected rectangle. If not, you may need to adjust the optimization algorithm parameters or reselect the optimization method and repeat the above steps.

[0054] Based on the optimized Q r , Q p , Q y , X w , Y w , Z w , update the external parameter matrix [R,T];

[0055] Using the HSV color model, by adjusting the brightness, chroma, and saturation thresholds, the position of the bottom edge of the flame, that is, the contact position of the flame and the grate, is extracted; the HSV representation simulates the way different colors are mixed together. H (hue) and S (saturation) are similar to the ratio of the current color purity to the maximum purity, ranging from [0,1], with 0 being gray, and V (value, brightness) representing a mixture of color with different amounts of black or white.

[0056] Use the projection formula [u,v,1] T =A*[R,T]*[X,Y,0] T, convert the edge contour points of the flame and grate contact obtained on the image into actual positions on the grate, and draw these points on the grate, such as Figure 2 As the video is updated in real time, these points will also be updated in real time, so as to obtain the fire video from the BEV perspective. This process combines photogrammetry, computer vision and image processing technology to achieve accurate shooting, measurement and analysis of the grate combustion position.

[0057] The present invention introduces the installation position and installation angle information of the camera, and switches the angle of viewing the fire video from the original two-dimensional plane to the 2.5D BEV perspective. After the camera calibration is completed, the fire line of the contact position between the flame and the grate in the video is extracted in real time, and the extracted fire line information is projected onto the grate plane. The combustion distribution of the flame on the grate is calculated in real time, and the fire line and distribution information of the flame are parameterized and provided to the ACC (combustion automatic control system). This not only overcomes the biased burning, material slippage and combustion control in actual combustion, but also can accurately and standardly reflect the fuel consumption and combustion distribution of the flame on the grate.

[0058] BEV: bird eye view, a perspective of looking down at ground objects or scenes from a high altitude, similar to the perspective of birds or drones observing from the air. 2.5D: adds the depth effect of the Z axis on the basis of 2D technology, making the picture more three-dimensional and realistic. The 2.5D BEV perspective is a perspective technology that integrates two-dimensional and three-dimensional information. It can retain sufficient scene information while simplifying the processing process, providing more accurate and comprehensive spatial perception capabilities for various applications.

[0059] Figure 4The physical structure diagram of an electronic device is illustrated, and the electronic device may include: a processor (processor) 701, a communication interface (Communications Interface) 702, a memory (memory) 703 and a communication bus 704, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus. The processor can call the logic instructions in the memory to execute the flame combustion position detection method. In addition, the logic instructions in the above-mentioned memory can be implemented in the form of a software function unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk.

[0060] On the other hand, an embodiment of the present invention further provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the flame combustion position detection method provided by the above-mentioned method embodiments.

[0061] On the other hand, an embodiment of the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to execute the flame combustion position detection method provided in the above embodiments.

[0062] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A flame burning position detection method, characterized in that: The following steps are involved: Use a camera to photograph the burning position of the grate, and mark a rectangular area on the photographed image as the grate area; Measure the distance Z between the camera and the grate area w , and the left and right deviation from the bottom center of the grate area X w and upper and lower deviations Y w , at the same time, obtain the inclination angle Q of the grate; Establish a new coordinate system with the center of the rectangular area as the origin; Based on the above coordinate system and related measurements, the coordinates of the four vertices of the grate area are derived; Using the photographic geometry projection formula, the four vertex coordinates are projected into the camera image to obtain four projection points A c , B c , C c , D c ; By comparing the size of the projected rectangle and the actual selected rectangular area, the camera distance Z is optimized and adjusted w ; Optimize the camera's attitude angle Q r , Q p , Q y , adjust the left and right deviation X w and upper and lower deviations Y w , so that the projected quadrilateral A c B c C c D c Move the whole image until it matches the actual selected rectangle; Qr represents the roll angle. Adjust Qr to make the projected rectangle consistent with the actual space rectangle in the horizontal direction. Qp represents the pitch angle. Adjust Qp to make the upper and lower sides of the projected quadrilateral match. Qy represents the yaw angle. Adjust Qy to make the left and right sides of the projected quadrilateral match. Use the optimized camera attitude angle and deviation value to re-project and check the projected quadrilateral A. c B c C c D c Whether it matches the actual selected rectangle; if not, adjust the parameters of the optimization algorithm or reselect the optimization method, and then repeat the above adjustment steps; Based on the optimized Q r , Q p , Q y , X w , Y w , Z w , update the external parameter matrix [R,T]; The HSV color model is used to extract the position of the bottom edge of the flame; Using the projection formula, the edge contour points where the flame and the grate are in contact obtained on the image are converted into actual positions on the grate, and these points are drawn on the grate.

2. The flame burning position detection method according to claim 1, characterized in that: It also includes calibrating the intrinsic parameters of the camera in an offline state to obtain the camera intrinsic parameter matrix A.

3. The flame burning position detection method according to claim 2, characterized in that: The intrinsic parameter matrix Among them, f x and f y is the focal length parameter, (c x , c y ) are the coordinates of the optical center.

4. The flame burning position detection method according to claim 1, characterized in that: The photographic geometry projection formula is [u,v,1] T =A*[R,T]*[X,Y,Z] T , where [u,v,1] T is a three-dimensional vector representing a point in the coordinate system. u and v are image coordinates, and 1 is used to convert the formula into homogeneous coordinate form. [R, T] is composed of the rotation matrix R and the translation vector T, which describes the transformation from the world coordinate system to the camera coordinate system. [X, Y, Z] T It is a three-dimensional vector representing a point in the world coordinate system. X, Y, and Z are the coordinates of the point in the world coordinate system.

5. The flame burning position detection method according to claim 1, characterized in that: The optimization algorithm used for optimization is gradient descent, Levenberg-Marquardt algorithm, Newton method or quasi-Newton method.

6. The flame burning position detection method according to claim 1, characterized in that: The size of the projected rectangle and the actually selected rectangular area is compared by calculating the area difference, aspect ratio difference or position difference of edge pixels between the projected rectangle and the actually selected rectangular area.

7. The flame burning position detection method according to claim 1, characterized in that: Extracting the position of the lower edge of the flame using the HSV color model also includes: preprocessing the image and edge detection.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the flame burning position detection method according to any one of claims 1 to 7 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the flame burning position detection method according to any one of claims 1 to 7 is implemented.

Images