Method and apparatus for detecting and judging cracks in the refractory layer of molten iron ladle before it is subjected to iron treatment.

Abstract

This invention relates to the field of metallurgical equipment condition monitoring, and discloses a method and apparatus for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before it receives hot iron. The method includes: acquiring an image of the inner wall of the molten iron ladle; performing a first-stage processing on the image to quickly locate one or more suspected crack areas using a convolutional neural network; performing a second-stage processing on each suspected crack area by actively projecting an optical pattern and analyzing its geometric deformation to calculate the three-dimensional physical parameters of the crack, such as its physical width and depth; and comparing the calculated physical parameters with preset safety standards to automatically judge the molten iron ladle. This invention, through a two-stage processing strategy, combines the rapid recognition capability of deep learning with the precise quantification capability of active visual three-dimensional measurement, achieving accurate acquisition of the physical dimensions of cracks. This provides reliable data support for automated and objective ladle judgment decisions, significantly improving the accuracy and reliability of detection.

Description

Technical Field

[0001] This invention relates to the field of metallurgical equipment condition monitoring technology, specifically to a method and apparatus for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before it receives iron. Background Technology

[0002] In the steelmaking process, the ladle is a crucial container for holding and transferring molten iron at high temperatures. Its inner wall is lined with a refractory material layer, which inevitably suffers various forms of damage, such as cracking and spalling, under the cyclical effects of high temperatures, erosion, and chemical corrosion. If this damage is not detected and properly addressed in time, it can lead to refractory layer penetration, resulting in molten iron leakage and posing a significant production safety hazard. Therefore, inspecting the condition of the ladle's inner wall before each use is a vital safety measure.

[0003] Currently, some automated inspection solutions use industrial cameras to capture images of the inside of molten iron ladles and employ computer image processing technology to identify the presence of cracks on the inner wall. Compared to traditional manual visual inspection, these methods offer improvements in efficiency and safety. However, most of these methods remain at the level of two-dimensional image analysis, relying on the brightness, contrast, or texture features of the image to identify pixel areas suspected of containing cracks.

[0004] This type of technology, based on two-dimensional image analysis, inherently limits its ability to answer the questions of whether a crack "exists" and "is located," but it cannot provide quantitative information about the crack's geometric dimensions. In other words, it can detect the existence of a crack, but cannot directly measure its actual width or depth in the three-dimensional physical world. However, for the safety assessment of molten iron ladles, the physical dimensions of the crack are precisely the most critical indicator determining whether it can continue to serve safely. Due to the lack of objective and accurate physical dimension data, the final decision on whether to certify the ladle often still relies on human experience to reinterpret the two-dimensional image. This limits the objectivity, consistency, and automation of the entire inspection process, making it difficult to achieve fully reliable automated ladle certification. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and apparatus for detecting and judging cracks in the refractory layer of a molten iron ladle before it is fully sized and before it is used. This solves the problem that existing technologies rely heavily on two-dimensional image analysis for detecting cracks in the inner wall of molten iron ladles. While such methods can identify the existence of cracks, they cannot obtain physical size information of the cracks, such as their actual width or depth. Therefore, when making a decision on whether a molten iron ladle can continue to be used, there is a lack of objective quantitative basis, resulting in insufficient reliability of the judgment results.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The first aspect of this invention provides a method for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before it is subjected to iron treatment. The method includes the following steps:

[0008] Step 1: Obtain an image of the inner wall of the molten iron ladle;

[0009] Step 2: Perform the first stage of processing on the image, and use a convolutional neural network to quickly locate one or more suspected crack areas;

[0010] Step 3: Perform a second-stage processing on each of the suspected crack areas to obtain their physical parameters;

[0011] Step 4: Based on the physical parameters and preset safety standards, automatically judge the molten iron ladle.

[0012] In an optional embodiment of the present invention, the second stage processing includes: controlling a camera module integrating a controllable light source to project a preset optical pattern onto the suspected crack area; acquiring an image superimposed with the optical pattern; and calculating the physical parameters by analyzing the geometric deformation of the optical pattern in the image caused by the surface three-dimensional contour of the suspected crack area. This two-stage processing method first utilizes the efficiency of convolutional neural networks to complete a large-scale rapid screening and localization, and then performs targeted, active vision-based precise measurement only on local areas with potential risks. This ensures detection efficiency while quantifying the key physical dimensions of the crack, providing an objective data basis for subsequent automated packet judgment.

[0013] In an optional embodiment of the present invention, the physical parameters include at least one of the physical width and physical length of the crack.

[0014] In one optional embodiment of the present invention, the optical pattern is one or more parallel laser lines.

[0015] In an optional embodiment of the present invention, the step of calculating physical parameters by analyzing the geometric deformation of the optical pattern specifically includes: identifying a break or bend in the optical pattern in the image due to the physical depth of the suspected crack region; measuring the pixel distance at the break or bend feature; and converting the pixel distance into the physical parameter by combining it with a pre-calibrated pixel-to-physical size scale. For example, the physical width W of the crack. crack The following formula can be used for calculation:

[0016] W crack =d pixel ·S scale ;

[0017] In the formula, d pixel S is the measured pixel distance. scale The pixel-to-physical size scale is expressed in millimeters per pixel.

[0018] In an optional embodiment of the present invention, prior to the first stage of processing, a preprocessing step is included for the image of the inner wall of the molten iron ladle. The preprocessing step includes at least one of grayscale conversion and image noise reduction to reduce the computational load of subsequent convolutional neural network processing and suppress image noise interference.

[0019] In an optional embodiment of the present invention, the second stage processing further includes parallel calculation of the gradient magnitude image of the suspected crack region. The gradient magnitude image can highlight areas in the image with drastic grayscale changes, i.e., the edges of the crack, thereby assisting in more accurate identification and measurement of the geometric deformation of the optical pattern. Gradient magnitude M grad The calculation can be expressed by the following formula:

[0020]

[0021] In the formula, I x (x,y) and I y (x,y) represent the gradients of the image at pixel (x,y) along the horizontal and vertical directions, respectively.

[0022] In an optional embodiment of the present invention, the automated ladle judging step specifically involves: comparing the maximum crack width in the physical parameters with a preset safety standard threshold; if the maximum crack width is less than the safety standard threshold, the ladle is judged to be qualified; if the maximum crack width is greater than or equal to the safety standard threshold, the ladle is judged to be unqualified.

[0023] In an optional embodiment of the present invention, the method further includes: automatically issuing instructions to the production scheduling system based on the result of the automated ladle judgment, so as to perform corresponding production scheduling for the molten iron ladle.

[0024] A second aspect of the present invention provides a device for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before it is subjected to iron treatment, the device comprising:

[0025] An image acquisition unit is used to acquire images of the inner wall of a molten iron ladle.

[0026] A processing unit is electrically connected to the image acquisition unit, and the processing unit is configured to perform the following operations: perform a first-stage processing on the image to quickly locate one or more suspected crack regions using a convolutional neural network; perform a second-stage processing on each of the suspected crack regions to obtain its physical parameters; and automatically judge the molten iron ladle according to the physical parameters and a preset safety standard.

[0027] In the device of the present invention, the image acquisition unit is a camera module that integrates a controllable light source.

[0028] Furthermore, in the second stage of processing, the processing unit is also configured to: control the controllable light source to project a preset optical pattern onto the suspected crack area, and calculate the physical parameters by analyzing the geometric deformation of the optical pattern based on the image acquired by the image acquisition unit with the optical pattern superimposed on it.

[0029] This invention provides a method and apparatus for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before it is subjected to iron.

[0030] It has the following beneficial effects:

[0031] 1. This invention, through a first-stage and a second-stage processing, rapidly locates suspected crack areas using a convolutional neural network, and further obtains the actual physical parameters of the crack by projecting optical patterns and analyzing their geometric deformation. This method elevates the detection results from a qualitative "present or absent" judgment at the two-dimensional image level to a quantitative assessment of "width or length" in three-dimensional physical space, providing objective and repeatable data for automated bag identification, thereby avoiding the subjectivity and uncertainty caused by relying on human experience or simple two-dimensional image analysis.

[0032] 2. The two-stage collaborative processing mechanism employed in this invention first utilizes a high-speed convolutional neural network to perform a global and rapid screening of the inner wall of the molten iron ladle, quickly identifying potential problem areas. Then, it performs more precise but time-consuming targeted geometric quantization only on these small, suspected areas. This strategy avoids time-consuming three-dimensional scanning of the entire inner wall of the molten iron ladle, achieving optimized allocation of computing resources. Without sacrificing the measurement accuracy of key areas, it significantly shortens the cycle time of a single inspection, meeting the fast-paced demands of industrial production lines.

[0033] 3. This invention integrates crack location, quantification, and final ladle judgment into a complete automated process. By automatically comparing the quantified physical parameters with preset safety standards, the system can directly output a clear conclusion of "qualified" or "unqualified," and can further automatically issue instructions to the production scheduling system. This not only reduces the workload of manual verification and lowers the risk of misjudgment due to personnel fatigue or inconsistent standard implementation, but also makes the health status management of molten iron ladles more standardized and efficient. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the structure of the present invention;

[0035] Figure 2 This is a schematic diagram of the image preprocessing process of the present invention;

[0036] Figure 3 This is a flowchart of the molten iron ladle judging process of the present invention;

[0037] Figure 4 This is a schematic diagram illustrating the calculation of crack depth using the structured light triangulation principle of the present invention.

[0038] The components include: 1. Track; 2. Empty iron ladle; 3. Testing room; 4. Camera; 5. Computer; 6. Crane trolley; 7. Gantry rail; 8. Vehicle. Detailed Implementation

[0039] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] See attached document Figure 1 , Figure 1 This is a schematic diagram of a full-size device for detecting and judging cracks in the refractory layer of a molten iron ladle before it receives hot iron, according to an embodiment of the present invention. The device is deployed within a testing chamber 3 and is used to test an empty molten iron ladle 2. The empty molten iron ladle 2 is transported to a predetermined testing position within the testing chamber 3 via a carrier 8 on a track 1.

[0041] The device includes a portal track 7 spanning above the location of the empty molten iron ladle 2. A lifting trolley 6, which can move horizontally along the portal track 7, is mounted on the portal track 7. A multimodal sensor head 4 is suspended and fixed on the lifting trolley 6, and the multimodal sensor head 4 can move vertically via the lifting mechanism of the lifting trolley 6. Through the horizontal movement of the portal track 7 and the vertical movement of the lifting trolley 6, the multimodal sensor head 4 can enter the interior of the empty molten iron ladle 2 and reach any position on the inner wall to collect data.

[0042] The multimodal sensor head 4 integrates a camera and a controllable light source. The camera acquires images of the inner wall of the empty molten iron ladle 2, and the controllable light source projects a preset optical pattern onto the inner wall of the empty molten iron ladle 2 during specific processing stages. The multimodal sensor head 4 is electrically connected to a computer 5 via line 9. The computer 5 receives and processes the image data acquired by the multimodal sensor head 4 and executes the detection and ladle identification method of the present invention.

[0043] See attached document Figure 2 , Figure 2 This is a flowchart illustrating a method for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before iron loading, according to an embodiment of the present invention. The present invention provides a method for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before iron loading, which is executed by a computer 5 and may include the following steps:

[0044] Step S101: Acquire an image of the inner wall of the molten iron ladle. Specifically, after the empty molten iron ladle 2 is transported to the detection position, the computer 5 controls the gantry rail 7 and the lifting trolley 6 to drive the multimodal sensor head 4 to descend to a preset starting height inside the empty molten iron ladle 2. Subsequently, the multimodal sensor head 4 scans the entire inner wall of the empty molten iron ladle 2 along a predetermined trajectory (e.g., rotating and lifting), and its built-in camera continuously acquires images during this process, transmitting the image signals to the computer 5 in real time via line 9.

[0045] Step S102 involves performing a first-stage processing on the image, using a convolutional neural network to quickly locate one or more suspected crack regions. A processing program is deployed within computer 5, which includes a first-stage processing module a. This module a receives the image from step S101 and analyzes the image using a pre-trained convolutional neural network model to identify regions in the image with visual crack features, and outputs the location coordinates of these regions, forming a set of suspected crack regions.

[0046] Step S103 involves a second-stage processing of each suspected crack region to obtain its physical parameters. The computer 5 program also includes a second-stage processing module b. For each suspected crack region output in step S102, module b performs targeted processing. It first sends a command to the multimodal sensor head 4 via line 9, controlling its internal controllable light source to project a preset optical pattern onto the specific region. Simultaneously, the camera acquires an image superimposed with the optical pattern. Subsequently, module b analyzes the geometric deformation produced by the optical pattern on the image and calculates the physical parameters of the suspected crack region, such as its physical width.

[0047] Step S104: The molten iron ladle is automatically judged based on physical parameters and preset safety standards. The computer 5 program further includes an automated ladle judgment module c. This module c compares the physical parameters of all cracks calculated in step S103 with a safety standard stored locally or retrieved from a database. Based on the comparison result, module c generates a clear judgment conclusion of "qualified" or "unqualified".

[0048] After the ladle is judged, the method may include a follow-up step: based on the judgment result generated by the automated ladle judgment module c, computer 5 sends a corresponding scheduling instruction to the production scheduling system (not shown in the figure) via line 9. For example, if the judgment result is "qualified", the molten iron ladle is notified to be available for receiving iron; if the result is "unqualified", the scheduling room is notified to send the molten iron ladle to the maintenance station.

[0049] In one specific embodiment, after receiving the raw image acquired by the multimodal sensor head 4, the method first performs an image preprocessing step. The purpose of this step is to reduce the computational complexity of subsequent processing and suppress noise introduced during image acquisition. Preprocessing may include the following operations:

[0050] Step 1, Grayscale Conversion. Convert the multi-channel color RGB image to a single-channel grayscale image. gray This conversion is achieved by weighted summation of the intensity values ​​of each color channel, and the calculation process can be expressed as follows:

[0051] I gray (x,y)=W R ·C R (x,y)+W G ·C G (x,y)+W B ·C B (x,y);

[0052] In the formula, I gray (x,y) represents the output grayscale value at pixel coordinates (x,y); C R (x,y),CG (x,y),C B (x, y) represent the original intensity values ​​of the red, green, and blue channels at this pixel, respectively; W R′ W G′ W B These are preset, fixed weighting coefficients.

[0053] Step 2, Image Denoising. To reduce the impact of random noise in the image on subsequent crack identification, Gaussian filtering can be used to denoise the grayscale image I. gray After smoothing, the denoised image I is obtained. gauss This process is achieved by convolving the image with a Gaussian kernel. After preprocessing, the first-stage processing module a processes the denoised image I. gauss The processing proceeds. Module a contains a pre-trained convolutional neural network. Through a series of convolutional and pooling layers, this network extracts hierarchical features from the input image, ranging from shallow to deep, such as edge, texture, and shape combinations. By learning and analyzing these features, the network can identify regions in the image that match crack morphological characteristics. Finally, module a outputs the location information of one or more suspected crack regions, for example, providing the coordinate range of each suspected region in the image in the form of a bounding box.

[0054] Subsequently, the second-stage processing module b receives a set of suspected crack regions output from the first-stage processing module a, and performs targeted geometric quantization processing on each region in the set. (See attached diagram) Figure 3 , Figure 3 This is a schematic diagram illustrating the principle of a second-stage process according to an embodiment of the present invention. The process may include:

[0055] Step S201: Projecting an optical pattern. The second-stage processing module b sends a command to the multimodal sensing head 4 via line 9, controlling its internal controllable light source to project one or more laser lines with known geometric shapes only onto the currently processed suspected crack area, forming an optical pattern. In one embodiment, the optical pattern may be a set of parallel thin laser lines.

[0056] Step S202: Acquire deformation image. The built-in camera of the multimodal sensor head 4 acquires an image of the suspected crack area illuminated by an optical pattern. Since the crack itself is a three-dimensional structure with physical depth, when a straight laser line is projected onto its surface, it will appear as a discontinuous break or local bending in the two-dimensional acquired image. This phenomenon is called geometric deformation.

[0057] Step S203: Calculate physical parameters. The second-stage processing module b analyzes the image acquired in step S202. It first automatically identifies breakpoints or bending feature points on the optical pattern using image processing algorithms, and measures the pixel distance d of these feature points in the image coordinate system. pixel Subsequently, module b utilizes a pixel-to-physical size scale S pre-determined during the system calibration phase. scale The measured pixel distance is converted into the actual physical width W of the crack. crack Its calculation can be expressed by the following formula:

[0058] W crack =d pixel ·S scale ;

[0059] By measuring multiple locations on the optical pattern, the width of the crack at different positions can be obtained, and its maximum width can be determined. In an optional implementation, to aid in the accurate identification of geometric deformation, the second-stage processing module b can also calculate the gradient magnitude image M of the suspected crack region in parallel. grad Regions with high gradient magnitudes correspond to edges in the image where grayscale changes drastically, clearly outlining the contours of cracks. The gradient magnitude can be calculated as follows:

[0060]

[0061] In the formula, I x (x,y) and I y (x, y) represent the first derivatives of the original image at pixel (x, y) along the horizontal and vertical directions, respectively, which can be obtained by convolving the image with gradient operators such as the Sobel operator. After completing the second-stage processing of all suspected crack regions, the automated crack detection module c summarizes the physical parameters of all confirmed and quantized cracks. This module c takes the maximum value of the maximum physical width of all cracks and compares it with a configurable safety standard threshold T stored in computer 5 or retrieved from an external database. width Compare the values. If the maximum width value is less than the safety standard threshold T... width If the maximum width value is greater than or equal to the safety standard threshold T, then the package is judged as "qualified"; width If the result is not satisfactory, the ladle will be judged as "unqualified". This conclusion is the final automated judgment result for that ladle.

[0062] In one specific embodiment, after receiving the raw image acquired by the multimodal sensor head 4, the method first performs an image preprocessing step. The purpose of this step is to reduce the computational complexity of subsequent processing and suppress noise introduced during image acquisition. Preprocessing may include the following operations:

[0063] Step 1, Grayscale Conversion. Convert the multi-channel color RGB image to a single-channel grayscale image. gray This conversion is achieved by weighted summation of the intensity values ​​of each color channel, and the calculation process can be expressed as follows:

[0064] I gray (x,y)=R(x,y)·I R +G(x,y)·I G +B(x,y)·I B ;

[0065] In the formula, I gray (x,y) represents the output grayscale value at pixel coordinates (x,y); R(x,y), G(x,y), and B(x,y) are the red, green, and blue color levels of this pixel, respectively, and their values ​​are integers from 0 to 255; I R I G I B These are the grayscale coefficients for red, green, and blue color levels, respectively. In one embodiment, I... R The value is 0.31, I G The value is 0.60, I B The value is set to 0.11. Step two, image denoising. To reduce the impact of random noise in the image on subsequent crack identification, Gaussian filtering can be used on the grayscale image I. gray After smoothing, the denoised image I is obtained. gauss This process is achieved by convolving the image with a Gaussian kernel, and its mathematical model is as follows:

[0066]

[0067] In the formula, G kernel (i,j) represents the value of the Gaussian kernel function at coordinates (i,j), and k is the size of the Gaussian kernel. After preprocessing, the first-stage processing module a processes the denoised image I. gauss The processing proceeds. Module a contains a pre-trained convolutional neural network. This network extracts hierarchical features from the input image through a series of convolutional and pooling layers to identify regions in the image that match crack morphology. Finally, module a outputs the location information of one or more suspected crack regions.

[0068] Subsequently, the second-stage processing module b receives a set of suspected crack regions output from the first-stage processing module a, and performs targeted geometric quantization processing on each region in the set. (See attached diagram) Figure 3 and attached Figure 4 The process may include:

[0069] Step S201: Project an optical pattern and acquire a deformation image. The second-stage processing module b controls the controllable light source inside the multimodal sensor head 4 to project a set of parallel laser lines onto the suspected crack area, and the camera acquires an image superimposed with the optical pattern. Due to the physical depth and width of the crack, the straight laser lines will exhibit broken or bent geometric deformations on the two-dimensional image.

[0070] Step S202: Calculate physical parameters. The second-stage processing module b analyzes the acquired deformation images to quantify the physical size of the cracks.

[0071] For the physical width W of the crack crack Module b identifies the break in the laser line between the two edges of the crack and measures its lateral pixel distance d in the image coordinate system. pixel And using a pre-calibrated scale S scale Perform the conversion:

[0072] W crack =d pixel ·S scale ;

[0073] For the physical depth h of the crack, refer to Figure 4 The triangulation principle shown is calculated as follows. Module b identifies the pixel displacement d in the image caused by the laser line's depth change. shift And calculate based on the known system geometric parameters:

[0074]

[0075] In the formula, d shift S represents the observed pixel displacement. scale α is the pixel-to-physical size scale; α is the laser projection angle; β is the camera viewing angle.

[0076] In an optional implementation, to aid in the accurate identification of geometric deformation, the second-stage processing module b can also calculate the gradient magnitude image of the suspected crack region in parallel. This calculation first obtains the gradient I of the image in the x and y directions. x and I y The calculation method is to convert image I (e.g., I...) into a single image. gauss ) and gradient operators (such as the Sobel operator) G x and G y Perform convolution:

[0077]

[0078] Subsequently, the total gradient magnitude M at pixel (x,y) is calculated based on the obtained directional gradient. grad (x,y) to enhance crack boundaries:

[0079]

[0080] After completing the second-stage processing of all suspected crack areas, the automated ladle evaluation module c summarizes the physical parameters of all quantified cracks. This module c compares the maximum physical width and maximum physical depth of all cracks with a preset safety standard threshold. If all quantified parameters are less than the corresponding safety standard threshold, the ladle is evaluated as "qualified"; if any parameter is greater than or equal to its corresponding safety standard threshold, the ladle is evaluated as "unqualified". This conclusion is the final automated ladle evaluation result for the molten iron ladle.

[0081] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for detecting and judging cracks in a full-size ladle refractory lining before iron feeding, characterized in that, include: Step 1: Obtain an image of the inner wall of the molten iron ladle; Step 2: Perform the first stage of image processing, using a convolutional neural network to quickly locate one or more suspected crack areas; Step 3: Perform a second-stage treatment on each suspected crack area to obtain its physical parameters; The second stage of processing includes: The camera module, which integrates a controllable light source, is controlled to project a preset optical pattern onto the suspected crack area; Acquire an image with the optical pattern superimposed on it; The physical parameters are calculated by analyzing the geometric deformation of the optical pattern in the suspected crack region. The optical pattern is one or more parallel laser lines; The step of calculating the physical parameters by analyzing the geometric deformation produced by the optical pattern in the suspected crack region specifically includes: Identify the breaks or bends in the optical pattern caused by the physical depth of the suspected crack region; For the physical depth of the crack The calculation method is as follows: Identify the pixel displacement on the image caused by the laser line due to depth changes. And calculate based on the known system geometric parameters: ； In the formula, The observed pixel displacement; Pixel-to-physical size scale; The laser projection angle; This is the camera's viewing angle; Measure the pixel distance at the break or bend; By combining a pre-calibrated pixel-physical size scale, the pixel distance is converted into the physical parameters; Step 4: Based on physical parameters and preset safety standards, automatically judge the molten iron ladle.

2. The method for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before iron loading, as described in claim 1, is characterized in that... The physical parameters also include at least one of the physical width and physical length of the crack.

3. The method for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before iron loading, as described in claim 1, is characterized in that... Before the first stage of processing, the process also includes a step of preprocessing the image of the inner wall of the molten iron ladle; The preprocessing steps include at least one of grayscale conversion and image noise reduction.

4. The method for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before iron loading, as described in claim 1, is characterized in that... The second stage of processing also includes parallel calculation of the gradient magnitude image of the suspected crack region to enhance the crack boundary and assist in the calculation of the physical parameters.

5. The method for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before iron loading, as described in claim 1, is characterized in that... The automated packet identification steps are as follows: The maximum crack width in the physical parameters is compared with a preset safety standard threshold. If the maximum crack width is less than the safety standard threshold, the molten iron ladle is deemed qualified. If the maximum crack width is greater than or equal to the safety standard threshold, the molten iron ladle is deemed unqualified.

6. The method for detecting and judging cracks in the refractory layer of a full-size molten iron ladle before iron loading, as described in claim 1, is characterized in that... The method further includes: Based on the results of the automated ladle identification, instructions are automatically sent to the production scheduling system to schedule the production of the molten iron ladle accordingly.

7. A device for detecting and judging cracks in the refractory layer of a full-size ladle before iron loading, used to perform the method for detecting and judging cracks in the refractory layer of a full-size ladle before iron loading as described in any one of claims 1-6, characterized in that, include: The image acquisition unit is used to acquire images of the inner wall of the molten iron ladle; The processing unit is electrically connected to the image acquisition unit, and the processing unit is configured to: The image is processed in the first stage, and one or more suspected crack areas are quickly located using a convolutional neural network; A second-stage processing is performed on each of the suspected crack regions to obtain its physical parameters; The molten iron ladle is automatically judged based on the physical parameters and preset safety standards. The image acquisition unit is a camera module that integrates a controllable light source; Furthermore, in the second stage of processing, the processing unit is also configured to control the controllable light source to project a preset optical pattern onto the suspected crack area, and to calculate the physical parameters by analyzing the geometric deformation of the optical pattern based on the acquired image superimposed with the optical pattern.

Images