Slag volume flow measuring device and measuring method

Abstract

This invention discloses a slag volumetric flow rate measuring device and method. The slag volumetric flow rate measuring device includes: a binocular camera, with one camera on each side of the center of the slag discharge direction, and the two cameras arranged front and back in the slag discharge direction; a green laser source for emitting green laser lines; a red laser source for emitting red laser lines; one of the red and green laser lines is parallel to the slag discharge direction, and the other is perpendicular to the slag discharge direction; each camera of the binocular camera acquires a slag discharge image simultaneously displaying both green and red laser lines. This invention's measuring device uses two lasers (red and green) and a binocular camera. The cross-sectional area of ​​the slag on the conveyor belt is calculated using the R-channel image, and the displacement is calculated by cross-correlation using the matching results of two consecutive frames of the G-channel image. The conveyor belt speed is calculated by combining the sampling time, and the cross-sectional area and speed are coupled to achieve online detection of slag volumetric flow rate.

Description

Technical Field

[0001] This invention relates to the field of measurement technology, specifically a method for measuring the flow rate of a dry slag discharge machine. Background Technology

[0002] Dry ash removal systems are ash removal systems that use air to cool high-temperature slag. They offer advantages such as energy saving, water saving, and high comprehensive utilization value of dry ash, and are currently widely used in major coal-fired power plants. In a dry ash removal system, cooling air enters the ash remover through air inlets on both sides and the head of the machine casing. This cools the high-temperature slag discharged from the boiler, raising its temperature before it enters the boiler furnace. Simultaneously, the slag, cooled to a specified temperature, is transported to the ash bin. The cooling air drawn into the boiler furnace by the dry ash removal system has two effects on boiler efficiency. On the one hand, it improves boiler efficiency by recovering heat from the slag; on the other hand, due to air leakage at the bottom of the furnace, under a fixed load, to ensure the air-coal ratio meets the set value, the air volume through the air preheater is reduced, leading to an increase in flue gas temperature and a decrease in boiler efficiency. Therefore, real-time detection of the ash volume at the bottom of the boiler, and subsequent effective adjustment of the cooling air volume of the dry ash remover based on the ash volume, is of great significance for improving the reliability and economy of the dry ash removal machine.

[0003] Currently, there are two main methods for measuring the slag discharge volume of dry slag machines: the slag truck weighing method and the multi-point level gauge method in the slag bin. The slag truck weighing method involves weighing the slag transported by the slag truck after it has accumulated in the slag bin for a certain period, thus obtaining slag volume data over that time period. This method provides relatively accurate slag volume data. However, in power plant operation, the slag bin typically needs to be emptied every 6 hours or more. The multi-point level gauge method accurately measures the slag level at different locations. Based on the regularity of slag accumulation, it calculates the volume change within the slag bin to ultimately calculate the boiler slag discharge volume. This method has advantages such as simple structure, convenient and quick installation and maintenance, and low cost. However, it also suffers from the problem of intermittent measurement and poor accuracy, failing to meet the requirements for real-time accurate measurement. Neither method can achieve real-time monitoring of the hot slag volume at the bottom of the boiler, thus preventing online adjustment of the cooling air volume. Slag characteristics can only be inferred from the online monitoring data of the slag discharge weighing system, resulting in significant lag.

[0004] Binocular stereo vision is an important component of machine vision. It utilizes binocular cameras to acquire two images of a measured object from different positions and calculates the positional deviation between corresponding points in the images based on the parallax principle to obtain the object's three-dimensional geometric information. Numerous scholars both domestically and internationally have researched this field, and the results have been applied to areas such as robot vision systems and industrial inspection. In recent years, some scholars have applied binocular vision to material conveying measurement, achieving dynamic volume measurement of bulk materials conveyed by belt conveyors. However, this method has limitations, including requirements on the size of the measured object and significant errors in measuring materials such as coal. Nevertheless, the binocular vision method provides an important approach for the online continuous detection of slag on steel belts in dry slag discharge systems.

[0005] The existing patent, CN 114111574 A, entitled "A Method for Three-Dimensional Measurement of High-Temperature Red-Hot Targets Using Binocular Laser Vision," discloses a method for three-dimensional measurement of high-temperature red-hot targets using binocular laser vision. The method includes the following steps: Step 1, acquiring images using a binocular camera and defining the acquired images as the left and right images respectively; Step 2, processing the images obtained in Step 1 to obtain the laser line coordinates; Step 3, calculating the centroids Dr and Dl of the laser lines in the new left and right images obtained in Step 2 respectively; Step 4, calculating the distance Di from the binocular camera to the target object based on the results obtained in Step 3. This invention enables real-time three-dimensional contour detection of forgings with high precision.

[0006] Existing technologies for measuring the volumetric flow rate of materials on conveyor belts have three main shortcomings: 1) They require significant modifications to the conveyor belt equipment itself, which is unsuitable for most industrial sites; 2) They have a narrow range of applications, with most technologies having high applicability requirements and not being widely applicable to various industrial sites; 3) The measuring equipment is relatively expensive. Taking the aforementioned patent, "Method and System for Measuring the Volumetric Flow Rate of Materials on Conveyor Belts," as an example, implementing this method requires both a two-dimensional laser rangefinder and a rotating reflector. In some harsh industrial environments, the placement of the rotating reflector is difficult. Furthermore, the patented method relies on significant deformation of the conveyor belt to obtain relevant data. This method is unsuitable for conveying low-density items such as tobacco, or in industrial settings where the conveyor belt is made of steel sheets and deformation is almost zero, such as in coal-fired power plant boiler slag conveyors. The equipment requires a two-dimensional laser rangefinder, which is generally expensive, especially for high-precision models.

[0007] Existing binocular vision technology suffers from low discriminative power when dealing with objects with weak or no texture, resulting in uniform pixel values, few feature points, and poor distinguishability. This can easily lead to mismatches and reduce the accuracy of disparity calculation. The aforementioned patent, "Binocular Linear Laser Vision Three-Dimensional Measurement Method for High-Temperature Red-Hot Targets," essentially calculates images captured by a single camera separately, which is not a conventional form of binocular vision and is unsuitable for the dynamic, texture-poor, feature-scarce, and porous characteristics of slag. Summary of the Invention

[0008] The technical problem to be solved by the present invention is to provide a low-cost and highly accurate slag volume flow rate measurement device and method based on vision.

[0009] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0010] A slag volume flow rate measuring device includes:

[0011] A binocular camera is used, with one camera on each side of the center of the slag discharge direction, and the two cameras are positioned one in front of the other in the slag discharge direction.

[0012] A green laser source, used to emit green laser beams;

[0013] A red laser source, used to emit red laser beams;

[0014] The red laser source emits a red laser line and the green laser source emits a green laser line, one of which is parallel to the slag discharge direction and the other is perpendicular to the slag discharge direction; each camera of the binocular camera acquires a slag discharge image that simultaneously contains both green and red laser lines.

[0015] The two cameras, positioned one in front of the other, are arranged at approximately a 45-degree angle to the conveyor belt.

[0016] Methods for measuring slag volumetric flow rate, including:

[0017] Two cameras capture images of the object with red and green laser lines, respectively.

[0018] The captured image is separated into RGB three channels to obtain the R channel image and the G channel image; the red laser line skeleton line is extracted from the obtained R channel image, and the green laser line skeleton line is extracted from the obtained G channel image.

[0019] Calculate the cross-sectional area of ​​the object perpendicular to the conveyor belt plane using the R-channel image of the extracted red laser line skeleton; calculate the velocity using the G-channel image of the extracted green laser line skeleton.

[0020] The volumetric flow rate is calculated based on the cross-sectional area of ​​the object calculated from the R-channel plot and the velocity calculated from the G-channel plot.

[0021] The steps for calculating the cross-sectional area of ​​the object perpendicular to the conveyor belt plane from the R-channel image of the extracted red laser skeleton lines include:

[0022] Parallax calculation is performed on the skeleton line of the red laser line extracted from the R-channel diagram:

[0023] d R =x RL -x RR

[0024] Where, d R For parallax; x RL Let x be the coordinates of the bright spot in the left image. RR The brightness coordinates are shown in the right figure;

[0025] Calculate the object depth z based on the calculated parallax. R :

[0026]

[0027] Where f is the camera focal length; B is the distance between the optical centers of the left and right cameras;

[0028] Based on the depth of the conveyor belt plane and the depth of the object, the thickness distribution r of the object along the red laser line is obtained:

[0029] r = z R0 -z Rx

[0030] Among them, z R0 For the reference planar conveyor belt depth, z Rx The depth after the object is placed;

[0031] Calculate the actual size of the object:

[0032]

[0033] Where h is the distance of the object projected onto the image width direction, which is obtained by summing the spatial distances Δx corresponding to all pixels of the object, and Δx is the depth z corresponding to each pixel. Rx The function of θ, where θ is the angle between the laser line and the horizontal direction;

[0034] Based on the calculated actual size and thickness distribution of the object, calculate the cross-sectional area S of the object perpendicular to the conveyor belt plane:

[0035]

[0036] Where r represents the thickness distribution, and r(x) represents the thickness at pixel x.

[0037] Velocity calculations were performed on the G-channel plot of the extracted green laser line skeleton, including:

[0038] Parallax calculation is performed on the green laser line skeleton line extracted from the machine G channel diagram:

[0039] d G =x GL -x GR

[0040] Where, d G For parallax; x GL Let x be the coordinates of the bright spot in the left image. GR The brightness coordinates are shown in the right figure;

[0041] Calculate the object depth z based on the calculated parallax. G :

[0042]

[0043] Where f is the camera focal length; B is the distance between the optical centers of the left and right cameras;

[0044] Based on the depth of the conveyor belt plane and the depth of the object, the thickness distribution g of the object along the green laser line is obtained:

[0045] g = z G0 -z Gx

[0046] Where z0 is the depth of the conveyor belt in the reference plane, z x The depth after the object is placed;

[0047] Based on the thickness distribution of the same object in two consecutive frames, the displacement of the object on the conveyor belt between the two frames is obtained.

[0048] The velocity of the object is determined based on the displacement of the object in the two frames before and after it is obtained.

[0049] The displacement of an object on a conveyor belt between two consecutive frames is obtained based on the object thickness distribution in the same image. This includes:

[0050] Perform cross-correlation calculation on the object thickness distribution of two consecutive frames to obtain the number of displacement pixels M corresponding to the maximum value of the cross-correlation function;

[0051] Based on the number of displacement pixels M corresponding to the maximum value of the cross-correlation function, the displacement L of the object on the conveyor belt between the previous and next frames is obtained:

[0052] L=MΔx(z G0 )

[0053] In the step of obtaining the displacement of the object in two consecutive frames, the cross-correlation function used for cross-correlation calculation is:

[0054]

[0055] Where n is the number of displacement pixels; N is the length of the thickness distribution sequence; g1 i Let g1 be the object thickness at the i-th displacement pixel in the object thickness distribution of the previous frame image, and g2 be the object thickness at the i-th displacement pixel. i Let g2 be the object thickness at the i displacement pixels in the object thickness distribution of the next frame image, where i = 1, 2, 3, ..., N.

[0056] The measuring device of this invention employs two lasers (red and green) and a binocular camera. The red and green lasers project linear laser beams onto the slag surface of the conveyor belt. Brightness and texture information are actively added manually, and the R and G channels are separated to achieve accurate binocular visual matching. Based on this accurate matching, the cross-sectional area of ​​the slag on the conveyor belt is calculated using the R channel image, and the displacement is calculated by cross-correlation of the matching results of two consecutive G channel images. The conveyor belt speed is calculated by combining the sampling time, and the cross-sectional area and speed are coupled to achieve online detection of slag volumetric flow rate.

[0057] Compared with the prior art, the present invention has the following advantages:

[0058] (1) It can realize the volumetric flow rate measurement of slag without texture. By actively adding laser lines, strong features are added to the slag without texture, solving the problem of difficult matching of stereo matching of binocular vision for slag without texture, and realizing the measurement of slag volumetric flow rate.

[0059] (2) The system has high measurement accuracy. It falls within the scope of optical measurement, and the overall system accuracy is high.

[0060] (3) Real-time measurement is possible. This invention can quickly process binocular vision images, processing 2-3 sets of images per second, meeting the real-time requirements of general industrial sites.

[0061] (4) Simple structure. The invention consists of two cameras and lasers of different colors. The lasers only need to emit linear lasers, without the need for complex other structures such as galvanometers, structured light, etc.

[0062] (5) Minimal modification required to existing equipment. Minimal modification required to the dry slag discharge machine itself, no impact on the conveyor belt, no interference with its operation, and no issues affecting the safety or efficiency of dry slag discharge.

[0063] (6) High cost performance. The device of this invention can use similar general equipment on the market, with no special requirements, low cost, and accurate and fast measurement results, thus offering high cost performance.

[0064] (7) Wide range of applications. Due to the addition of a line laser, this invention has strong environmental adaptability. In addition to dry slag discharge systems, it is also applicable to various industries that require materials to be transported by conveyor belts, such as coal, mining, steel, construction, grain, and tobacco. Attached Figure Description

[0065] Figure 1 Schematic diagram of dry slag remover and arrangement of binocular cameras;

[0066] Figure 2 Schematic diagram of a binocular camera system;

[0067] Figure 3 Dual-laser-line binocular vision system for measuring the volumetric flow rate of textureless slag;

[0068] Figure 4 RGB channel separated image and preprocessing; (a) real object photo, (b) R channel image, (c) G channel image, (d) binarized R channel image;

[0069] Figure 5 Skeleton extraction image;

[0070] Figure 6 Search for feature highlights in the left and right images of the R channel. (a) Highlight coordinates in the left image, (b) Highlight coordinates in the right image;

[0071] Figure 7 Schematic diagram of binocular vision principle;

[0072] Figure 8 Actual laser line.

[0073] Figure descriptions: 1: Extrusion shut-off gate; 2: Cooling vent; 3: Binocular camera; 4: Red laser; 5: Green laser; 6: Steel belt conveyor; 7: Slag crusher; 8: Slag bin; 9: Conveyor belt; 10: Binocular camera; 11: Red laser; 12: Green laser. Detailed Implementation

[0074] The solution for achieving slag volumetric flow rate consists of two parts: a measuring device and a measuring method.

[0075] Description of the measuring device:

[0076] This invention applies to dry ash removal machines for power plant boilers. A schematic diagram of the dry ash removal machine structure after equipping it with an active binocular stereo vision device is shown below. Figure 1 As shown: A dry ash removal machine transports hot ash falling from the boiler tail end via a steel belt conveyor. The steel belt is in a sealed state, and a binocular camera is positioned directly above the steel belt to capture images. Dual lasers are used. Figure 1 The principle of using a black and white cylindrical active light source is illustrated in the diagram below. Figure 2 As shown, a laser device is added between conventional binocular camera devices to generate two distinct colored linear laser marks with strong features on the surface of a textureless object, thereby improving the matching accuracy of the binocular camera for textureless objects.

[0077] Figure 2In this system, a red laser line is perpendicular to the conveyor belt and is used to calculate the cross-sectional area of ​​the object; a green laser line is parallel to the conveyor belt and is used to calculate the conveyor belt speed. A binocular camera is arranged in parallel, with its overall orientation at approximately a 45-degree angle to the conveyor belt. Images captured by the binocular cameras and incorporating the laser lines are transmitted to a host computer via network cable. The host computer then processes these images using subsequent algorithms to obtain the required volumetric flow rate. This measuring device is versatile; the cameras used have no special requirements, and the specific camera and lens selection is determined based on the range of objects being measured. The two laser light sources only need to be of different colors, and the laser shapes do not need to be complexly encoded into structured light. The power of the light source is determined based on the absorptivity of the object being measured.

[0078] Method implementation details:

[0079] Method implementation as follows Figure 3 The flowchart is shown below:

[0080] The first step is to complete the equipment installation, ensuring the laser line is installed as specified, and to complete the binocular camera calibration. The essence of binocular machine vision is to use the two-dimensional planar position information captured by the camera to analyze the position of the measured object in three-dimensional space. This analysis is achieved through coordinate system transformation, which is described by the camera's intrinsic and extrinsic parameters. These parameters need to be determined through camera calibration. Furthermore, because the camera uses lenses to form images, the manufacturing process of lenses and cameras cannot be perfectly ideal, resulting in significant distortion at image edges. This causes a large deviation in the obtained three-dimensional position, so calibration is also necessary to obtain distortion correction parameters and correct the image. The camera calibration is performed using the Zhang Zhengyou calibration method in MATLAB. This method only requires the camera to take multiple images of the black and white checkerboard calibration board at different angles and distances. Calibration is completed by understanding the correspondence between each feature point on the calibration board and its image point on the image plane.

[0081] The Zhang Zhengyou calibration method for single-camera systems is already quite complete and widely used. Based on obtaining the internal and external parameters of the single camera, such as the focal length f, it can be used to calibrate stereo cameras. Specifically, the external parameters of the stereo system, the rotation matrix R and the translation matrix T, can be obtained from the rotation matrices R of the left and right cameras after single-camera calibration. L and R R and translation vector T L and T R Let P be the coordinates of a point P in space within the left and right camera coordinate systems. L and P R Based on single-camera calibration, it can be expressed as follows:

[0082]

[0083] Among them, P W It is the coordinate of point P in the world coordinate system. Equation (2) eliminates P WAfterwards, one can obtain,

[0084] P R =R R R L T [P L -R L (T R -T L (3)

[0085] Based on the relationship between the external parameter rotation matrix, translation matrix and spatial points, the required solution matrix for bi-target calibration can be obtained from equation (3).

[0086]

[0087] Where R is the rotation matrix of the stereo system to be calibrated, and T is the translation vector of the stereo system to be calibrated. With the cameras arranged nearly parallel, R is approximately the identity matrix, and the vector T can be expressed as in equation (5).

[0088] T = [B00] (5)

[0089] Where B is the baseline length.

[0090] The second step is to calibrate the depth of the conveyor belt reference plane. The calibration process is shown in steps three through seven below, thereby obtaining the depth of the conveyor belt at the laser line.

[0091] The third step involves using both cameras to capture images of the object with red and green laser lines. For textureless objects, the actual situation is as follows: Figure 4 (a).

[0092] The fourth step involves calibrating the captured image using the specified parameters, then separating the RGB channels to obtain the R channel. Figure 4 (b) and G channel Figure 4 As shown in (c).

[0093] The fifth step involves converting the R and G channel images to grayscale separately. Because there is a significant difference in brightness between the red and green laser lines in the R and G channels, the images can be binarized using a threshold setting. The binarized R channel image is shown below. Figure 4 As shown in (d):

[0094] Since the laser line occupies multiple pixels in the image, to further eliminate matching ambiguity, it is necessary to extract the laser line skeleton. This ensures that each row of pixels uses only a pixel-wide line as input, achieving a unique match. The skeleton is widely used to describe the refined form of a binary image while also representing the overall shape of the binary region and information such as size, orientation, and connectivity. This paper uses median transformation for skeleton extraction.

[0095] Let A be a planar region, and let MA be the central axis of A, defined as follows:

[0096]

[0097] Where p, q1, and q2 are points in region A, and d(p, q2) and d(p, q2) are Euclidean distances.

[0098] The median transformation of A refers to the ordered pairs consisting of a point in MA(A) and its distance to the boundary of A, which can be expressed by equation (7):

[0099]

[0100] MAT(A) is the skeleton of the central axis MA(A) of plane A, p is a point on the skeleton of plane A, q is an arbitrary point on the boundary of plane A, and r is the infimum of the Euclidean distance d(p,q) between p and q.

[0101] After binarizing the R channel Figure 4 (d) Perform a mid-axis transformation to extract the skeleton, as shown in the example. Figure 5 As shown:

[0102] Step 6: Based on the calibrated and corrected skeleton image obtained from preprocessing, bright pixel searches can be performed on the left and right images for matching. Because the left and right images are row-aligned after calibration and correction, and the laser line after binarized skeleton extraction is only one pixel wide, accurate matching can be achieved by searching only a single pixel in one row when searching the left and right images. This solves the problem of mismatching or no matching for textureless objects in traditional binocular vision. After searching for pixels in the left and right images, the coordinates of the bright spots in each row are recorded, and then disparity calculation is performed. Taking the R channel as an example... Figure 6 As shown:

[0103] The coordinates x of the bright spot in the left image are obtained by searching for the maximum value. RL And the brightness coordinate x in the right figure RR The parallax d can be obtained. R As shown in equation (8):

[0104] d R =x RL -x RR (8)

[0105] Step 7, based on the binocular vision principle model diagram as shown below. Figure 7 As shown in the figure, point p L and p R Let x be the image point of a point P in space on the imaging planes of the left and right cameras, respectively, and let x be the line segment x. L and x R denoted as , where are the distances from the left and right image points to the edge of the camera's imaging plane, f is the camera's focal length, and O is the distance from the left and right image points to the edge of the camera's imaging plane. L and O RLet B be the optical centers of the left and right cameras, respectively, and let B be the distance between the optical centers of the left and right cameras, i.e., the system baseline length. Based on the triangle relationship, the distance z from a point P in space to the camera is calculated through the R channel. R , can be represented as,

[0106]

[0107] The focal length f and baseline B are obtained from the calibration results, and the parallax d is obtained from equation (8). R This gives the depth z of the object at the spatial point. R The thickness r distribution of the conveyor belt object can then be obtained, as shown in equation (10).

[0108] r = z R0 -z Rx (10)

[0109] Among them, z R0 For the reference planar conveyor belt depth, z Rx The depth after placing the object.

[0110] Step 8: Based on the pinhole imaging principle of a camera, the actual spatial size corresponding to each pixel can be calculated by determining the spatial depth of the image junction point after calibration and correction. Since the left-right matching process only searches and pairs rows, it's equivalent to projecting a tilted laser line onto the image width direction. Therefore, the tilt angle needs to be considered when calculating the actual size of the object. Figure 8 As shown:

[0111] The actual size of the object can then be calculated using equation (11):

[0112]

[0113] Where h is the distance from the object's projection onto the image width direction, and is the spatial distance Δx(z) between all pixels of the object. Rx The values ​​are obtained by summing the values ​​of Δx and θ, where Δx is a function of the depth of the pixel and θ is the angle between the laser line and the horizontal direction. Based on the actual size l and thickness of the object, the cross-sectional area S of the object perpendicular to the conveyor belt plane can be calculated, as shown in equation (11).

[0114]

[0115] Where r(x) is the thickness at pixel x, Δx(z) Rx ) represents a depth of z Rx One pixel corresponds to the actual spatial distance.

[0116] Step 9: The speed calculation principle for the G channel image is the same as that for the R channel, yielding the object thickness distribution g. The object's thickness distribution differs between consecutive frames as it moves along the conveyor belt. Based on the object thickness distribution in the two consecutive frames, a cross-correlation operation is performed to obtain the displacement of the object between the two frames, thus determining the object's speed. The cross-correlation principle is as follows: For two consecutive frames with a sampling time interval of T, after thickness calculation, the thickness distribution g1 of the same object in the two frames is obtained. i With g2 i For i = 1, 2, 3, ..., N, their cross-correlation function is:

[0117]

[0118] Where n is the number of displacement pixels, N is the length of the thickness distribution sequence, and the number of displacement pixels corresponding to the maximum value of the cross-correlation function is denoted as M;

[0119] In the tenth step, the speed can be calculated as follows:

[0120]

[0121] Finally, the volumetric flow rate V is obtained. flow :

[0122] V flow =vS (14)

Claims

1. A method for measuring slag volumetric flow rate based on a slag volumetric flow rate measuring device, the slag volumetric flow rate measuring device comprising: A binocular camera system is used, with one camera on each side of the center of the slag discharge direction, and the two cameras are positioned one in front of the other in the slag discharge direction. A green laser source, used to emit green laser beams; A red laser source, used to emit red laser beams; The red laser source emits a red laser line and the green laser source emits a green laser line, one of which is parallel to the slag discharge direction and the other is perpendicular to the slag discharge direction; each camera of the binocular camera acquires a slag discharge image that simultaneously displays both the green and red laser lines. The method for measuring slag volumetric flow rate is characterized by: Two cameras capture images of the object with red and green laser lines, respectively. The captured image is separated into RGB three channels to obtain the R channel image and the G channel image; the red laser line skeleton line is extracted from the obtained R channel image, and the green laser line skeleton line is extracted from the obtained G channel image. Calculate the cross-sectional area of ​​the object perpendicular to the conveyor belt plane using the R-channel image of the extracted red laser line skeleton; calculate the velocity using the G-channel image of the extracted green laser line skeleton. The volumetric flow rate is calculated based on the cross-sectional area of ​​the object calculated from the R-channel plot and the velocity calculated from the G-channel plot. The steps for calculating the cross-sectional area of ​​the object perpendicular to the conveyor belt plane from the R-channel image of the extracted red laser line skeleton include: Parallax calculation is performed on the skeleton line of the red laser line extracted from the R-channel diagram: wherein, d R is the parallax; x RL is the left image bright point coordinate, x RR is the right image bright point coordinate; Calculate the object depth based on the calculated parallax. z R ： in, f The focal length of the camera; B The distance between the optical centers of the left and right cameras; Based on the depth of the conveyor belt plane and the depth of the object ， Obtain the thickness distribution of the conveyor belt object along the red laser line. r : in, z R0 For reference planar conveyor belt depth, z Rx The depth after the object is placed; Based on the calculated thickness distribution, calculate the cross-sectional area of ​​the object perpendicular to the conveyor belt plane. S : in, r For thickness distribution, h This represents the distance the object projects onto the image along its width. θ The angle between the laser line and the horizontal direction; Velocity calculations were performed on the G-channel plot of the extracted green laser line skeleton, including: Parallax calculation is performed on the green laser line skeleton line extracted from the machine G channel diagram: Where, d G For parallax; x GL The coordinates of the bright spot in the left image are... x GR The coordinates of the bright spot in the right image; Calculate the object depth based on the calculated parallax. z G ： in, f The focal length of the camera; B The distance between the optical centers of the left and right cameras; Based on the depth of the conveyor belt plane and the depth of the object ， Obtain the thickness distribution of the conveyor belt object along the green laser line. g : in, z 0 For reference planar conveyor belt depth, z x The depth after the object is placed; Based on the thickness distribution of the same object in two consecutive frames, the displacement of the object on the conveyor belt between the two frames is obtained. The velocity of the object is determined by the displacement of the object in the two frames before and after the object is obtained. The displacement of an object on the conveyor belt between two consecutive frames is obtained based on the object thickness distribution in two consecutive images of the same object, including: Perform cross-correlation calculation on the object thickness distribution of two consecutive frames to obtain the number of displacement pixels M corresponding to the maximum value of the cross-correlation function; Based on the number of displacement pixels M corresponding to the maximum value of the cross-correlation function, the displacement L of the object on the conveyor belt between the previous and next frames is obtained: in, Δx(z G0 ) For depth z G0 One pixel corresponds to the actual spatial distance. In the step of obtaining the displacement of the object in two consecutive frames, the cross-correlation function used for cross-correlation calculation is: Where n is the number of displacement pixels; N is the length of the thickness distribution sequence; g1 i The thickness distribution of objects in the previous frame image g 1 i The object thickness per displacement pixel g 2 i For the thickness distribution of objects in the later frame image g 2 i The object thickness per displacement pixel i = 1, 2, 3, ..., N.

2. The method for measuring slag volume flow rate according to claim 1, characterized in that, The velocity of the object is to be determined as follows: in, v M represents the velocity of the moving object; M represents the number of displacement pixels corresponding to the maximum value of the cross-correlation function used for cross-correlation calculation; T represents the sampling time interval between two consecutive frames. z G0 The depth of the conveyor belt plane; Δx(z G0 ) For depth z G0 One pixel corresponds to the actual spatial distance.

3. The method for measuring slag volume flow rate according to claim 2, characterized in that, The volumetric flow rate calculated based on the cross-sectional area of ​​the object calculated from the R-channel plot and the velocity calculated from the G-channel plot is: Finally, the volumetric flow rate is obtained. V flow ： in, V flow This refers to volumetric flow rate.

4. The method for measuring slag volume flow rate according to claim 1, characterized in that, The two cameras, positioned one in front of the other, are arranged at a 45-degree angle to the conveyor belt.

Images