Slag block contour recognition method and device, storage medium and electronic equipment

Abstract

The invention discloses a slag block contour recognition method and device, a storage medium and electronic equipment. An initial image of a slag block on a grid plate at the bottom of a slag well is obtained through at least two cameras, brightness gradient analysis is conducted through the brightness difference between the slag block and the low-temperature environment, and the contour is preliminarily recognized. The camera can cover the whole area of the slag block, is not limited by a fixed point, can capture irregular edge details, and avoids the loss of contour information. Contours recognized by multiple cameras are fused to obtain a three-dimensional contour, a three-dimensional form is constructed, the space size is completely reflected, the limitation of a single view angle is broken through, and the overall form is comprehensively captured. And continuously acquiring image capture brightness dynamic change, correcting the contour based on the relevance of the image capture brightness dynamic change and the contour, tracking state change in real time, and ensuring that the contour is consistent with an actual form. The corrected contour is used for the hydraulic shutoff door to adjust the extrusion depth, accurate slag extrusion is achieved, the problem of inaccurate operation caused by incomplete contour recognition and large size estimation deviation is solved, and a reliable size basis is provided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image recognition, and particularly relates to a slag block contour identification method and device, a storage medium and an electronic device. BACKGROUND

[0002] In the prior art, a plurality of retractable mechanical probes are arranged below the grid plate at the bottom of the slag well, the probe ends are connected to pressure sensors, when the slag block falls on the grid plate, part of the probes will be pressed by the slag block to produce displacement, the pressure sensor converts the pressure signal into an electrical signal, according to the number, position and pressure value of the triggered probes, the coverage range and approximate size of the slag block are judged. Specifically, when the pressure value exceeds a preset threshold, it is determined that the position of the probe is covered by the slag block, the distribution density of the triggered probes is counted to estimate the length and width of the slag block, and the duration of the pressure signal is combined to judge whether the slag block is stably present.

[0003] However, due to the limited and fixed distribution density of the probes, for irregularly shaped slag blocks, the edge part of the slag block may not contact any probe, which leads to the inability to completely capture the actual contour of the slag block, and further makes it difficult to accurately judge the boundary range of the slag block, so that the estimated size of the slag block has a large deviation from the actual size. SUMMARY

[0004] In view of the above problems, the present application provides a slag block contour identification method and device, a storage medium and an electronic device.

[0005] To solve the above technical problems, the present application provides the following solutions:

[0006] In a first aspect, the present application provides a slag block contour identification method, which is realized based on at least two cameras arranged outside the slag well, and the method comprises: acquiring an initial image of a slag block on a grid plate at the bottom of the slag well through the at least two cameras, performing brightness gradient analysis based on the brightness difference between the slag block and the low-temperature environment in the slag well in the initial image, and preliminarily identifying the contour of the slag block; fusing the contours identified based on the at least two cameras to obtain a three-dimensional contour of the slag block; continuously acquiring images of the slag block through the at least two cameras to capture the dynamic change in brightness of the slag block; and correcting the contour of the slag block based on the correlation between the dynamic change in brightness and the contour, so as to adjust the extrusion depth according to the contour of the slag block for a hydraulic shut-off door to realize precise slag extrusion.

[0007] In a second aspect, the present application provides a slag block contour identification device, which comprises:

[0008] The identification module is configured to acquire an initial image of a slag block on a grid plate at the bottom of the slag well through at least two cameras arranged outside the slag well, perform brightness gradient analysis based on the brightness difference between the slag block and the low-temperature environment in the slag well in the initial image, and preliminarily identify the contour of the slag block.

[0009] a fusion module configured to fuse the contours identified by the at least two cameras to obtain a three-dimensional contour of the slag block;

[0010] a capture module configured to continuously capture images of the slag block by the at least two cameras to capture dynamic changes in brightness of the slag block;

[0011] a correction module configured to correct the contour of the slag block based on the correlation between the dynamic changes in brightness and the contour, so that the hydraulic shut-off gate adjusts the extrusion depth according to the contour of the slag block to achieve accurate slag extrusion.

[0012] To achieve the above-mentioned purposes, according to the third aspect of the present application, a storage medium is provided, the storage medium comprising a stored program, wherein the program controls the device where the storage medium is located to execute the slag block contour identification method of the first aspect when the program is running.

[0013] To achieve the above-mentioned purposes, according to the fourth aspect of the present application, an electronic device is provided, the device comprising at least one processor, and at least one memory connected with the processor, a bus; wherein the processor, the memory complete mutual communication through the bus; the processor is used to call the program instruction in the memory, to execute the slag block contour identification method of the first aspect.

[0014] By the above technical solution, the technical solution provided by the present application has at least the following advantages:

[0015] The present application obtains the initial image of the slag block on the grid plate at the bottom of the slag well by at least two cameras, and performs brightness gradient analysis by using the brightness difference between the slag block and the low-temperature environment in the slag well to preliminarily identify the contour of the slag block. Since the camera can cover the whole area of the slag block through image acquisition, it is not limited by the arrangement of fixed points, and can capture the edge details of the slag block, including the irregular edge parts, avoiding the missing of contour information due to the limited monitoring range. Then, the contours identified by the at least two cameras are fused to obtain a three-dimensional contour of the slag block. Through the fusion of multiple view images, the three-dimensional shape of the slag block can be constructed, and the spatial size of the slag block can be fully reflected, breaking through the limitations of single-view planar images in contour presentation and ensuring the comprehensive capture of the overall shape of the slag block. Subsequently, the images of the slag block are continuously captured by the at least two cameras to capture the dynamic changes in brightness of the slag block, and the contour of the slag block is corrected based on the correlation between the dynamic changes in brightness and the contour. This process can track the state changes of the slag block in real time, and the contour is corrected dynamically to ensure that the corrected contour is always consistent with the actual shape of the slag block. Finally, the corrected contour can be used to adjust the extrusion depth of the hydraulic shut-off gate to achieve accurate slag extrusion, thereby solving the problem of inaccurate operation caused by incomplete contour identification and large size estimation deviation. Through complete and dynamic contour identification, a reliable size basis is provided for subsequent operation.

[0016] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0017] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0018] Figure 1 A flowchart illustrating a slag block contour recognition method provided in an embodiment of this application is shown.

[0019] Figure 2 This illustration shows a schematic diagram of the camera's shooting range and the boundary of a slag block according to an embodiment of this application;

[0020] Figure 3 This illustration shows a schematic diagram of the relative position of a camera and a slag well according to an embodiment of this application;

[0021] Figure 4 This illustration shows a schematic diagram of the initial outline and boundary point coordinates of a slag block according to an embodiment of this application;

[0022] Figure 5 This illustration shows a schematic diagram of the contours of a slag block before and after deformation, and the monitoring range of a shut-off door and a camera, provided in an embodiment of this application.

[0023] Figure 6 This illustration shows a structural schematic diagram of a slag block contour recognition device provided in an embodiment of this application;

[0024] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown. Detailed Implementation

[0025] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application will be thorough and complete, and will fully convey the scope of the present application to those skilled in the art.

[0026] In the embodiments of this application, the terms "first," "second," etc., do not have a logical or temporal dependency, nor do they limit the quantity or execution order. It should also be understood that although the following description uses the terms "first," "second," etc., to describe various elements, these elements should not be limited by the terms. These terms are merely used to distinguish one element from another.

[0027] In this application, the term "at least one" means one or more, and the term "multiple" means two or more.

[0028] It should also be understood that the term “if” can be interpreted as “when” or “upon”, or “in response to determination” or “in response to detection”. Similarly, depending on the context, the phrase “if determination…” or “if detection [the stated condition or event]” can be interpreted as “when determination…” or “in response to determination…” or “when detection [the stated condition or event]” or “in response to detection [the stated condition or event]”.

[0029] In existing technologies, slag block contour recognition is achieved by deploying multiple retractable mechanical probes below the grid plate at the bottom of the slag well, with pressure sensors connected to the tips of each probe. When a slag block falls onto the grid plate, some probes are displaced by compression. The pressure sensors convert the pressure signal into an electrical signal, and the coverage area and approximate size of the slag block are determined based on the number, location, and pressure value of the triggered probes. Specifically, when the pressure value exceeds a preset threshold, it is determined that the location of that probe is covered by slag. The length and width of the slag block are estimated by statistically analyzing the distribution density of the triggered probes, and the stability of the slag block is determined by combining the duration of the pressure signal. However, this technology has significant limitations: due to the limited and fixed density of the probes, the edges of irregularly shaped slag blocks may not contact any probes, making it impossible to completely capture the actual contour of the slag block. Consequently, it is difficult to accurately define the boundary range of the slag block, resulting in a large deviation between the estimated slag block size and the actual size.

[0030] Based on this, this application provides a method for identifying the contour of slag blocks. The method will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a slag block contour recognition method provided in this application. Specifically, it includes the following steps:

[0031] Step 110: Obtain initial images of slag blocks on the grid plate at the bottom of the slag well using at least two cameras. Perform brightness gradient analysis based on the brightness difference between the slag blocks and the low-temperature environment inside the slag well in the initial images to preliminarily identify the outline of the slag blocks.

[0032] When the slag discharged from the boiler falls onto the grid plate at the bottom of the slag well due to its own weight (the initial temperature of the slag is usually 800-1200℃, higher than the low temperature environment of 50-80℃ inside the slag well), two cameras pre-positioned outside the slag well simultaneously start image acquisition. These two cameras are respectively installed at the midpoint of adjacent side walls of the slag well (for example, for a square slag well with a side length of 2m, camera 1 is fixed at the midpoint of the north side wall of the slag well, 1.5m above the bottom of the slag well; camera 2 is fixed at the midpoint of the east side wall of the slag well, at the same height), and both use industrial cameras with photosensitive capabilities (such as a CMOS camera with a resolution of 1920×1080, equipped with an 8mm fixed-focus lens, and a photosensitive range covering 400-1100nm visible light and near-infrared bands). The center line of their lenses is pointed towards the center area of ​​the slag well, ensuring complete imaging of the slag on the grid plate. Figure 2 and Figure 3 As shown, the camera's shooting range covers the area where slag blocks may exist in the slag well in the form of rays. The overall range of the camera corresponds to the image area covered by the camera's field of view, the slag block boundary is the outline of the slag block in the image, and the rays correspond to the center line and edge light of the lens, intuitively showing the camera's shooting coverage of the slag blocks.

[0033] After the camera acquires the initial image (image format: RGB color image, frame rate: 25fps), the initial image is first processed to grayscale: the weighted average method is used to convert the RGB three-channel data into a single-channel grayscale image. The conversion formula is Gray = 0.299 × R + 0.587 × G + 0.114 × B (where R, G, and B are the pixel values ​​of the red, green, and blue channels, respectively, with a value range of 0-255). This removes color information interference and retains the grayscale value information that reflects the brightness characteristics. At this time, the grayscale value of each pixel in the grayscale image represents the brightness value at that location, forming the brightness channel data.

[0034] When delineating the low-temperature environment region within the slag well based on brightness channel data, after eliminating grid structure interference by using a preset grid coordinate range (e.g., a rectangular area corresponding to pixel coordinates (x1, y1) to (x2, y2) in the image), the remaining region is clustered using the K-means clustering algorithm. Pixels are divided into two clusters based on grayscale values ​​(low-brightness cluster and potential slag block cluster). The low-brightness cluster, with pixel grayscale values ​​concentrated in the range of 30-70, is identified as the low-temperature environment region within the slag well. The average grayscale value of all pixels in this region is calculated to obtain the average environmental brightness value Avg. L (For example, Avg is calculated) L =50).

[0035] The first threshold Th1 is set based on the average ambient brightness value, specifically Th1 = Avg. L+1.5×σ (σ is the standard deviation of the brightness value in the low-temperature environment area, here σ=10, so Th1=50+15=65), and the pixels with gray values ​​greater than Th1 in the brightness channel data are marked as suspected slag block areas. Suspected slag block areas appear as continuous high-brightness patches (gray value 80-200) in the grayscale image, corresponding to the location of the actual high-temperature slag block.

[0036] To further extract the edges of the debris, the brightness difference between adjacent pixels within the suspected debris area is calculated. For each pixel (x, y) in the suspected area, the absolute difference between its grayscale value and that of the pixel to its right (x+1, y) and the pixel below it (x, y+1) is calculated, yielding the horizontal difference ΔH and the vertical difference ΔV. The brightness gradient value is then calculated using the Sobel gradient operator, i.e., G(x, y) = sqrt(ΔH² + ΔV²). The brightness gradient value quantifies the degree of brightness change at the pixel (the brightness gradient value at the edge is significantly higher than that in the inner area).

[0037] A second threshold Th2 (Th2>Th1) is set based on the average brightness value, where Th2 = Avg. L +3×σ=50+30=80. Pixels with brightness gradient values ​​greater than the second threshold within the suspected slag block area are marked as edge candidate points. These points correspond to locations of brightness abrupt changes and are potential locations of slag block edges.

[0038] Furthermore, spatial continuity verification is performed on the candidate edge points. Eight-neighborhood connectivity analysis is used to determine whether each candidate point has at least two adjacent candidate points (up, down, left, right, and diagonal). Continuous point sets that meet this condition are considered valid edge segments. Discrete, isolated candidate points (e.g., only one point or no adjacent points) are discarded. Subsequently, morphological closure operations (3×3 structuring elements) are applied to all valid edge segments to connect the small gaps, and the edge segments are fitted using the least squares method to ultimately form a closed polygonal contour, which is the initial contour of the slag block. The initial contour completely encloses the suspected slag block region in the image, and its boundary point coordinates can be directly used for subsequent slag block size calculations.

[0039] In summary, this application can accurately distinguish the brightness differences between slag blocks and the low-temperature environment within the slag well based on the brightness channel data extracted through grayscale processing, providing a reliable foundation for subsequent identification. By dividing the low-temperature environment region and calculating the average brightness value, a stable environmental benchmark can be established, making the setting of the first and second thresholds more targeted and accurate, effectively reducing the impact of environmental interference on slag block contour identification. Marking pixels with brightness values ​​greater than the first threshold as suspected slag block areas can initially lock the possible range of slag blocks, narrowing the target area for subsequent processing. By using a gradient operator to calculate the brightness gradient value and combining it with the second threshold to screen edge candidate points, the brightness abrupt change characteristics of slag block edges can be accurately captured, ensuring the sensitivity of edge identification. Verifying the spatial continuity of edge candidate points and forming a closed initial contour can completely present the shape of the slag block, avoiding contour distortion caused by isolated points or broken edges, thereby achieving accurate and stable identification of slag block contours.

[0040] In another implementation, after the slag block falls into the slag well, an initial color image is acquired by industrial cameras positioned on the top of opposite sides of the slag well. The red channel data in the image is directly extracted (because the brightness characteristics of the high-temperature slag block are more significant in the red band) as the basis for brightness difference analysis, without the need for grayscale processing. From the red channel data, an area in the corner of the slag well not covered by the slag block is selected as a low-temperature environment reference area. The maximum and minimum brightness values ​​of the pixels in this area are counted, and the average of the two is taken as the environmental reference brightness. Pixels in the red channel data whose brightness values ​​exceed twice the environmental reference brightness are marked as potential slag block pixels. The Prewitt gradient operator is used to calculate the brightness gradient in the horizontal and vertical directions for these pixels, and the pixels with the top 30% gradient values ​​are selected as edge points. Adjacent edge points are connected into continuous line segments using a chain code tracking algorithm. After removing short line segments with a length of less than 5 pixels, the remaining line segments are closed to form the initial outline of the slag block.

[0041] Step 120: Fuse the contours recognized by at least two cameras to obtain the three-dimensional contour of the slag block.

[0042] The contours identified by at least two cameras are fused to obtain the three-dimensional contour of the slag block. This requires using the initial two-dimensional contours collected by each camera as a basis, combined with the spatial geometric parameters obtained in the early stage, and through a series of operations such as coordinate transformation, point cloud registration, region fusion and surface reconstruction, to achieve a precise mapping from the planar contour to the three-dimensional shape. The specific implementation method is as follows.

[0043] First, the transformation from two-dimensional contours to three-dimensional coordinates needs to be completed based on the key parameters obtained in the early stage. These parameters include the distance parameters of each camera relative to the slag well (e.g., ...). Figure 4The installation height H1 and horizontal distance h3 of camera 1, the installation height H2 and horizontal distance h4 of camera 2, etc., are determined by laser ranging or equipment calibration. The pixel coordinates of the initial contour boundary points identified by each camera (such as the left and right boundary points identified by camera 1, and the front and rear boundary points identified by camera 2), and the angles between the boundary points and the center lines of the corresponding cameras (such as α1, α2, φ1, φ2, etc., which are calculated by pixel coordinates and camera intrinsic parameters). A three-dimensional world coordinate system is established with the center point at the bottom of the slag well as the origin (X-axis along the length of the slag well, Y-axis along the width, and Z-axis perpendicular to the grid plate and upwards). For each camera's two-dimensional boundary point, coordinates are converted based on its corresponding angle and distance parameters. For example, the pixel coordinates of the left boundary point of camera 1 in the two-dimensional image are converted to physical dimensions using a conversion coefficient (e.g., 1 pixel corresponds to 0.01 meters). Combined with the horizontal distance h3 and the angle α1, the X and Y coordinates of this point in the world coordinate system are calculated using the trigonometric function h2 = h3 × tanα1. The Z coordinate is the bottom height of the slag block (i.e., the height of the grid plate). Similarly, the three-dimensional coordinates of the right boundary point are calculated based on the angle α2, and the Z coordinate of the top boundary point is determined based on the vertical angle and the camera installation height H1. Using the same method, all boundary points identified by camera 2 are also converted to three-dimensional coordinates in the world coordinate system, forming two sets of three-dimensional points from different perspectives. Each set of points corresponds to the edge features of the slag block from that perspective.

[0044] Next, the 3D point sets from the two perspectives are registered to eliminate spatial positional deviations. Since the two cameras are installed on adjacent sides of the slag heap (e.g., the north and east sides), their shooting perspectives differ, potentially causing translational or rotational errors in the converted 3D point sets. The Iterative Closest Point (ICP) algorithm is used for registration: common feature points in the two point sets (such as the top vertex of the slag block, edge turning points, etc., points that can be identified from both perspectives) are selected, and the coordinate deviations of these points in the world coordinate system are calculated. The transformation matrix is ​​iteratively optimized using the least squares method to control the average distance error between the two point sets within 0.05 meters. For example, if the 3D coordinates of a vertex identified by camera 1 are (1.2, 0.8, 1.0), and the coordinates of the same point identified by camera 2 are (1.22, 0.79, 1.01), after iterative adjustment using the ICP algorithm, the coordinates of the two points can be corrected to (1.21, 0.795, 1.005), ensuring accurate alignment of boundary points from different perspectives in 3D space, laying the foundation for subsequent fusion.

[0045] Then, the registered 3D point sets are fused to integrate edge information from different perspectives. For overlapping areas in the two point sets (i.e., the edge of the slag block detected by both cameras), points with a coordinate deviation of less than 0.03 meters are retained, while isolated points caused by camera noise or local brightness interference are removed (e.g., a point is considered isolated if its distance from surrounding points exceeds 0.1 meters). For non-overlapping areas (i.e., edge parts unique to a single camera's perspective), they are fully incorporated into the fused point set to ensure coverage of the entire slag block surface. For example, if camera 1 fails to capture the southern edge of the slag block due to its limited perspective, while camera 2 can clearly identify this part, the southern boundary points of camera 2 (e.g., continuous points from 3D coordinates (2.0, 1.5, 0.9) to (2.5, 1.5, 0.9)) need to be added to the point set during fusion to avoid missing edge information. Meanwhile, the continuity of the fused point set is verified by combining the physical characteristics of the slag block (such as the continuous edge of the high-temperature slag block without obvious breakage). Discrete points are eliminated by 8-neighborhood connectivity analysis to ensure that the point set is continuous as a whole and accurately reflects the edge morphology of the slag block.

[0046] Finally, a closed 3D contour is constructed based on the fused 3D point set. The Delaunay triangulation algorithm is used to reconstruct the surface of the 3D point set, connecting discrete points into continuous triangular facets: first, all points are sorted according to their X and Y coordinates; then, a triangular mesh is constructed by maximizing the minimum angle principle, ensuring that the vertices of each triangular facet are points from the fused 3D point set, and that there is no overlap or gap between facets. Subsequently, a morphological closure operation (using a 5×5×5 spatial structuring element) is performed on the triangular mesh to fill tiny gaps (such as gaps smaller than 0.02 meters due to insufficient point density), and Gaussian filtering (with a standard deviation of 0.01 meters) is used to smooth the mesh edges, reducing edge distortion caused by point coordinate errors. The final three-dimensional profile must fully reflect the length, width, and height characteristics of the slag block. For example, based on the parameters of the two cameras in the previous example, the fused three-dimensional profile is approximately 1.58 meters long, 1.31 meters wide, and 1.2 meters high. Furthermore, the curvature and edge smoothness of the profile surface are consistent with the actual physical shape of the slag block, which can intuitively present the three-dimensional structure of the slag block. This provides a reliable three-dimensional morphological benchmark for subsequent profile correction based on dynamic brightness changes and precise slag squeezing operation of the hydraulic shut-off gate.

[0047] Step 130: Continuously acquire images of the debris block using at least two cameras to capture the dynamic changes in the brightness of the debris block.

[0048] Using the initial outline of the slag block determined in step 110 as a reference, a dynamic tracking window is set. The window boundary extends 20 pixels in each direction from the initial outline to form a rectangular tracking area, ensuring that even if the slag block is displaced due to minor vibrations, it can still be completely contained within the window. The two cameras used continuously acquire images at fixed time intervals, with the acquisition frequency set to 1 frame / second. The camera parameters are consistent with those in step 110.

[0049] For each frame, firstly, an image region within the dynamic tracking window is extracted, and the luminance channel data of this region is extracted using the same method as in step 110. Then, based on the initial contour boundary coordinates determined in step 110, a blocky region corresponding to the initial contour is defined in the current frame image. Only the pixel luminance values ​​within this region are statistically analyzed to calculate the average luminance value of the blocky region in that frame image. For example, the average luminance value of frame t is denoted as Lt, and it is calculated by summing the luminance values ​​of all pixels within the region and dividing by the total number of pixels.

[0050] To quantify the brightness change trend, the brightness decay rate of the slag block region corresponding to the initial contour is calculated. Specifically, the average brightness values ​​of five consecutive frames (Lt-4, Lt-3, Lt-2, Lt-1, Lt) are selected, and a linear fitting method is used to obtain the slope k of the brightness change over time. The brightness decay rate α is then defined as... It is used to characterize the rate of decrease in brightness relative to the initial value per unit time, where L0 is the average brightness value of the initial frame.

[0051] The preset attenuation threshold is 5% / second (i.e., the brightness decreases by more than 5% of the initial brightness per second). When the real-time calculated brightness attenuation rate α exceeds this threshold, the slag block is determined to have entered the cooling stage. At this time, the temperature of the slag block has dropped below 600℃, and the difference between the brightness and the environment gradually decreases, providing a status basis for subsequent slag block position identification.

[0052] This application effectively narrows the image acquisition and processing range by setting a dynamic tracking window based on the initial contour, reducing interference from irrelevant background areas on slag block monitoring, improving image processing efficiency, and ensuring that the slag block is always within the monitoring range, avoiding tracking loss due to minor displacement of the slag block. By continuously acquiring images within the dynamic tracking window at fixed time intervals and extracting brightness channel data, the brightness information of the slag block area can be captured in real time. Based on this, the average brightness value of the slag block area corresponding to the initial contour can be calculated, accurately reflecting the overall brightness level change of the slag block. The brightness decay rate is calculated based on the average brightness value of consecutive frames, which can quantify the trend of slag block brightness change over time. When the brightness decay rate exceeds the decay threshold, it indicates that the slag block has entered the cooling stage, allowing for timely judgment of the slag block's state transition. This state indication is crucial for maintaining slag block position recognition, because after the slag block enters the cooling stage, its brightness will gradually approach or even fall below the ambient brightness. Knowing this state in advance can prevent slag block contour recognition failure due to brightness reduction, ensuring continuous and stable tracking of the slag block and providing a consistent state basis for subsequent dynamic adjustment of slag block boundaries and size judgment.

[0053] In another implementation, the dynamic tracking window is set as a rectangular area extending 15 pixels outward from the initial contour, using the initial contour as the boundary. Industrial cameras positioned on the top of opposite sides of the slag well continuously capture color images within the window at a fixed interval of 0.5 seconds per frame. The red channel data within the tracking window of each frame is directly extracted, and the slag block area is located based on the initial contour coordinates. The red channel brightness values ​​of all pixels within this area are statistically analyzed and their average values ​​are calculated. Based on the average red channel brightness values ​​of 20 consecutive frames, the slope of the brightness change curve over time is obtained through linear regression fitting, and this slope is used as the brightness change rate. The preset brightness change rate threshold is -5 (i.e., the average red channel brightness decreases by 5 units every 0.5 seconds). When the calculated brightness change rate is less than this threshold, it is determined that the slag block brightness is in a state of significant decline.

[0054] Step 140: Based on the correlation between dynamic changes in brightness and the contour, correct the contour of the slag block.

[0055] Based on the brightness decay rate calculated in step 130, the brightness decay characteristics of the pixel region corresponding to the initial contour edge point are analyzed in depth. Since the edge of the block dissipates heat faster, its brightness decay rate will theoretically be higher than that of the inner region. A contour correction threshold is set, for example, to 15%. When the difference between the brightness decay rate of an initial contour edge point and the brightness decay rate of the adjacent inner region exceeds this contour correction threshold, the edge point is determined to be a suspected misjudged edge point. This is because such a large difference in decay rate may mean that the edge point is not a true block edge, but a misjudgment caused by image noise, local brightness interference, or other factors.

[0056] For regions identified as potentially misjudged edge points, the brightness gradient is recalculated. During the calculation, brightness channel data from consecutive frames within the dynamic tracking window are used to further filter edge candidate points. For example, the Sobel gradient operator is used to calculate the brightness gradient value, and pixels with brightness gradient values ​​greater than a second threshold (e.g., the second threshold is set to 80) are marked as edge candidate points. To ensure the accuracy and reliability of the selected edge candidate points, the candidate points must maintain a brightness gradient value greater than the threshold in three consecutive frames and exhibit spatial continuity (by using 8-neighborhood connectivity analysis to determine if each candidate point has at least two adjacent candidate points).

[0057] Based on the re-screened edge candidate points, suspected misjudged edge points are replaced or deleted. After replacement or deletion, the corrected edge segments undergo rigorous spatial continuity verification. Through 8-neighborhood connectivity analysis, it is ensured that each point on an edge segment can be effectively connected to its neighboring points, forming continuous line segments. Discrete points or short line segments that do not meet the spatial continuity requirements are eliminated or repaired, ultimately forming new valid edge segments. Based on these new valid edge segments, the initial contour of the slag block is updated. Simultaneously, the contours corrected by at least two cameras are re-fused. The fusion process is similar to that used when acquiring the 3D contour, employing the Iterative Closest Point (ICP) algorithm for registration, controlling the average distance error between the two point sets to within 0.05 meters, thus obtaining the corrected 3D contour.

[0058] In another implementation, the initial contour is used as a reference to divide the area into several sub-regions. The brightness change rate (the amount of change in brightness value per unit time) of each sub-region is calculated in real time, and a dynamic correlation threshold is set (e.g., 30% of the difference in brightness change rate between adjacent sub-regions). When the difference in brightness change rate between a sub-region and its adjacent regions exceeds the threshold and remains at this difference for three consecutive frames, it is determined that the region has a contour deviation. If the brightness of a sub-region drops sharply and the absolute value of the change rate is greater than that of the surrounding area, it is speculated that the edge of the debris has peeled off, and the boundary of the region is shrunk inward to a pixel position with stable brightness. If the brightness of a sub-region drops slowly and the change rate is lower than that of the surrounding area, it is speculated that a new debris has attached, and the boundary is expanded outward to the point of brightness change. At the same time, the correction magnitude is adjusted in combination with the area ratio of the sub-region to ensure that the contour correction matches the actual shape change.

[0059] After completing the construction and initial correction of the three-dimensional contour of the slag block, in order to capture the accumulation of new slag blocks in real time, it is necessary to further expand the monitoring range and continuously collect image data.

[0060] Using the initial contour of the slag block determined in step 110 as a reference, the dynamic tracking window is set as a rectangular area extending 30 pixels outwards from the initial contour boundary in all directions. The window coordinate range remains consistent with that during the previous contour recognition to ensure coverage of areas where the slag block may expand or new slag blocks may fall. The camera continuously acquires images within the window at a fixed interval of 0.5 seconds per frame. After preprocessing (such as noise reduction and grayscale conversion), the luminance channel data is extracted from the acquired images. Using the same grayscale processing method as in step 110, the RGB three channels are converted into single-channel grayscale data using a weighted average method.

[0061] When monitoring new high-brightness regions, the brightness channel data of the current frame and the previous 5 frames are first differentially analyzed using the background subtraction method: the difference between the brightness value of each pixel in the current frame and the average brightness value of the corresponding pixels in the previous 5 frames is calculated. When the difference is greater than 20 (in grayscale units), the region where the pixel is located is marked as a suspected new region. Subsequently, the blocky regions corresponding to the initial contour (i.e., the contour range identified in step 110) are excluded from the suspected new regions, and the remaining regions are the new high-brightness candidate regions to be verified. This step can be achieved through masking operations, marking the pixels within the initial contour as "identified" and retaining only the suspected regions outside the contour for subsequent analysis.

[0062] For new high-brightness candidate regions, calculate their average brightness value and compare it with the average brightness value of the low-temperature environment region determined in step 110. Assume the average brightness value Avg of the low-temperature environment region obtained earlier through K-means clustering is used. L =50, with the preset multiplier set to 1.8, resulting in a brightness threshold of 50 × 1.8 = 90. If the average brightness value of the candidate area is greater than 90 (for example, the statistical average brightness of a candidate area is 110), then its brightness is determined to meet the characteristics of a new slag block. Because newly fallen slag blocks have a higher temperature (usually above 800℃), their brightness is significantly higher than that of cooled old slag blocks and low-temperature environments. This threshold can effectively distinguish new slag blocks from environmental interference (such as steam reflection).

[0063] Further analysis is needed to determine the spatial overlap between the high-brightness area and the initial contour. The overlap is determined by calculating the proportion of pixels intersecting the high-brightness area with the initial contour to the total number of pixels in the high-brightness area. For example, if a high-brightness area has 200 pixels, and 70 of them overlap with the initial contour, then the overlap is 70 / 200 = 35%. When the overlap exceeds 30%, spatial overlap is considered to exist, meaning that new slag blocks have fallen on top of or to the edge of old slag blocks, forming an accumulation. This overlap threshold is set based on the characteristic in actual working conditions that "new slag blocks usually come into contact with existing slag blocks," avoiding misjudging independently falling small slag blocks as accumulations.

[0064] When all the above conditions are met (brightness meets the standard and overlap exceeds 30%), it indicates the existence of a new debris block accumulation. At this time, the boundary coordinates of the high-brightness area (pixel coordinates obtained through edge detection) are merged with the initial contour, and the dynamic tracking window is updated. The window boundary is adjusted to expand outward by 30 pixels after merging, ensuring that the new area is fully included in the monitoring. At the same time, the new area is included in the brightness dynamic monitoring range. Starting from the next frame, the average brightness value and brightness decay rate of the merged area are calculated (the calculation method is the same as in step 130, that is, the decay rate is obtained by fitting the slope of the average brightness of 5 consecutive frames), and the initial brightness characteristics of the new area (such as the average brightness of 110 detected for the first time) are retained for subsequent differentiation from the brightness change trend of the old debris block area. The brightness decay rate of the new debris block is usually lower than that of the old debris block (due to the higher initial temperature), and this feature can help verify the authenticity of the accumulated area.

[0065] To more accurately correct the contour, it is necessary to first calculate the displacement amplitude and brightness range overlap of the boundary points in the slag block region corresponding to the initial contour. When calculating the displacement amplitude of the boundary points, four vertices (A, B, C, D) of the initial contour in 20 consecutive frames are selected as feature points. The pixel coordinates of each point in each frame are obtained through the SIFT feature matching algorithm. The coordinate difference between each frame and the first frame is calculated (e.g., the coordinates of point A in the 10th frame are (305, 203), and the coordinates of the first frame are (300, 200), so Δx = 5 pixels and Δy = 3 pixels). Then, according to the conversion coefficient between pixels and actual size (e.g., 0.01 meters / pixel), the pixel displacement is converted into the actual physical displacement (Δx actual = 5 × 0.01 = 0.05 meters, Δy actual = 3 × 0.01 = 0.03 meters). The average displacement value of each point in 20 frames is taken as the displacement amplitude of the boundary points (e.g., the average displacement of point A is 0.04 meters, and the average displacement of the overall boundary points is 0.035 meters). This parameter is used to distinguish between the natural displacement of slag blocks caused by minor vibrations and settling and the actual size changes, so as to avoid misjudging the displacement as contour expansion.

[0066] When calculating the overlap of brightness ranges, the brightness histograms of the initial contour region in the current frame and the previous 5 frames are statistically analyzed (the horizontal axis represents brightness values ​​0-255, and the vertical axis represents the number of pixels). The overlap is obtained by calculating the ratio of the intersection area to the union area of ​​the two histograms. For example, if the intersection area of ​​the current frame's brightness histogram and the average histogram of the previous 5 frames is 7200, and the total area is 10000, then the overlap is 72%. A higher overlap indicates better continuity of brightness characteristics in the initial region and less influence from external disturbances (such as dust or changes in lighting), providing a reliable brightness benchmark for subsequent corrections.

[0067] When new slag blocks accumulate, collect the following correction information: the overlap area between the new high-brightness area and the initial outline (obtained by pixel count, e.g., 800 pixels corresponds to an actual area of ​​800 × 0.01 × 0.01 = 0.08 square meters), the overlap position parameters (e.g., located in the upper right corner of the initial outline, with an actual distance of 0.3 meters from point B); the expansion distance of the overall outline after the new area merges with the initial outline compared to the initial outline (maximum 70 pixels in the X direction, i.e., 0.7 meters, maximum 80 pixels in the Y direction, i.e., 0.8 meters), and the expansion area ratio (expansion area 1500 pixels, total merged area 6500 pixels, expansion ratio 1500 / 6500≈23%).

[0068] Based on the above parameters, the length, width, and height of the slag block are corrected: the initial length (X direction) is 4.0 meters (400 pixels × 0.01), combined with the X-direction expansion distance of 0.7 meters and the average displacement of the boundary points of 0.035 meters, the corrected length is 4.0 + 0.7 + 0.035 × 2 = 4.77 meters (double the displacement amplitude to compensate for boundary fluctuations); the initial width (Y direction) is 3.0 meters (300 pixels × 0.01), combined with the Y-direction expansion distance of 0... With an overlap of 0.8 meters and 72% in brightness range (high overlap indicates stable expansion), the corrected width is 3.0 + 0.8 + 0.035 × 2 = 3.87 meters; the initial height is 1.5 meters. Based on the overlap area ratio of the new area and the initial contour of 16% (0.08 / 0.5) and the expansion area ratio of 23%, calculated using the height expansion coefficient (expansion area ratio × 0.5), the corrected height is 1.5 + 1.5 × 23% × 0.5 ≈ 1.67 meters. Through the fusion of multi-dimensional parameters, it is ensured that the corrected contour not only includes the accumulated amount of new slag blocks but also eliminates the interference of natural displacement, accurately reflecting the actual physical shape of the slag blocks.

[0069] Through this series of operations, the accumulation process of new slag blocks can be captured in real time, and the outline can be dynamically updated based on brightness characteristics and spatial relationships. This provides more accurate dimensional basis for the slag extrusion operation of the hydraulic shut-off gate, avoiding incomplete slag extrusion or equipment overload caused by the failure to identify new slag blocks.

[0070] After monitoring and contour correction of the new slag accumulation, in order to accurately track the dynamic movement of independent small slag blocks and their interaction with the initial slag block, the dynamic tracking window needs to be adjusted to adapt to the movement characteristics of the small slag blocks.

[0071] Using the initial contour of the slag block determined in step 110 as a reference, the dynamic tracking window is set to a rectangular area extending 25 pixels outward from the boundary of the initial contour in all directions, ensuring coverage of the possible movement range of the small slag block. The camera continuously acquires images at an interval of 0.5 seconds / frame. The acquired images are then processed into grayscale and used as the basis data for subsequent analysis.

[0072] When identifying independent high-brightness slag block regions, the slag block regions corresponding to the initial contour are first excluded from the dynamic tracking window (pixels within the initial contour are marked as "identified" using a masking operation), and only the regions outside the contour are analyzed. The average brightness value Avg of the low-temperature environment within this region is calculated. L (Same as step 110, obtained through K-means clustering, for example, Avg) L =50), set the preset multiplier to 2, thus obtaining the brightness threshold = 50 × 2 = 100; for pixels outside the window outline, filter out continuous pixel areas with a brightness value greater than 100, and the area of ​​this area is less than 500 pixels (corresponding to an actual area of ​​0.05 square meters, based on the conversion coefficient of 1 pixel = 0.01 meters), and judge them as dynamically moving small debris blocks. This area threshold setting is based on the characteristic that "the size of independent small debris blocks is usually less than 0.1 cubic meters" in actual working conditions, to avoid misjudging large new debris blocks as small debris blocks.

[0073] When tracking the movement trajectory of the small debris, the SIFT feature matching algorithm is used to extract the center pixel coordinates of the independent high-brightness regions identified in each frame of the image. The minimum bounding rectangle of the region is determined by edge detection, and the center of the rectangle is the center coordinate of the small debris (e.g., the center coordinate of the t-th frame is (x...). t ,y t Linear fitting is performed on the center coordinates of 10 consecutive frames to obtain the motion trajectory equation. For example, the fitting result is y = 0.6x + 120 (pixel coordinates). This equation can characterize the movement path of the small debris in the image. During the fitting process, if the region is not detected in a certain frame, its possible position is predicted by the trajectory of the previous 3 frames. If it is not predicted in 3 consecutive frames, the trajectory is terminated (excluding the case that it has fallen out of the window).

[0074] When determining whether a small debris block is occluded by the debris block corresponding to the initial contour, the system first monitors whether its trajectory enters the range of the initial contour. The initial contour is represented as a closed polygon in the image. The coordinate point-in-polygon algorithm is used to determine whether the center coordinates of the small debris block fall within this polygon (for example, the center coordinates (650, 320) in frame 8 fall within the pixel range (500-700, 200-400) of the initial contour). Simultaneously, the system checks whether the brightness channel data of the high-brightness area is interrupted: if the average brightness of the original high-brightness area drops sharply from 120 to 80 (below the brightness threshold of 100) in three consecutive frames, and no other area inherits its brightness characteristics, then the brightness data is considered interrupted. This is because after the small debris block is occluded by the large debris block, the high-temperature surface cannot be captured by the camera, resulting in a significant reduction in brightness.

[0075] When both of the above conditions are met, it indicates that the independent high-brightness slag block region is occluded by the slag block corresponding to the initial contour. At this time, the parameters of the occluded region are extracted: the boundary point coordinates are determined by the brightness gradient change at the interruption position, and the brightness gradient value (horizontal gradient) of the region is calculated using the Sobel operator. Vertical gradient Combined gradient Pixels with gradient values ​​greater than 80 (the second threshold, the same as the edge detection threshold in step 110) are marked as boundary points. For example, the pixel coordinates of the boundary points are (650, 320), (660, 330), (640, 335), etc. Then, they are converted to actual coordinates (6.50, 3.20) meters, (6.60, 3.30) meters, etc., using the pixel-to-actual-size conversion coefficient (1 pixel = 0.01 meters).

[0076] In the shape parameters, curvature features are obtained by calculating the second derivative of the curve fitted to the boundary points, for example, by performing polynomial fitting on the boundary point sequence (such as a cubic polynomial y = ax). 3 +bx 2 Given (+cx+d), calculate the second derivative y”=6ax+2b, whose absolute value is the curvature value, and the average value is used to characterize the curvature of the occluded area; the edge smoothness is obtained by calculating the variance of the angle change of the line connecting adjacent boundary points. The smaller the variance, the smoother the edge (e.g., if the variance of the angle change of an occluded area is 5°, it is judged as a relatively smooth edge). The area parameter is obtained by counting the total number of pixels in the occluded area (e.g., 400 pixels) and combining it with the conversion coefficient to obtain the actual area = 400×0.01×0.01=0.04 square meters.

[0077] When revising the initial contour based on the above parameters, the actual coordinates of the boundary points of the occluded area are first fused with the corresponding edge points of the initial contour. For example, if the right edge point of the initial contour is (6.40, 3.10) meters and the boundary point of the occluded area is (6.50, 3.20) meters, the contour edge point is adjusted to (6.50, 3.20) meters to include the occluded portion. Based on the area parameter, the width of the initial contour is increased by the corresponding dimension of 0.04 square meters (e.g., expanding the width by 0.1 meters). Combining the curvature and smoothness parameters in the shape parameters, a Gaussian filter (standard deviation 0.02 meters) is applied to the corrected edge to ensure a smooth transition between the new edge segment and the original contour (e.g., the variance of the angle change of the edge point after filtering decreases from 10° to 6°). The final corrected contour includes the occluded small slag chunks while maintaining spatial continuity, accurately reflecting the actual shape of the slag chunks.

[0078] This process can accurately capture the occlusion interaction between small and large slag blocks, dynamically update the outline size and shape, and provide a more realistic size basis for the slag squeezing operation of the hydraulic shut-off gate, avoiding the problem of underestimation of the outline due to occlusion.

[0079] After monitoring and correcting the outline of individual small slag blocks, in order to further accurately update the slag block outline in conjunction with the mechanical extrusion process, it is necessary to introduce the operating parameters of the hydraulic shut-off gate as the basis for correction.

[0080] During the process of correcting the profile by adjusting the extrusion depth in conjunction with the hydraulic shut-off gate, the morphological changes of the slag block under mechanical extrusion can be achieved through... Figure 5 Presented intuitively. Figure 5 The spatial relationship between the contours of the slag block before and after deformation and the monitoring range of the shut-off gate and camera is clearly displayed: the solid-line frame represents the initial contour of the slag block before compression, and its boundary points (such as vertices B and D) define the shape of the slag block when it is not subjected to external force; the dashed-line frame represents the deformed contour after compression, reflecting the contraction, displacement, and shape distortion of the slag block edges after the shut-off gate is applied (such as shortening in length, narrowing in width, and increased local edge curvature). The shut-off gate is illustrated with a rectangular structure, and the distance between it and the initial contour of the slag block represents its relative position before compression, while the degree of closeness between it and the deformed contour after compression intuitively reflects the compression depth; the overall range of the camera is marked with a dashed-line frame, completely covering the area of ​​the slag block before and after compression, verifying the effective capture capability of the dynamic tracking window for changes in the shape of the slag block, and ensuring that the image acquisition during the compression process always includes the complete slag block area.

[0081] When collecting data on the penetration depth of the hydraulic shut-off gate in the slag well, a magnetostrictive displacement sensor needs to be installed on the hydraulic drive cylinder of the shut-off gate to collect the expansion and contraction of the gate in real time. This expansion and contraction is the penetration depth data. For example, when the shut-off gate advances from its initial position (not in contact with the slag block) towards the slag block, the sensor feedback value gradually increases from 0 to 0.45m. This 0.45m reflects the degree of compression of the slag block by the shut-off gate; the larger the value, the more thorough the compression. The sensor data is transmitted to the control terminal via an interface and synchronized with the image data collected by the camera to ensure the correspondence between the compression action and the image frame.

[0082] When acquiring images within the dynamic tracking window, the initial contour determined in step 110 is used as a reference, and the window boundary is extended by 20 pixels in each direction to form a rectangular monitoring area. The same camera used in step 110 continuously acquires images at a frequency of 1 frame / second. The initial image before slag extrusion is the first frame within 10 seconds before the shut-off gate starts (denoted as frame t0), and the continuous image frames after slag extrusion are 20 frames within 20 seconds after the shut-off gate starts (denoted as frames t1 to t20). The acquired images are preprocessed: first, they are converted to grayscale using a weighted average method (Gray = 0.299 × R + 0.587 × G + 0.114 × B), then noise is removed using a 3 × 3 Gaussian filter (standard deviation 0.8), and finally, single-channel brightness data is extracted for subsequent analysis.

[0083] When determining the boundary point coordinates before and after slag extrusion based on luminance channel data, for the initial image before slag extrusion (frame t0) and each frame image after slag extrusion (such as frame t10), the Sobel gradient operator (horizontal convolution kernel Gx = [-1,0,1; -2,0,2; -1,0,1], vertical convolution kernel Gy = [-1,-2,-1; 0,0,0; 1,2,1]) is used to calculate the luminance gradient value. Pixels with gradient values ​​greater than 80 (the second threshold, the same as in step 110) are selected as edge points. Then, continuous edge segments are formed through 8-neighborhood connectivity analysis, and the vertices of the edge segments are taken as boundary points. For example, the pixel coordinates of a boundary point on the initial contour before slag extrusion are (300, 200). According to the conversion coefficient of 0.01m per pixel (determined by camera calibration), its actual coordinates are (3.00, 2.00)m. After slag extrusion (frame t10), the pixel coordinates of this boundary point become (310, 210), and the actual coordinates are (3.10, 2.10)m, reflecting the positional shift caused by extrusion.

[0084] When calculating the displacement amplitude, profile morphology change, and area change of the boundary points before and after slag extrusion, the displacement amplitude of the boundary points is calculated using the Euclidean distance formula: for the same boundary point before and after slag extrusion (such as the vertex mentioned above), The average displacement amplitude of all boundary points is calculated (e.g., 0.35m). Changes in profile morphology include changes in perimeter (14.0m before slag extrusion, 12.5m after, a decrease of 1.5m) and changes in average curvature (calculated by the change in the angle between adjacent boundary points, e.g., an increase from 0.05 / m to 0.08 / m, indicating a more tortuous profile). Area changes are calculated by counting the total number of pixels within the profile before and after slag extrusion: 120,000 pixels before slag extrusion, corresponding to an actual area of ​​120,000 × 0.01 × 0.01 = 12.0m². 2 After descaling, the pixel count is 98,000, corresponding to an area of ​​9.8m². 2 The area reduction rate is approximately (12.0-9.8) / 12.0 ≈ 18.3%.

[0085] When correcting the profile based on data on indentation depth, displacement amplitude, morphological changes, area changes, and brightness channel data, a multi-parameter correlation model needs to be established: First, based on the empirical relationship between indentation depth and profile dimensions (e.g., experimental data shows that for every 0.1m increase in indentation depth, the average length of the slag block decreases by 0.3m), combined with the actual indentation depth of 0.45m, the initial estimate of the length reduction is 0.45×3=1.35m; then, combined with the boundary point displacement amplitude (average 0.35m), the estimated value is corrected (e.g., 1.35+0.35=1.70m). Second, referring to the area change (reduction of 18.3%), the width and height are adjusted (the width reduction ratio is taken as 0.8 times the area reduction ratio, i.e., 14.6%; the height reduction ratio is taken as 1.2 times, i.e., 22.0%). Simultaneously, the rationality of the correction was verified by combining brightness channel data: after extrusion, due to heat dissipation from breakage, the average brightness of the slag decreased from the initial 150 to 127.5 (a decrease of 15%), consistent with the trend of morphological changes. Ultimately, the initial length of 4.0m was corrected to 4.0 - 1.70 = 2.30m, the width of 3.0m to 3.0 × (1 - 14.6%) = 2.56m, and the height of 1.2m to 1.2 × (1 - 22.0%) = 0.94m. The corrected contour needed to pass an 8-neighborhood connectivity check to ensure edge continuity and complete matching with the extrusion depth of the shut-off gate and the brightness distribution characteristics in the image, providing a precise basis for subsequent parameter adjustments in the slag extrusion operation.

[0086] Furthermore, as a response to the above Figure 1 The implementation of the method embodiment shown in this application provides a slag block contour recognition device. This device embodiment corresponds to the foregoing method embodiments. For ease of reading, this embodiment will not repeat the details of the foregoing method embodiments one by one, but it should be clear that the device in this embodiment can correspondingly implement all the contents of the foregoing method embodiments. Specifically, as shown... Figure 6 As shown, the slag block contour recognition device 600 includes:

[0087] The identification module 610 is used to acquire an initial image of the slag block on the grid plate at the bottom of the slag well through at least two cameras set outside the slag well, and to perform brightness gradient analysis based on the brightness difference between the slag block and the low-temperature environment inside the slag well in the initial image to initially identify the outline of the slag block.

[0088] The fusion module 620 is used to fuse the contours recognized by at least two cameras to obtain the three-dimensional contour of the slag block.

[0089] The capture module 630 is used to continuously acquire images of the debris block through at least two cameras and capture the dynamic changes in the brightness of the debris block;

[0090] The correction module 640 is used to correct the contour of the slag block based on the correlation between the dynamic change of brightness and the contour, so that the hydraulic shut-off gate can adjust the extrusion depth according to the contour of the slag block to achieve precise slag extrusion.

[0091] Furthermore, such as Figure 6 As shown, the recognition module 610 is specifically used to process the initial image to extract brightness channel data, distinguish between slag blocks and low-temperature environment areas within the slag well based on the brightness channel data, and mark suspected slag block areas. The suspected slag block areas are defined by setting a first threshold based on the brightness difference between the slag blocks and the low-temperature environment within the slag well, and marking the areas where the brightness values ​​of the pixels in the brightness channel data are greater than the first threshold. Brightness gradient calculation is performed on the suspected slag block areas to filter out edge candidate points. Spatial continuity analysis is performed on the edge candidate points to form effective edge segments. All effective edge segments are connected and closed to form the initial outline of the slag block.

[0092] Furthermore, such as Figure 6 As shown, the identification module 610 is specifically used to calculate the brightness difference between adjacent pixels in the suspected slag block area, and to calculate the brightness gradient value of the pixels in the suspected slag block area through a gradient operator based on the brightness difference. The brightness gradient value is used to quantify the brightness change rate between the edge of the slag block and the low-temperature environment inside the slag well. Pixels in the suspected slag block area with a brightness gradient value greater than a second threshold are marked as edge candidate points, and the second threshold is greater than the first threshold.

[0093] Furthermore, such as Figure 6 As shown, step module 630 is specifically used to set a dynamic tracking window based on the initial contour of the slag block, and continuously acquire images within the dynamic tracking window at fixed time intervals using at least two cameras; extract the brightness channel data within the dynamic tracking window in each frame image, calculate the average brightness value of the slag block area corresponding to the initial contour based on the brightness channel data; and calculate the brightness attenuation rate of the slag block area corresponding to the initial contour based on the average brightness value of consecutive frames.

[0094] Furthermore, such as Figure 6As shown, the correction module 640 is specifically used to analyze the brightness decay characteristics of the pixel region corresponding to the edge point of the initial contour based on the brightness decay rate. The brightness decay characteristics include the difference in brightness decay rate between the edge point and the internal region. When the difference between the brightness decay rate of the initial contour edge point and the brightness decay rate of the adjacent internal region exceeds the contour correction threshold, the edge point is determined to be a suspected misjudged edge point. The brightness gradient of the region where the suspected misjudged edge point is located is recalculated, and the edge candidate points are re-selected by combining the brightness channel data of the continuous frames in the dynamic tracking window. The suspected misjudged edge points are replaced or deleted according to the re-selected edge candidate points, and the spatial continuity of the corrected edge segment is checked to form a new effective edge segment. The initial contour of the block is updated based on the new effective edge segment, and the contours corrected by at least two cameras are re-fused to obtain the corrected three-dimensional contour.

[0095] Furthermore, such as Figure 6 As shown, the correction module 640 is also used to monitor whether a new high-brightness area appears in the dynamic tracking window, other than the slag block area corresponding to the initial contour; if a new high-brightness area appears, it is determined whether the brightness value of the new high-brightness area is greater than a preset multiple of the average brightness value of the low-temperature environment area; if it is greater, it is determined whether the new high-brightness area and the initial contour have spatial overlap; if there is spatial overlap, it indicates that there is new slag block accumulation, and the new high-brightness area is included in the brightness dynamic monitoring range.

[0096] Furthermore, such as Figure 6 As shown, the correction module 640 is also used to calculate the boundary point displacement amplitude and brightness range overlap of the slag block region corresponding to the initial contour based on the brightness channel data in the dynamic tracking window; when there is a new slag block accumulation, the overlap area between the new high brightness region and the initial contour, the expansion distance and expansion area ratio of the overall contour after the new high brightness region and the initial contour are merged compared with the initial contour, are used as correction information; the contour of the slag block is corrected based on the boundary point displacement amplitude, brightness range overlap and correction information.

[0097] Furthermore, such as Figure 6As shown, the correction module 640 is also used to track the motion trajectory of newly emerging independent high-brightness slag block regions outside the slag block region corresponding to the initial contour in the dynamic tracking window, in the initial image and continuous image frames acquired within the dynamic tracking window. The motion trajectory is determined based on the brightness channel data corresponding to the independent high-brightness slag block region. The independent high-brightness slag block region is a dynamically moving small slag block. The brightness value of the small slag block is greater than the average brightness value of the low-temperature environment region by a preset multiple, and it has no spatial overlap with the initial contour. When the motion trajectory enters the range of the slag block region corresponding to the initial contour, and the brightness channel data of the independent high-brightness slag block region is interrupted, it indicates that the independent high-brightness slag block region is occluded by the slag block corresponding to the initial contour. The boundary point coordinates, shape parameters, and area parameters of the occluded region are extracted. The boundary point coordinates are determined based on the brightness gradient change at the interruption position. The shape parameters include the curvature features and edge smoothness of the boundary of the occluded region. The area parameters include the total number of pixels contained in the occluded region and the actual area calculated based on the total number of pixels. The contour of the slag block is corrected based on the boundary point coordinates, shape parameters, and area parameters.

[0098] Furthermore, such as Figure 6 As shown, the correction module 640 is also used to collect the squeezing depth data of the hydraulic shut-off gate of the slag well. The squeezing depth data is used to reflect the degree of squeezing of the slag block by the hydraulic shut-off gate. It acquires the initial image before slag squeezing and the continuous image frames after slag squeezing within the dynamic tracking window. Based on the brightness channel data of the initial image before slag squeezing and the continuous image frames after slag squeezing, it determines the boundary point coordinates of the slag block area corresponding to the initial contour before slag squeezing and the boundary point coordinates of the slag block area after slag squeezing, respectively. It determines the boundary point displacement amplitude, contour shape change and area change before and after slag squeezing based on the boundary point coordinates before and after slag squeezing. Based on the squeezing depth data, boundary point displacement amplitude, contour shape change, area change and brightness channel data, it corrects the contour of the slag block.

[0099] Furthermore, such as Figure 6 As shown, the recognition module 610 is also used to acquire the distance parameters of at least two cameras relative to the slag well, as well as the boundary points of the three-dimensional contour in the images captured by each camera; using the center line of each camera as a reference, calculate the angle formed between each boundary point of the three-dimensional contour and the corresponding camera; calculate the length, width and height of the slag block according to the distance parameters and the angle; determine the coordinate information of the slag block according to the length, width and height, so that the hydraulic shut-off gate can optimize the slag squeezing path based on the coordinate position of the slag block.

[0100] Optionally, the slag block contour recognition device may be an electronic device with data processing capabilities, or a functional module within the electronic device, without limitation.

[0101] For example, the electronic device can be a server, which can be a single server or a server cluster consisting of multiple servers. As another example, the electronic device can be a mobile phone, tablet computer, desktop computer, laptop computer, handheld computer, notebook computer, ultra-mobile personal computer (UMPC), netbook, as well as cellular phones, personal digital assistants (PDAs), augmented reality (AR) devices, virtual reality (VR) devices, and other terminal devices. As yet another example, the electronic device can also be a recording device, video surveillance equipment, etc. This application does not impose any special limitations on the specific form of the electronic device.

[0102] The following example uses a slag block contour recognition device as an electronic device. Figure 7 As shown, Figure 7 The hardware structure of an electronic device 700 provided in this application.

[0103] like Figure 7 As shown, the electronic device 700 includes a processor 710, a communication line 720, and a communication interface 730.

[0104] Optionally, the electronic device 700 may also include a memory 740. The processor 710, memory 740, and communication interface 730 can be connected via a communication line 720.

[0105] The processor 710 can be a central processing unit (CPU), a network processor (NP), a digital signal processor (DSP), a microprocessor, a microcontroller, a programmable logic device (PLD), or any combination thereof. The processor 710 can also be any other device with processing capabilities, such as a circuit, device, or software module, without limitation.

[0106] In one example, processor 710 may include one or more CPUs, for example Figure 7 CPU0 and CPU1 in the CPU.

[0107] As an optional implementation, the electronic device 700 includes multiple processors; for example, in addition to processor 710, it may also include processor 770. A communication line 720 is used to transmit information between the components included in the electronic device 700.

[0108] Communication interface 730 is used for communicating with other devices or other communication networks. These other communication networks can be Ethernet, Radio Access Network (RAN), Wireless Local Area Networks (WLAN), etc. Communication interface 730 can be a module, circuit, transceiver, or any device capable of enabling communication.

[0109] Memory 740 is used to store instructions. These instructions can be computer programs.

[0110] The memory 740 can be a read-only memory (ROM) or other type of static storage device that can store static information and / or instructions; it can also be a random access memory (RAM) or other type of dynamic storage device that can store information and / or instructions; it can also be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, etc., without limitation.

[0111] It should be noted that the memory 740 can exist independently of the processor 710, or it can be integrated with the processor 710. The memory 740 can be used to store instructions, program code, or some data, etc. The memory 740 can be located inside or outside the electronic device 700, without restriction.

[0112] The processor 710 is configured to execute instructions stored in the memory 740 to implement the communication method provided in the following embodiments of this application. For example, when the electronic device 700 is a terminal or a chip in a terminal, the processor 710 can execute instructions stored in the memory 740 to implement the steps performed by the transmitting end in the following embodiments of this application.

[0113] As an optional implementation, the electronic device 700 also includes an output device 750 and an input device 760. The output device 750 can be a display screen, speaker, or other device capable of outputting data from the electronic device 700 to the user. The input device 760 can be a keyboard, mouse, microphone, joystick, or other device capable of inputting data into the electronic device 700.

[0114] It should be pointed out that, Figure 7 The structure shown does not constitute a limitation on the electronic device, except... Figure 7 In addition to the components shown, the electronic device may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0115] The slag block contour recognition device and application scenarios described in this application are for the purpose of more clearly illustrating the technical solutions of this application, and do not constitute a limitation on the technical solutions provided in this application. As those skilled in the art will know, with the evolution of slag block contour recognition devices and the emergence of new business scenarios, the technical solutions provided in this application are also applicable to similar technical problems.

[0116] This application provides a storage medium storing a program that, when executed by a processor, implements the slag block contour recognition method.

[0117] It is understood that, in order to achieve the functions in the above embodiments, the computer device includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, based on the units and method steps of the various examples described in conjunction with the embodiments disclosed in this application, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed by hardware or by computer software driving hardware depends on the specific application scenario and design constraints of the solution.

[0118] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0119] In a typical configuration, the device includes one or more processors (CPUs), memory, and a bus. The device may also include input / output interfaces, network interfaces, etc.

[0120] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM, and memory includes at least one memory chip. Memory is an example of computer-readable media.

[0121] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0122] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0123] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0124] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for recognizing the contour of slag blocks, characterized in that, The method is based on at least two cameras installed outside the slag well, and the method includes: Initial images of slag blocks on the grid plate at the bottom of the slag well are acquired using at least two cameras. Based on the brightness difference between the slag blocks and the low-temperature environment inside the slag well in the initial images, brightness gradient analysis is performed to preliminarily identify the outline of the slag blocks. The contours identified by the at least two cameras are fused to obtain the three-dimensional contour of the slag block; The at least two cameras continuously acquire images of the slag block, capturing the dynamic changes in the brightness of the slag block; Based on the correlation between the dynamic change in brightness and the contour, the contour of the slag block is corrected so that the hydraulic shut-off gate can adjust the extrusion depth according to the contour of the slag block, thereby achieving precise slag extrusion.

2. The method according to claim 1, characterized in that, Initial images of slag blocks on the grid plate at the bottom of the slag well are acquired using at least two cameras. Based on the brightness difference between the slag blocks and the low-temperature environment inside the slag well in the initial images, brightness gradient analysis is performed to preliminarily identify the outline of the slag blocks, including: The initial image is processed to extract brightness channel data. Based on the brightness channel data, the slag block and the low-temperature environment area in the slag well are distinguished and the suspected slag block area is marked. The suspected slag block area is the area marked by the pixels in the brightness channel data whose brightness value is greater than the first threshold, based on the brightness difference between the slag block and the low-temperature environment in the slag well. Brightness gradient calculations are performed on the suspected slag block area to filter out edge candidate points; Spatial continuity analysis is performed on the candidate edge points to form effective edge segments. All effective edge segments are then connected and closed to form the initial outline of the slag block.

3. The method according to claim 2, characterized in that, Brightness gradient calculations are performed on the suspected slag block area to filter out candidate edge points, including: Calculate the brightness difference between adjacent pixels in the suspected slag block area, and calculate the brightness gradient value of the pixels in the suspected slag block area based on the brightness difference using a gradient operator. The brightness gradient value is used to quantify the brightness change rate between the edge of the slag block and the low-temperature environment inside the slag well. Pixels with a brightness gradient value greater than a second threshold within the suspected slag block area are marked as edge candidate points, where the second threshold is greater than the first threshold.

4. The method according to claim 3, characterized in that, Continuously acquiring images of the slag block using at least two cameras, and capturing dynamic changes in the brightness of the slag block, including: Based on the initial outline of the slag block, a dynamic tracking window is set, and images are continuously acquired at fixed time intervals within the dynamic tracking window by at least two cameras; Extract the brightness channel data within the dynamic tracking window of each frame image, and calculate the average brightness value of the slag block region corresponding to the initial contour based on the brightness channel data; The brightness attenuation rate of the slag block region corresponding to the initial contour is calculated based on the average brightness value of consecutive frames.

5. The method according to claim 4, characterized in that, Based on the correlation between the dynamic change in brightness and the contour, the contour of the slag block is corrected, including: Based on the brightness attenuation rate, the brightness attenuation characteristics of the pixel region corresponding to the initial contour edge point are analyzed, and the brightness attenuation characteristics include the difference in brightness attenuation rate between the edge point and the inner region. When the difference between the brightness decay rate of the initial contour edge point and the brightness decay rate of the adjacent internal region exceeds the contour correction threshold, the edge point is determined to be a suspected misjudged edge point. The brightness gradient of the region where the suspected misjudged edge point is located is recalculated, and the edge candidate points are re-selected by combining the brightness channel data of consecutive frames within the dynamic tracking window. Based on the re-selected edge candidate points, suspected misjudged edge points are replaced or deleted, and the spatial continuity of the corrected edge segments is checked to form new valid edge segments. The initial contour of the slag block is updated based on the new effective edge segments, and the contours corrected by at least two cameras are re-fused to obtain the corrected three-dimensional contour.

6. The method according to claim 4, characterized in that, The method further includes: Monitor whether a new high-brightness area appears within the dynamic tracking window, excluding the slag block area corresponding to the initial contour. If a new high-brightness area appears, determine whether the brightness value of the new high-brightness area is greater than a preset multiple of the average brightness value of the low-temperature environment area. If it is greater than the initial contour, determine whether the new high-brightness area and the initial contour have spatial overlap. If spatial overlap exists, it indicates the accumulation of new slag blocks, and the new high-brightness area is included in the dynamic brightness monitoring range.

7. The method according to claim 6, characterized in that, The method further includes: Calculate the displacement amplitude and brightness range overlap of the boundary points of the slag block region corresponding to the initial contour based on the brightness channel data within the dynamic tracking window; When new slag blocks accumulate, the overlapping area of ​​the new high-brightness area with the initial contour, the expansion distance and expansion area ratio of the overall contour after the new high-brightness area is merged with the initial contour compared to the initial contour, are used as correction information. The contour of the slag block is corrected based on the displacement amplitude of the boundary point, the overlap of the brightness range, and the correction information.

8. The method according to claim 4, characterized in that, The method further includes: The motion trajectory of newly emerging independent high-brightness slag block regions, excluding the slag block region corresponding to the initial contour, is tracked within the dynamic tracking window in the initial image and continuous image frames acquired within the dynamic tracking window. The motion trajectory is determined based on the brightness channel data corresponding to the independent high-brightness slag block region. The independent high-brightness slag block region is a dynamically moving small slag block. The brightness value of the small slag block is greater than the average brightness value of the low-temperature environment region by a preset multiple and has no spatial overlap with the initial contour. When the motion trajectory enters the slag block area corresponding to the initial contour, and the brightness channel data of the independent high-brightness slag block area is interrupted, it indicates that the independent high-brightness slag block area is occluded by the slag block corresponding to the initial contour. Extract the boundary point coordinates, shape parameters, and area parameters of the occluded region. The boundary point coordinates are determined based on the brightness gradient change at the interruption position. The shape parameters include the curvature features and edge smoothness of the occluded region boundary. The area parameters include the total number of pixels contained in the occluded region and the actual area calculated based on the total number of pixels. The contour of the slag block is corrected based on the boundary point coordinates, the shape parameters, and the area parameters.

9. The method according to claim 4, characterized in that, The method further includes: Data on the insertion depth of the hydraulic shut-off valve in the slag well is collected, and the insertion depth data is used to reflect the degree of compression of the slag block by the hydraulic shut-off valve. Acquire the initial image before slag extrusion and the continuous image frames after slag extrusion within the dynamic tracking window. Based on the brightness channel data of the initial image before slag extrusion and the continuous image frames after slag extrusion, the boundary point coordinates of the slag block region corresponding to the initial contour before slag extrusion and the boundary point coordinates of the slag block region after slag extrusion are determined respectively. The displacement amplitude, contour shape change and area change of the boundary points before and after slag extrusion are determined based on the coordinates of the boundary points before and after slag extrusion. The contour of the slag block is corrected based on the extrusion depth data, the boundary point displacement amplitude, the contour shape change, the area change, and the brightness channel data.

10. The method according to claim 1, characterized in that, Before fusing the contours identified by the at least two cameras, the method further includes: Obtain the distance parameters between the at least two cameras and the slag well, as well as the boundary points of the three-dimensional contour in the images captured by each camera; Using the centerline of each camera as a reference, calculate the angle between each boundary point of the three-dimensional contour and the corresponding camera; The length, width, and height of the slag block are calculated based on the distance parameter and the included angle, respectively. The coordinate information of the slag block is determined based on the length, width, and height.

11. A slag block contour recognition device, characterized in that, The device includes: The identification module is used to acquire initial images of slag blocks on the grid plate at the bottom of the slag well through at least two cameras set outside the slag well, and to perform brightness gradient analysis based on the brightness difference between the slag blocks and the low-temperature environment inside the slag well in the initial image to initially identify the outline of the slag blocks. A fusion module is used to fuse the contours recognized by the at least two cameras to obtain the three-dimensional contour of the slag block. A capture module is used to continuously acquire images of the slag block through the at least two cameras and capture the dynamic changes in the brightness of the slag block; The correction module is used to correct the contour of the slag block based on the correlation between the dynamic change in brightness and the contour, so that the hydraulic shut-off gate can adjust the extrusion depth according to the contour of the slag block to achieve precise slag extrusion.

12. A storage medium, characterized in that, The storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the storage medium to perform the slag block contour recognition method as described in any one of claims 1-10.

13. An electronic device, characterized in that, The device includes at least one processor, at least one memory connected to the processor, and a bus; wherein the processor and the memory communicate with each other through the bus; the processor is used to call program instructions in the memory to execute the slag block contour recognition method as described in any one of claims 1-10.

Images