Automatic dismounting and mounting method and system for argon pipe opening in steel ladle hot repair

Abstract

The invention discloses an automatic disassembly and assembly method for an argon pipe opening in steel ladle hot repair. Comprising the following steps: acquiring a real-time image of an argon pipe orifice and extracting an interface edge area in the real-time image; extracting a strong edge and a weak edge in the interface edge region through edge detection, and connecting the strong edge and the weak edge to form a target edge region; extracting a plurality of feature regions in the target edge region, and performing pixel matching on the plurality of feature regions and a corresponding preset standard template image to calculate a contrast value; performing threshold judgment on the similarity values obtained by matching, then sorting the similarity values, and taking a feature region position corresponding to the highest similarity value meeting the threshold judgment as an optimal matching position; according to the optimal matching position, selectively adjusting the real-time image through at least one of multi-scale template matching or template matching based on rotation invariant, so as to determine the positioning information of the argon pipe base; and according to the positioning information, the robot arm is controlled to clamp the argon pipe to be automatically installed to the argon pipe opening or detach the argon pipe from the argon pipe opening.

Description

Technical Field

[0001] The invention relates to the field of ladle automation equipment, and in particular to an automatic disassembly and assembly method and system for an argon gas pipe opening in hot repair of a ladle. Background Art

[0002] During the hot repair process of the ladle, accurate identification and installation detection of the argon pipe orifice are the key to ensuring operational safety and efficiency. However, existing technologies face multiple challenges in practical applications. The high-temperature operating environment places strict requirements on the high-temperature resistance of the equipment. At the same time, interference factors such as thermal radiation and light changes lead to a decrease in the stability of the visual recognition algorithm. In addition, the dependence of deep learning algorithms on high-quality data, the complexity of the deployment of automation systems, the need for real-time response, and the impact of changes in operating conditions on the adaptability of the algorithm have all become bottlenecks restricting the development of the technology. At the same time, insufficient human-computer interaction still requires manual intervention in some complex situations, reducing the level of automation. Therefore, the development of an algorithm that can overcome the above problems and achieve efficient, stable, and adaptive argon pipe orifice identification and installation detection has become a technical problem that needs to be solved urgently.

[0003] Based on this, this application is filed. Summary of the invention

[0004] In view of the above-mentioned technical problems, the purpose of the present application is to provide a method for automatically disassembling and assembling an argon pipe nozzle during hot repair of a ladle, aiming to solve the problem of automatic disassembly and assembly of the argon pipe nozzle.

[0005] In a first aspect of the present application, a method for automatically disassembling and assembling an argon pipe mouth in hot repair of a ladle is provided, and the automatic disassembly and assembly method comprises: acquiring a real-time image of the argon pipe mouth and extracting an interface edge area in the real-time image; extracting strong edges and weak edges in the interface edge area by edge detection, and connecting the strong edges and weak edges to form a target edge area; extracting a plurality of feature areas in the target edge area, and performing pixel matching on the plurality of feature areas and corresponding preset standard template images to calculate a contrast value; performing threshold judgment on the similarity values ​​obtained by matching and then sorting them, and taking the feature area position corresponding to the highest similarity value that meets the threshold judgment as the best matching position; selectively adjusting the real-time image by at least one of multi-scale template matching or rotation-invariant-based template matching according to the best matching position to determine the positioning information of the argon pipe base; and controlling a robotic arm to clamp the argon pipe for automatic installation to the argon pipe mouth, or to remove the argon pipe from the argon pipe mouth according to the positioning information.

[0006] In an embodiment of the present application, the strong edges and weak edges in the interface edge area are extracted by edge detection, and the strong edges and weak edges are connected to form a target edge area, including: convolving the pixels of the target edge area with a two-dimensional Gaussian kernel to obtain a processed edge image; calculating the horizontal gradient and vertical gradient of the processed edge image to obtain the gradient amplitude and direction of each pixel; obtaining the local maximum of the processed edge image by non-maximum suppression; extracting the local maximum values ​​greater than a preset first intensity threshold as strong edges, and extracting the local maximum values ​​less than a preset second intensity threshold as weak edges, and connecting the strong edges and weak edges to form a target edge area.

[0007] In an embodiment of the present application, the pixel points of the target edge area are convolved with a two-dimensional Gaussian kernel to obtain a processed edge image by the following formula:

[0008]

[0009] Among them, G(x, y) is the data representation of the edge image after processing, σ is the standard deviation, and x and y are the coordinate points of each pixel.

[0010] In the embodiment of the present application, extracting a plurality of feature regions from the target edge region, and performing pixel matching between the plurality of feature regions and a corresponding preset standard template image to calculate a contrast value includes:

[0011] The contrast value between the feature area and the standard template image is obtained using the following formula:

[0012]

[0013] Among them, R(x, y) is the calculated contrast value, T(i, j) is the pixel value of the corresponding standard template, I(x+i, y+j) is the local pixel value of the feature area at the position (x, y), and are the average values ​​of the standard template and the feature area, respectively.

[0014] In an embodiment of the present application, the similarity values ​​obtained by matching are sorted after threshold judgment, and the feature area position corresponding to the highest similarity value that satisfies the threshold judgment is taken as the best matching position, including: performing threshold judgment on the similarity values, screening out valid similarity values ​​that exceed a preset first similarity threshold; sorting the valid similarity values, obtaining the highest similarity value therein; obtaining the feature area position corresponding to the highest similarity value as the best matching position.

[0015] In an embodiment of the present application, the real-time image can be selectively adjusted according to the best matching position through at least one of multi-scale template matching or template matching based on rotation invariance to determine the positioning information of the argon tube base, including: selecting at least three feature points in the feature area of ​​the best matching position, the feature points at least including the center point of the argon tube base, and using a feature point detection algorithm to extract corresponding points in the standard template image corresponding to the feature points; determining whether there is a rotational change in the argon tube base according to the center point and the corresponding points, and determining whether the real-time image is scaled according to the circumscribed graphics of the target edge area; when either the argon tube base is rotated or the real-time image is scaled, calculating an affine transformation matrix according to the detected feature points and the corresponding points, and using the affine transformation matrix to transform the real-time image to align it with the standard template image, and outputting the positioning information in the aligned real-time image.

[0016] In an embodiment of the present application, the step of calculating an affine transformation matrix according to the detected feature points and the corresponding points includes:

[0017] The affine transformation matrix is ​​calculated using the following formula:

[0018]

[0019] Where M is a 2x3 affine transformation matrix, including rotation, translation, and scaling.

[0020] In an embodiment of the present application, the use of a feature point detection algorithm to extract corresponding points in the standard template image corresponding to the feature points includes: generating a SIFT feature descriptor for each detected feature point; using a matching algorithm to preliminarily match the feature descriptor in the real-time image with the feature descriptor in the standard template image, and returning the preliminarily screened corresponding points; obtaining preliminary screening corresponding points from the preliminary screened corresponding points through a threshold judgment of a ratio test; and using a RANSAC algorithm to eliminate mismatched point pairs from the preliminary screening corresponding points, thereby outputting the final corresponding points.

[0021] In an embodiment of the present application, controlling the robotic arm to clamp the argon tube for automatic installation to the argon tube port, or removing the argon tube from the argon tube port according to the positioning information includes: obtaining a first current position of the robotic arm actuator and a second current position of the argon tube; when receiving an installation signal, determining a first motion path according to the first current position, the second current position and the positioning information, so as to control the robotic arm to clamp the argon tube and automatically install it to the argon tube port according to the first motion path; when receiving a removal signal, determining a second motion path according to the first current position, the second current position and the positioning information, so as to control the robotic arm to remove the argon tube from the argon tube port according to the second motion path.

[0022] In an embodiment of the present application, determining whether there is a rotational change in the argon tube base according to the center point and the corresponding point, and determining whether there is a scaling in the real-time image according to the circumscribed graphics of the target edge area, includes: fitting a circle using the least squares method to estimate the rotation angle between the feature point and the corresponding point; when the rotation angle is greater than a preset rotation threshold, determining that there is a rotational change in the base; and obtaining the size of the circumscribed rectangle of the target edge area, and comparing the size of the circumscribed rectangle with the preset edge size corresponding to the standard template image to calculate the scaling ratio; if the scaling ratio exceeds the preset scaling threshold range, determining that there is a scaling change in the base.

[0023] In the second aspect of the present application, there is also provided an automatic disassembly and assembly system for the argon pipe mouth in the hot repair of a ladle, comprising: a robotic arm; an end effector for clamping or releasing the argon pipe; an image acquisition device for acquiring a real-time image of the argon pipe mouth; an argon pipe storage rack for placing the argon pipe; and a controller electrically connected to the robotic arm, the end effector and the image acquisition device, and configured to execute the automatic disassembly and assembly method as described in any one of the above items.

[0024] In summary, the automatic disassembly and assembly method provided by the embodiment of the present invention includes at least the following technical effects:

[0025] Through high-precision edge detection and feature extraction technology, the interface edge and feature area of ​​the argon tube mouth can be accurately identified, providing an accurate basis for subsequent matching; pixel matching technology can ensure the accuracy of the disassembly and assembly position, and multi-scale and rotation-invariant template matching further improves the flexibility and fault tolerance of matching, improves the robustness of matching, and makes the disassembly and assembly process more efficient and stable; and the combination of strong edges and weak edges improves the integrity of the edges by reducing errors caused by environmental changes (such as light and pollution).

[0026] The entire disassembly and assembly process is automated, reducing manual intervention, lowering operational difficulty and error rates; by real-time monitoring of the disassembly and assembly process of the argon gas pipe, abnormal situations can be discovered and handled in a timely manner to ensure the smooth progress of the production process; the automated disassembly and assembly method reduces the risk of errors caused by human operation, such as plugging in the wrong position, damaging the interface, etc., and improves operational safety.

[0027] Other features and advantages of the embodiments of the present invention will be described in the following specific implementation examples. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0029] Figure 1 A flow chart of the general concept of the automatic disassembly and assembly method provided in the embodiment of the present application;

[0030] Figure 2 The step node diagram of the automatic disassembly and assembly method of the present application corresponding to its process is shown;

[0031] Figure 3 A sub-flow chart of step S200 in the automatic disassembly and assembly method provided in one embodiment of the present application;

[0032] Figure 4 This is a sub-flowchart of step S400 in the automatic disassembly and assembly method provided in one embodiment of the present application.

[0033] Figure 5 A sub-flow chart of step S500 in the automatic disassembly and assembly method provided in one embodiment of the present application;

[0034] Figure 6 A sub-flow chart of step S501 in the automatic disassembly and assembly method provided in one embodiment of the present application;

[0035] Figure 7 A sub-flow chart of step S600 in the automatic disassembly and assembly method provided in one embodiment of the present application;

[0036] Figure 8 This is a module diagram of an automatic disassembly and assembly system provided in one embodiment of the present application. DETAILED DESCRIPTION

[0037] The terms "second direction", "first direction", "third direction", "inside", "outside" and the like that appear below to indicate orientation or positional relationships, unless otherwise specified, are to be understood as being based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0038] In addition, if there is a feature defined as "first" or "second", it is only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Features defined as "first" or "second" may explicitly or implicitly include at least one of the defined features. If the description of "plurality" appears, the general meaning is to include at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0039] In this application, unless otherwise clearly specified and limited, the terms such as "installed", "connected", "connected", "fixed" and so on should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integrated connection; it can be a mechanical connection or an electrical connection, it can be a direct connection, or it can be an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For those skilled in the art, the specific meanings of the above terms in this application can be understood according to the specific circumstances.

[0040] In the description of this specification, if the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" appear, it means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0041] Reference Figure 1 and Figure 2 The present application first provides a method for automatically disassembling and assembling an argon pipe orifice during hot repair of a ladle, comprising at least the following steps:

[0042] Step S100, obtaining a real-time image of the argon gas pipe opening and extracting an interface edge area in the real-time image;

[0043] Step S200, extracting strong edges and weak edges in the interface edge region by edge detection, and connecting the strong edges and weak edges to form a target edge region;

[0044] Step S300, extracting a plurality of feature regions in the target edge region, and performing pixel matching between the plurality of feature regions and a corresponding preset standard template image to calculate a contrast value;

[0045] Step S400, sorting the similarity values ​​obtained by matching after threshold judgment, and taking the feature region position corresponding to the highest similarity value that meets the threshold judgment as the best matching position;

[0046] Step S500, adjusting the real-time image according to the best matching position and optionally by at least one of multi-scale template matching or template matching based on rotation invariance, so as to determine the positioning information of the argon tube base;

[0047] Step S600: Control the robot arm to clamp the argon gas tube according to the positioning information and automatically install the argon gas tube to the argon gas tube port, or remove the argon gas tube from the argon gas tube port.

[0048] Among the terms appearing in the general concept of this application, "argon pipe port" refers to the pipe interface on the ladle for connecting and transporting argon; argon is an inert gas, which is commonly used to protect molten steel from oxidation during steel production; "real-time image" refers to the image of the current scene captured in real time by a camera or other image acquisition device, which can be referred to as Figure 2 ; Pixel matching refers to comparing the feature area in the real-time image with the guaranteed template image pixel by pixel to evaluate the similarity between them, such as square difference matching, normalized square difference matching, etc.; In the present application, the target edge area is formed by connecting the strong edge and the weak edge to represent the edge contour of the argon tube orifice. The target edge area is the basis for subsequent feature extraction and matching; and the "standard template image" is a pre-made image with specific features, which is used to match the feature area in the real-time image, which can usually be understood as the ideal state or known state of the argon tube base.

[0049] Based on the general concept of the present application, a high-resolution camera is first used to capture the current state image of the argon pipe opening, and the boundary between the argon pipe opening and the surrounding environment, i.e., the interface edge area, is extracted through an image processing algorithm (such as Canny edge detection); within the interface edge area, clear (strong edges) and blurred (weak edges) edges are further distinguished, and these edges are connected to form a complete and continuous target edge area to prepare for subsequent feature extraction; within the target edge area, an area with significant features (such as the bottom, corner points, edge intersections, etc.) is selected as a feature area, and the feature area is matched with a preset standard template image at the pixel level, and the contrast value (or similarity value) between the two is calculated to evaluate the accuracy of the match.

[0050] Then, all the similarity values ​​obtained by matching are threshold-screened, and only the values ​​above the threshold are retained, and these values ​​are sorted. The feature area position corresponding to the highest similarity value selected in this application is used as the best matching position, which reflects the precise position of the argon tube base in the real-time image; according to the best matching position, the real-time image may be further adjusted selectively, such as using multi-scale template matching to adapt to argon tube bases of different sizes, or using template matching based on rotation invariance to process the rotation change of the base, so as to more accurately determine the positioning information of the argon tube base; finally, according to the positioning information, the robot arm is controlled to accurately clamp the argon tube and install it to the argon tube port or remove it from it. This step realizes the automated disassembly and assembly operation of the argon tube port.

[0051] Based on the embodiments of the present invention, through high-precision edge detection and feature extraction technology, the interface edge and feature area of ​​the argon tube mouth can be accurately identified, providing an accurate basis for subsequent matching; pixel matching technology can ensure the accuracy of the disassembly and assembly position, and multi-scale and rotation-invariant template matching further improves the flexibility and fault tolerance of matching, making the disassembly and assembly process more efficient and stable.

[0052] In the present application, the entire disassembly and assembly process is automated, reducing manual intervention, lowering operational difficulty and error rates; by real-time monitoring of the disassembly and assembly process of the argon gas pipe nozzle, abnormal situations can be discovered and handled in a timely manner to ensure the smooth progress of the production process; the automated disassembly and assembly method reduces the risk of errors caused by human operation, such as plugging in the wrong position, damaging the interface, etc., thereby improving operational safety.

[0053] Based on the above general concept, a specific embodiment is provided below:

[0054] In an optional embodiment of step S200, strong edges and weak edges in the interface edge region are extracted by edge detection, and the strong edges and weak edges are connected to form a target edge region, including:

[0055] Step S201, convolving the pixel points in the target edge area with the two-dimensional Gaussian kernel to obtain a processed edge image;

[0056] Specifically, convolve each pixel in the target edge region with a two-dimensional Gaussian kernel; the size and standard deviation of the Gaussian kernel can be adjusted according to the severity of the image noise; the convolved image will have smooth edges and reduced noise. Convolve each pixel in the target edge region with the Gaussian kernel. For each pixel in the target edge region, multiply the pixel values ​​around it (according to the size of the Gaussian kernel) by the corresponding weight in the Gaussian kernel, and add the results to get the new value of the pixel. This process is performed on the entire target edge region, resulting in a smoothed edge image.

[0057] Step S202, calculating the horizontal gradient and vertical gradient of the processed edge image to obtain the gradient amplitude and direction of each pixel;

[0058] The horizontal gradient and vertical gradient of the processed edge image are calculated, and the Sobel operator, Prewitt operator or other gradient operators can be used to implement it. It can be understood that the gradient amplitude represents the strength of the edge, and the gradient direction represents the direction of the edge, the purpose of which is to identify the edge position in the image.

[0059] Step S203, obtaining the local maximum value of the processed edge image by non-maximum suppression;

[0060] For each pixel, compare its gradient magnitude with that of adjacent pixels along its gradient direction. If the gradient magnitude of the current pixel is not the local maximum, suppress it (i.e., set it to 0 or a smaller value); that is, only the local maximum in the gradient direction is retained to form a refined edge and remove redundant pixels. In a specific scheme, if the gradient magnitude of the current pixel is the local maximum (i.e., greater than or equal to the gradient magnitude of the adjacent pixels), retain the pixel; otherwise, suppress it (i.e., set it to 0 or a smaller value).

[0061] Step S204: extract the local maximum values ​​greater than a preset first intensity threshold as strong edges, extract the local maximum values ​​less than a preset second intensity threshold as weak edges, and connect the strong edges and the weak edges to form a target edge region.

[0062] According to the preset first intensity threshold, the edge pixels after non-maximum suppression are divided into strong edges and weak edges; strong edges are those pixels whose gradient amplitude is greater than the threshold, while weak edges are those pixels whose gradient amplitude is less than the threshold; then, appropriate algorithms (such as morphological operations, contour tracking, etc.) are used to connect strong edges and weak edges to form a continuous and complete target edge area.

[0063] Based on step S201 to step S203, in a specific embodiment provided in the present application, two thresholds T are set by using double threshold processing (Double Thresholding) low and T high Points with gradient intensity greater than the high threshold are considered "strong edges", points less than the low threshold are ignored, and points in between are considered "weak edges", thereby eliminating false edges caused by noise, retaining important edges, and finally connecting edges, taking strong edges as the basis of edges, starting from these strong edges, if a weak edge is connected to a strong edge, it is considered an edge, otherwise it is suppressed. This step ensures the coherence of the edge, thereby achieving denoising of the picture with the argon gas tube mouth.

[0064] Since Gaussian kernel convolution and gradient calculation are robust to illumination changes, the algorithm can maintain a stable edge detection effect under different illumination conditions; the horizontal and vertical gradients of the edge image after calculation can accurately capture the changes in pixel values ​​in the image, thereby accurately identifying the edge position, and the gradient amplitude and direction provide edge strength and direction information, which is helpful for subsequent edge classification and connection; through non-maximum suppression, redundant pixels on the edge can be removed, and only the local maximum value in the gradient direction is retained, thereby obtaining a more refined edge and reducing the width of the edge, making it clearer and more accurate; according to the preset intensity threshold, the edges are divided into strong edges and weak edges, which helps to handle different types of edges more flexibly in subsequent processing; when connecting strong edges and weak edges, the weak edge information can be fully utilized to fill the gaps between strong edges, thereby obtaining a more complete and continuous edge contour.

[0065] In a specific scheme, the pixel points in the target edge area are convolved with a two-dimensional Gaussian kernel to obtain a processed edge image by the following formula:

[0066]

[0067] Among them, G(x, y) is the data representation of the edge image after processing, σ is the standard deviation, and x and y are the coordinate points of each pixel.

[0068] In a feasible embodiment of step S300, a plurality of feature regions are extracted from the target edge region, and pixel matching is performed between the plurality of feature regions and corresponding preset standard template images to calculate contrast values, including:

[0069] The following formula is used to obtain the contrast value between the feature area and the standard template image:

[0070]

[0071] Among them, R(x, y) is the calculated contrast value, T(i, j) is the pixel value of the corresponding standard template, I(x+i, y+j) is the local pixel value of the feature area at the position (x, y), and are the average values ​​of the standard template and the feature area, respectively.

[0072] This method can more accurately reflect the contrast between the feature area and the standard template image by local pixel matching, that is, calculating the difference in pixel values ​​at corresponding positions. The local matching method avoids the errors that may be caused by global contrast and improves the accuracy of contrast calculation. In the contrast calculation formula, the average values ​​of the standard template and the feature area are introduced for adjustment. This helps to eliminate the contrast calculation error caused by the overall brightness difference and make the result more accurate and reliable. Since this method uses local pixel matching and average value adjustment, it has a certain robustness to changes in lighting conditions. Therefore, under different lighting conditions, it can still maintain a high detection accuracy; it is suitable for different types of argon gas pipe opening environments.

[0073] In an embodiment of step S400, the similarity values ​​obtained by matching are sorted after threshold judgment, and the feature region position corresponding to the highest similarity value satisfying the threshold judgment is taken as the best matching position, including:

[0074] Step S401, performing threshold determination on the similarity values, and screening out valid similarity values ​​exceeding a preset first similarity threshold;

[0075] In the present application, a first similarity threshold is preset, and the matching result is considered valid only when the similarity value exceeds this threshold; the threshold can be adjusted according to the needs of the actual application scenario.

[0076] Step S402: sort the valid similarity values ​​to obtain the highest similarity value;

[0077] Find the highest similarity value among the valid similarity values. This represents the location of the feature area that best matches the standard template image; sort the valid similarity values, usually from high to low. The highest similarity value after sorting is what you want.

[0078] Step S403: Obtain the feature region position corresponding to the highest similarity value as the best matching position.

[0079] The best matching position is determined according to the highest similarity value. This position represents the precise position of the argon tube orifice in the image and can be used for subsequent image processing, object recognition or positioning tasks. The feature area position corresponding to the highest similarity value after sorting is directly obtained, which is the best matching position.

[0080] Through threshold judgment, low-quality matching results can be removed to reduce the possibility of mismatching. At the same time, the sorting operation ensures that only the highest similarity value is selected, further improving the accuracy of matching; once the highest similarity value is determined, the best matching position can be directly obtained without additional calculation or processing.

[0081] In an embodiment of step S500, the real-time image may be selectively adjusted by at least one of multi-scale template matching or rotation-invariant template matching according to the best matching position to determine the positioning information of the argon tube base, including:

[0082] Step S501, selecting at least three feature points in the feature area of ​​the best matching position, the feature points at least including the center point of the argon tube base, and using a feature point detection algorithm to extract corresponding points in the standard template image corresponding to the feature points;

[0083] When selecting feature points, the center point of the argon tube base should be given priority, because the center point is crucial to determine the position and orientation of the base. At the same time, two or more other feature points, such as edge points and corner points of the base, can also be selected to ensure the accuracy and stability of the transformation; feature point detection algorithms, such as SIFT, SURF or ORB, can be used to extract these feature points and their corresponding points.

[0084] Step S502, determining whether the argon tube base has a rotation change according to the center point and the corresponding point, and determining whether the real-time image has a zoom according to the external graphics of the target edge area;

[0085] According to the positional relationship between the center point and the corresponding point, it is determined whether the argon tube base has a rotation change; at the same time, according to the size change of the circumscribed graphics (such as rectangles, ellipses, etc.) in the target edge area, it is determined whether the real-time image has been scaled.

[0086] Step S503: When the argon tube base is rotated or the real-time image is scaled, an affine transformation matrix is ​​calculated based on the detected feature points and corresponding points, and the real-time image is transformed using the affine transformation matrix to align it with the standard template image, and the positioning information in the aligned real-time image is output.

[0087] When rotation or scaling is detected, the affine transformation matrix is ​​calculated and used to transform the real-time image to align it with the standard template image. The positioning information of the argon tube base can be accurately extracted from the aligned image; the affine transformation matrix can be obtained by solving a set of linear equations, which contains transformation information such as rotation, scaling and translation. After transforming the real-time image using the matrix, an image aligned with the standard template image can be obtained, and finally the positioning information of the argon tube base in the aligned image can be extracted through image processing techniques (such as edge detection, contour extraction, etc.).

[0088] In a specific embodiment, calculating the affine transformation matrix according to the detected feature points and corresponding points includes:

[0089] The affine transformation matrix is ​​calculated using the following formula:

[0090]

[0091] Where M is a 2x3 affine transformation matrix, including rotation, translation, and scaling.

[0092] In step S501, a feature point detection algorithm is used to extract corresponding points corresponding to the feature points in the standard template image, including:

[0093] Step S5011: Generate a SIFT feature descriptor for each detected feature point;

[0094] First, locate the feature points in the real-time image, which usually involves detecting corners, edge points or areas of significant changes in the image. Then, generate 128-dimensional (or other dimensions, depending on the specific implementation) SIFT feature descriptors for these feature points, which reflect the local gradient information around the feature points.

[0095] Step S5012: Use a matching algorithm to preliminarily match the feature descriptors in the real-time image with the feature descriptors in the standard template image, and return the preliminarily screened corresponding points;

[0096] Using a descriptor matching algorithm (such as brute force matching, FLANN matching, etc.), the similarity (such as Euclidean distance, Hamming distance, etc.) between the descriptors in the real-time image and the standard template image is calculated. Subsequently, the corresponding points of the preliminary match are screened out according to the similarity threshold.

[0097] Step S5013, obtaining the initial screening corresponding points from the initial screening corresponding points through the threshold judgment of the ratio test;

[0098] For each initially matched corresponding point, calculate its similarity ratio with the nearest neighbor descriptor (i.e., the ratio of the similarity of the corresponding point to the similarity of the next nearest neighbor descriptor). If the ratio is less than a preset threshold (such as 0.75), the corresponding point is retained; otherwise, it is removed. This step helps to eliminate false matches caused by factors such as image noise and repeated textures.

[0099] Step S5014: Use the RANSAC algorithm to eliminate mismatched point pairs in the initial screening of corresponding points, thereby outputting the final corresponding points.

[0100] In this step, the algorithm attempts to randomly select three non-collinear points from the initial screening corresponding points, calculate the affine transformation matrix, and count the number of corresponding points that satisfy the matrix (i.e., the number of inliers). After multiple iterations, the affine transformation matrix with the largest number of inliers is selected as the final model, and the corresponding points that do not satisfy the matrix (i.e., mismatched point pairs) are removed.

[0101] Determine whether the argon tube base has rotational changes based on the center point and the corresponding point, and determine whether the real-time image has scaling based on the external graphics of the target edge area, including:

[0102] Fitting a circle using the least squares method to estimate the rotation angle between the feature point and the corresponding point; when the rotation angle is greater than a preset rotation threshold, it is determined that the base has a rotation change; and

[0103] The size of the bounding rectangle of the target edge area is obtained, and the size of the bounding rectangle is compared with the preset edge size corresponding to the standard template image to calculate the scaling ratio; if the scaling ratio exceeds the preset scaling threshold range, it is determined that there is a scaling change in the base.

[0104] In step S600, the robot arm is controlled to clamp the argon gas tube and automatically install it to the argon gas tube port, or to remove the argon gas tube from the argon gas tube port according to the positioning information, including:

[0105] Step S601, obtaining a first current position of a robot arm actuator and a second current position of an argon gas tube;

[0106] Step S602: upon receiving the installation signal, determining a first motion path according to the first current position, the second current position and the positioning information, so as to control the robot arm to clamp the argon gas tube and automatically install it to the argon gas tube port according to the first motion path;

[0107] Step S603: upon receiving the disassembly signal, determining a second motion path according to the first current position, the second current position and the positioning information, so as to control the robot arm to disassemble the argon gas tube along the second motion path to the argon gas tube outlet.

[0108] Determine the first motion path when the installation signal is received: Determine the optimal motion path based on the current position of the robot arm actuator (first current position), the current position of the argon tube (second current position) and the target position (positioning information). Factors to be considered may include path length, time, obstacle avoidance, etc. Control the robot arm to clamp the argon tube and automatically install it according to the first motion path: According to the planned path, the robot arm performs clamping, moving and installation actions; this step may involve precise force control and position control to ensure the accuracy and reliability of the installation.

[0109] When the disassembly signal is received, similarly, the motion path for disassembly is determined according to the current position of the robot arm actuator, the current position of the argon tube, and the positioning information. The robot arm is controlled to disassemble the argon tube according to the second motion path to the argon tube outlet: according to the planned path, the robot arm performs actions such as moving to the disassembly position, clamping, separating, and moving back to the original position; similarly, this step also requires precise force control and position control.

[0110] like Figure 8The embodiment of the present application also provides a second aspect of the present application which also provides an automatic disassembly and assembly system 100 for the argon pipe mouth in the hot repair of a ladle, including a robotic arm 10, an end effector 20, an image acquisition device 30, an argon pipe storage rack 40 and a controller 50.

[0111] The end effector 20 is used to clamp or release the argon tube; the image acquisition device 30 is used to obtain a real-time image of the argon tube outlet; the argon tube storage rack 40 is used to place the argon tube; the controller is electrically connected to the robotic arm 10 (or robot), the end effector 20 and the image acquisition device 30, and is configured into the above-mentioned automatic disassembly and assembly method.

[0112] The image acquisition device 30 obtains a real-time image of the argon pipe opening and transmits the image information to the controller 50; the controller 50 analyzes the image information to determine the position, state and disassembly strategy of the argon pipe opening; the controller 50 controls the robot arm 10 to move the end effector 20 to the position of the argon pipe opening; the end effector 20 clamps or releases the argon pipe according to the instruction of the controller 50; after completing the disassembly operation, the robot arm 10 moves the end effector 20 back to its original position and waits for the next operation.

[0113] Furthermore, those skilled in the art should understand that if the automatic disassembly and assembly system 100 provided in the embodiment of the present application is combined or replaced by all or part of the sub-modules involved in each product through embodiments such as fusion, simple change, mutual transformation, etc., such as the components are placed and moved, or designed to be detachable; or the products they constitute are set as one, such as an integrated design; all the combined components can form equipment / devices / systems with specific functions, and using such equipment / devices / systems to replace the corresponding components of the present application also falls within the protection scope of the present application.

[0114] The various technical features described above can be combined arbitrarily. Although all possible combinations of these technical features are not described, any combination of these technical features should be considered to be covered by this specification as long as there is no contradiction in such combination.

[0115] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still adjust the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these adjustments or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for automatically disassembling and assembling an argon pipe orifice during hot repair of a ladle, characterized in that: include: Acquire a real-time image of the argon gas pipe opening and extract an interface edge region in the real-time image; Extracting strong edges and weak edges in the interface edge region by edge detection, and connecting the strong edges and weak edges to form a target edge region; Extracting a plurality of characteristic regions in the target edge region, and performing pixel matching between the plurality of characteristic regions and corresponding preset standard template images to calculate contrast values; The similarity values ​​obtained by matching are sorted after threshold judgment, and the feature area position corresponding to the highest similarity value that meets the threshold judgment is taken as the best matching position; According to the best matching position, the real-time image can be selectively adjusted by at least one of multi-scale template matching or template matching based on rotation invariance to determine the positioning information of the argon gas tube base; According to the positioning information, the robot arm is controlled to clamp the argon tube and automatically install it to the argon tube port, or to remove the argon tube from the argon tube port.

2. The automatic disassembly and assembly method according to claim 1, characterized in that: The step of extracting strong edges and weak edges in the interface edge region by edge detection, and connecting the strong edges and weak edges to form a target edge region, comprises: Convolving the pixel points in the target edge area with a two-dimensional Gaussian kernel to obtain a processed edge image; Calculating the horizontal gradient and the vertical gradient of the processed edge image to obtain the gradient amplitude and direction of each pixel; Obtaining the local maximum of the processed edge image by non-maximum suppression; The local maximum values ​​with a strength greater than a preset first intensity threshold are extracted as strong edges, and the local maximum values ​​with a strength less than a preset second intensity threshold are extracted as weak edges, and the strong edges and weak edges are connected to form a target edge region.

3. The automatic disassembly and assembly method according to claim 1, characterized in that: The pixel points in the target edge area are convolved with a two-dimensional Gaussian kernel to obtain a processed edge image by the following formula: Among them, G(x, y) is the data representation of the edge image after processing, σ is the standard deviation, and x and y are the coordinate points of each pixel.

4. The automatic disassembly and assembly method according to claim 3, characterized in that: The step of extracting a plurality of feature regions from the target edge region and performing pixel matching between the plurality of feature regions and corresponding preset standard template images to calculate contrast values ​​comprises: The contrast value between the feature area and the standard template image is obtained using the following formula: Among them, R(x, y) is the calculated contrast value, T(i, j) is the pixel value of the corresponding standard template, I(x+i, y+j) is the local pixel value of the feature area at the position (x, y), and are the average values ​​of the standard template and the feature area, respectively.

5. The automatic disassembly and assembly method according to claim 2, characterized in that: The similarity values ​​obtained by matching are sorted after threshold judgment, and the feature region position corresponding to the highest similarity value satisfying the threshold judgment is taken as the best matching position, including: Performing a threshold judgment on the similarity values ​​to filter out valid similarity values ​​exceeding a preset first similarity threshold; Sort the valid similarity values ​​and obtain the highest similarity value; The feature region position corresponding to the highest similarity value is obtained as the best matching position.

6. The automatic disassembly and assembly method according to claim 2, characterized in that: The method of selectively adjusting the real-time image by at least one of multi-scale template matching or template matching based on rotation invariance according to the best matching position to determine the positioning information of the argon gas tube base includes: Select at least three feature points in the feature area of ​​the best matching position, wherein the feature points at least include the center point of the argon tube base, and use a feature point detection algorithm to extract corresponding points in the standard template image corresponding to the feature points; Determine whether there is a rotation change in the argon tube base according to the center point and the corresponding point, and determine whether there is scaling in the real-time image according to the external graphics of the target edge area; When the argon tube base is rotated or the real-time image is scaled, an affine transformation matrix is ​​calculated based on the detected feature points and the corresponding points, and the real-time image is transformed using the affine transformation matrix to align it with the standard template image, and then the positioning information in the aligned real-time image is output.

7. The automatic disassembly and assembly method according to claim 6, characterized in that: The calculating an affine transformation matrix according to the detected feature points and the corresponding points comprises: The affine transformation matrix is ​​calculated using the following formula: Where M is a 2x3 affine transformation matrix, including rotation, translation, and scaling.

8. The automatic disassembly and assembly method according to claim 6, characterized in that: The extracting corresponding points in the standard template image corresponding to the feature points using a feature point detection algorithm includes: For each detected feature point, generate a SIFT feature descriptor; Using a matching algorithm to preliminarily match the feature descriptors in the real-time image with the feature descriptors in the standard template image, and returning preliminarily screened corresponding points; The corresponding points of the preliminary screening are obtained from the corresponding points of the preliminary screening by judging the threshold value of the ratio test; The RANSAC algorithm is used to eliminate mismatched point pairs from the initial screening of corresponding points, thereby outputting the final corresponding points.

9. The automatic disassembly and assembly method according to claim 6, characterized in that: The method of controlling the robot arm to clamp the argon tube and automatically install it to the argon tube port, or removing the argon tube from the argon tube port according to the positioning information, includes: Obtaining a first current position of the robot arm actuator and a second current position of the argon gas tube; When receiving the installation signal, determining a first motion path according to the first current position, the second current position and the positioning information, so as to control the robot arm to clamp the argon gas tube and automatically install it to the argon gas tube port according to the first motion path; When the disassembly signal is received, a second motion path is determined according to the first current position, the second current position and the positioning information, so as to control the robot arm to disassemble the argon gas tube along the second motion path to the argon gas tube outlet. The determining whether there is a rotation change of the argon gas tube base according to the center point and the corresponding point, and determining whether there is scaling of the real-time image according to the external graphics of the target edge area, includes: Fitting a circle using the least squares method to estimate the rotation angle between the feature point and the corresponding point; when the rotation angle is greater than a preset rotation threshold, it is determined that the base has a rotation change; and The size of the circumscribed rectangle of the target edge area is obtained, and the size of the circumscribed rectangle is compared with the preset edge size corresponding to the standard template image to calculate the scaling ratio; if the scaling ratio exceeds the preset scaling threshold range, it is determined that there is a scaling change in the base.

10. An automatic disassembly and assembly system for argon pipe opening in ladle hot repair, characterized in that: include: Robotic arm; The end effector is used to clamp or release the argon tube; An image acquisition device, used to obtain a real-time image of the argon gas pipe opening; Argon tube storage rack, used to place argon tubes; The controller is electrically connected to the robot arm, the end effector and the image acquisition device, and is configured to execute the automatic disassembly and assembly method as described in any one of claims 1 to 9 above.

Images