An intelligent centering method for double-sided shearing of steel plates based on machine vision

Abstract

An intelligent centering method for a double-sided shearing machine of steel plates based on machine vision includes: Step 1: Convey the steel plate to the shearing area of the double-sided shearing machine and send a control signal; Step 2: After receiving the control signal, control the camera group to take pictures to obtain the contour position coordinates of the steel plate; Step 3: Based on the contour position coordinates of the steel plate, calculate the maximum inscribed rectangle of the steel plate using the particle swarm optimization algorithm; Step 4: Obtain the target cutting rectangle according to the maximum inscribed rectangle and the target cutting width; Step 5: Calculate the adjustment amount of the magnetic centering device according to the distance between the edge of the target cutting rectangle close to the driving side and the extension line of the shear blade; Step 6: After adjustment, repeat Steps 3 to 5 to determine whether the steel plate is adjusted in place. If it is adjusted in place, enter the next process; Step 7: If it is not adjusted in place and the maximum allowable number of times is not exceeded, repeat Steps 3 to 6; Step 8: If it is not adjusted in place and the maximum allowable number of times is exceeded, perform manual adjustment.

Description

Technical Field

[0001] The present invention belongs to the technical field of metal processing and relates to an intelligent centering method for double-sided shears of steel plates based on machine vision. Background Art

[0002] With the rapid development of emerging communication technologies, computer network technologies, and intelligent control technologies, it is driving the manufacturing industry towards intelligent production transformation. Steel production is the industry with the most urgent demand for intelligent manufacturing and is also the industry closest to intelligent manufacturing. Although there is no mature intelligent manufacturing standard system and solution in the industry, major steel enterprises are actively exploring and moving forward. In the medium and heavy plate production line, the efficiency of the shearing line has always been an important factor restricting the production efficiency of the production line. On traditional shearing lines, the centering operation of double-sided shears has always relied on the human eye, observing the position of the steel plate in cooperation with a laser line marker, and manually adjusting the position of the steel plate on the roller table. The centering operation takes a long time and is inaccurate.

[0003] Regarding the problem of automatic centering of double-sided shears for steel plates, domestic researchers have conducted many related studies. The Chinese invention patent "An Automatic Centering Control Method for a Magnetic Centering Device of Double-Sided Shears of Steel Plates" with the application number CN202110527605.6 uses a cold metal detector, two laser line markers, and several industrial cameras installed in the magnetic centering area of the double-sided shears to obtain the images of the steel plate and the laser line in the magnetic centering area of the double-sided shears, calculate the distance between the edge of the steel plate and the laser line, and then calculate the adjustment amount required for each electromagnet to straighten and center the steel plate. Finally, the automatic centering of the steel plate is achieved through the lifting and lateral movement control of the electromagnets. However, this method needs to rely on two laser line markers to determine the cutting edge position during the process of obtaining the adjustment amount. Therefore, the laser line marker needs to move with the shear blade of the double-sided shears. The marking direction of the laser line marker should be parallel to the length direction of the steel plate and the position of the laser line should ensure that the shearing position of the double-sided shears coincides, which poses high requirements for the installation and maintenance of the laser line marker. In addition, the calculation result of the adjustment amount of this method is obtained on the premise of defaulting that the steel plate is flat, and it cannot consider the influence of special situations such as "sickle bend" on the calculation result of the adjustment amount. Summary of the Invention

[0004] To solve the above technical problems, the purpose of the present invention is to provide an intelligent centering method for double-sided shears of steel plates based on machine vision, which uses several cameras to splice the contour of the steel plate by means of machine vision and obtain the position information of the steel plate, automatically calculates the distance that the centering device needs to move, and guides the centering device to achieve the automatic centering of the steel plate, improving the efficiency of the centering work, reducing the work intensity of workers, and reducing the accident rate.

[0005] The present invention provides an intelligent centering method for double-sided shears of steel plates based on machine vision, including:

[0006] Step 1: Convey the steel plate to the conveyor rollers in the to-be-sheared area at the front end of the entrance of the double-sided shearing device, and the basic automation system of the double-sided shear sends a control signal to the intelligent centering system;

[0007] Step 2: After receiving the control signal, the intelligent centering system controls the camera group located above the to-be-sheared area of the double-sided shear to take pictures to obtain the contour position coordinates of the steel plate;

[0008] Step 3: Based on the contour position coordinates of the steel plate obtained in Step 2, calculate the maximum inscribed rectangle of the steel plate based on the particle swarm optimization algorithm;

[0009] Step 4: Obtain the target cutting rectangle according to the maximum inscribed rectangle of the steel plate obtained in Step 3 and the target cutting width of the steel plate;

[0010] Step 5: Calculate the adjustment amount of each magnetic centering device according to the distance between the edge of the target cutting rectangle near the drive side and the extended line of the shear blade, and make adjustments;

[0011] Step 6: After this round of adjustment is completed, obtain the steel plate contour, the maximum inscribed rectangle and the target cutting rectangle again, and judge whether the steel plate is adjusted at this time. If the adjustment is in place, enter the next process;

[0012] Step 7: If the adjustment is not in place and the adjustment times of the steel plate do not exceed the maximum allowable adjustment times at this time, repeat Steps 3 to 6;

[0013] Step 8: If the adjustment is not in place and the adjustment times of the steel plate exceed the maximum allowable adjustment times at this time, the system stops automatic adjustment and notifies the operator that manual adjustment is required.

[0014] Further, Step 2 is specifically as follows:

[0015] Step 2.1: Stitch multiple photos taken by the camera group to obtain a complete steel plate image;

[0016] Step 2.2: Use the Canny operator for edge detection to obtain the steel plate contour image.

[0017] Further, Step 2.2 is specifically as follows:

[0018] Step 2.2.1: Perform binary segmentation on the strip steel image collected by the camera, set the gray value of the pixel points greater than the threshold to the maximum value 255, and set the gray value of the pixel points less than or equal to the threshold to the minimum value 0 to obtain the image A1 after binary segmentation;

[0019] Step 2.2.2: Perform morphological operations of dilation and erosion on the image A1 to fill the holes existing in the image A1 due to image noise to obtain the image A2 after morphological processing;

[0020] Step 2.2.3: Find the connected region B with the largest area in image A2. The region corresponding to the connected region B is the region where the strip is located in the image. Keep the image gray value of this connected region unchanged, and set the gray values of other connected regions in A2 to 0. At this time, an image A3 with only the connected region B as the foreground and a background gray value of 0 is obtained.

[0021] Step 2.2.4: Use the Canny edge detection algorithm to perform edge detection on image A3, and extract the boundary contour line Z between the connected region B and the background. The position coordinates of m pixel points on Z are (x1, y1), (x2, y2), ……, (x m , y m ).

[0022] Further, the specific steps of step 3 are as follows:

[0023] Step 3.1: Establish a coordinate system. The x-axis of the coordinate system is parallel to the extension line of the shear blade of the double-sided shearing device. The positive direction of the x-axis is consistent with the steel plate conveying direction, and the direction is from the waiting-to-be-sheared area of the double-sided shear to the working area of the double-sided shear. The positive direction of the y-axis is from the drive side to the operation side.

[0024] Step 3.2: Randomly generate N rectangles as particles to be optimized. The position P of particle i i is described by (x i , y i , L i , H i , θ i ). The meanings of these 5 attributes are as follows:

[0025] x i is the abscissa of the center position of the rectangle corresponding to the i-th particle; y i is the ordinate of the center position of the rectangle corresponding to the i-th particle; L i is the length of the rectangle corresponding to the i-th particle; H i is the width of the rectangle corresponding to the i-th particle; θ i is the inclination angle of the rectangle corresponding to the i-th particle.

[0026] Step 3.3: Randomly initialize the velocity of each particle. The velocity V of particle i i is represented by (V i,x , V i,y , V i,L , V i,H , V i,θ ). The meanings of these 5 velocity components are as follows:

[0027] V i,x is the moving velocity of the i-th particle in the abscissa component; V i,yis the moving speed of the $i$-th particle in the vertical coordinate component; $V$ i,L is the moving speed of the $i$-th particle in the $L$ component; $V$ i,H is the moving speed of the $i$-th particle in the $H$ component; $V$ i,θ is the moving speed of the $i$-th particle in the $\theta$ component;

[0028] Step 3.4: Initialize the optimal position $P'$ searched by each particle; Initialize the optimal position $P'$ searched among all particles; Initialize the maximum area $S'$ searched by each particle i to 0; Initialize the maximum area $S'$ searched among all particles to 0; i

[0029] Step 3.5: Calculate the area $S$ of the rectangle represented by each particle i , if the rectangle is completely inside the steel plate and the area is greater than the maximum area $S'$ searched by the particle i , then assign the current position $P$ of the particle i to the optimal position $P'$ of the particle, and assign the area $S$ of the rectangle represented by the particle i to the maximum area $S'$ searched by the particle i ; i

[0030] Step 3.6: Find the particle with the largest rectangle area among all current particles. If the rectangle area corresponding to the particle is greater than the maximum area $S'$ searched among all particles, then assign the current position $P$ of the particle i to the optimal position $P'$ searched among all particles, and assign the area $S$ of the rectangle represented by the particle i to the maximum area $S'$ searched by all particles;

[0031] Step 3.7: Update the speed of each particle:

[0032] $V$ i k+1 $=$ $\omega$ k $V$ i k $+$ $c_1r_1(P$ i $'$ $-$ $P$ i ) $+$ $c_2r_2(P'$ $-$ $P$ i )

[0033] where $V$ i k represents the speed of particle $i$ in the $k$-th iteration; $\omega$ k is the inertia weight in the $k$-th iteration; $c_1$ is the individual learning weight; $c_2$ is the global learning weight; $r_1$, $r_2$ are random numbers in the interval $[0, 1]$, which are updated in each iteration; ​​

[0034] Step 3.8: Update the position of each particle:

[0035] P i k+1 = P i k + V i k+1

[0036] where P i k represents the position of particle i in the k-th iteration;

[0037] Step 3.9: Update the inertia weight:

[0038] ω k+1 = γω k

[0039] where: γ is the decay rate of the weight;

[0040] Step 3.10: Repeat Steps 3.5 to 3.9 until the number of iterations reaches the maximum number of iterations M. Then, the rectangle corresponding to the optimal position P′(x′, y′, L′, H′, θ′) searched among all particles is the calculated maximum inscribed rectangle of the steel plate.

[0041] Furthermore, in Step 3.5, it is determined whether the rectangle is completely inside the steel plate according to the following method:

[0042] Step 3.5.1: Taking the center position (x i , y i ) of the rectangle as the center point O, emit rays in all directions;

[0043] Step 3.5.2: Each ray will intersect with the rectangle and the edge contour of the steel plate, and the intersections are denoted as n1 and n2 respectively;

[0044] Step 3.5.3: Calculate the distances u1 and u2 between O and n1 and between O and n2 respectively. If u1 is less than or equal to u2 in each direction, it is considered that the rectangle is completely inside the steel plate.

[0045] Furthermore, Step 4 is specifically as follows:

[0046] The maximum inscribed rectangle is expressed as P′(x′, y′, L′, H′, θ′), then the target cutting rectangle is expressed as P″(x′, y′, L′, w, θ′), where w is the target cutting width, that is, the distance between the two side blades of the double-sided shearing device.

[0047] Furthermore, Step 5 is specifically as follows:

[0048] Step 5.1: Let the extension line of the shear blade near the drive side of the double-sided shearing device be l1, the side of the target cutting rectangle near the drive side be eh, and there are n magnetic centering devices M on the conveyor roller n ;

[0049] Step 5.2: Given that the center point position of the target cutting rectangle is (x′, y′) and the inclination angle of the target cutting rectangle is θ′, then through geometric relationships, the expression of side eh is:

[0050]

[0051] The expression of the shear blade extension line l1 is:

[0052] y = h1

[0053] where h1 is the distance between the shear blade extension line l1 and the x-axis;

[0054] Step 5.3: Set the magnetic centering device M n The expression of the line where the moving direction is located is x = m n ; Then the intersection coordinates of eh and the central axis of M n are expressed as:

[0055]

[0056] The intersection of l1 and the magnetic centering device M n on the central axis is (m n , h1). Therefore, the distance between the two intersections, that is, the adjustment amount of the magnetic centering device M n is:

[0057]

[0058] An intelligent centering method for a steel plate double-sided shear based on machine vision in the present invention automatically identifies the contour edges of the steel plate before the double-sided shear based on machine vision and intelligent algorithms, and real-time feedbacks to the basic automation system of the double-sided shear to complete the closed-loop control function of the steel plate position, thereby realizing intelligent centering and ensuring the improvement of the steel plate shearing accuracy. At the same time, it can greatly reduce the labor intensity of workers in the double-sided shear area, improve the production efficiency of the shearing unit, effectively reduce the trimming loss caused by the failure of manual centering, and further improve the intelligent manufacturing level of the production line. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 is a schematic diagram of the adjustment process of an intelligent centering method for a steel plate double-sided shear based on machine vision in the present invention;

[0060] Figure 2 is a flowchart of an intelligent centering method for a steel plate double-sided shear based on machine vision in the present invention. Detailed implementation mode

[0061] Taking Figure 1 A certain situation shown as an example, the adjustment process of the present method will be described in detail. In this embodiment, magnetic centering devices are installed at positions M1 and M2 to adjust the position of the steel plate. If there are centering devices at more positions in actual application, these devices can also follow the same logic for adjustment. In the figure, the length of the steel plate is L, the width is H, and l1 and l2 are the extension lines of the two shear blades of the double-sided shearing device respectively. h1 and h2 are the distances between l1 and l2 and the position reference line respectively.

[0062] As Figure 2 shown, a kind of intelligent centering method for double-sided shearing of steel plates based on machine vision of the present invention includes:

[0063] Step 1: Convey the steel plate to the conveyor roller in the double-sided shearing waiting area at the front end of the entrance of the double-sided shearing device, and the basic automation system of the double-sided shearing sends a control signal to the intelligent centering system.

[0064] Step 2: After receiving the control signal, the intelligent centering system controls the camera group located above the double-sided shearing waiting area to take pictures to obtain the contour position coordinates of the steel plate. The specific content of step 2 is:

[0065] Step 2.1: Stitch multiple photos taken by the camera group to obtain a complete steel plate image;

[0066] Step 2.2: Use the Canny operator for edge detection to obtain the steel plate contour image. The specific content of step 2.2 is:

[0067] Step 2.2.1: Perform binary segmentation on the strip steel image collected by the camera, set the gray value of the pixel points greater than the threshold to the maximum value 255, and set the gray value of the pixel points less than or equal to the threshold to the minimum value 0 to obtain the image A1 after binary segmentation;

[0068] Step 2.2.2: Perform morphological operations of dilation and erosion on the image A1 to fill the holes existing in the image A1 due to image noise to obtain the image A2 after morphological processing;

[0069] Step 2.2.3: Find the largest connected domain B in the image A2. The area corresponding to the connected domain B is the area where the strip steel is located in the image; keep the gray value of the image of this connected domain unchanged, and set the gray values of other connected domains in A2 to 0. At this time, an image A3 with only the connected domain B as the foreground and a background gray value of 0 is obtained;

[0070] Step 2.2.4: Use the Canny edge detection algorithm to perform edge detection on image A3, and extract the boundary contour line Z between the connected domain B and the background. The position coordinates of m pixel points on Z are (x1, y1), (x2, y2), ……, (x m ,y m ).

[0071] Step 3: According to the position coordinates of the steel plate contour obtained in Step 2, calculate the maximum inscribed rectangle abcd of the steel plate based on the particle swarm optimization algorithm. The specific steps of Step 3 are as follows:

[0072] Step 3.1: Set the position reference line and establish a coordinate system. As Figure 1 shown, the x-axis of the coordinate system is parallel to the extension line of the shear blade of the double-sided shearing device, and the positive direction of the x-axis is consistent with the steel plate transportation direction, pointing from the to-be-sheared area of the double-sided shear to the working area of the double-sided shear; the positive direction of the y-axis points from the drive side to the operation side;

[0073] Step 3.2: Randomly generate N rectangles as particles to be optimized. The position P i of particle i is described by (x i ,y i ,L i ,H i ,θ i ). The meanings of these 5 attributes are as follows:

[0074] x i is the abscissa of the center position of the rectangle corresponding to the i-th particle; y i is the ordinate of the center position of the rectangle corresponding to the i-th particle; L i is the length of the rectangle corresponding to the i-th particle; H i is the width of the rectangle corresponding to the i-th particle; θ i is the inclination angle of the rectangle corresponding to the i-th particle;

[0075] Step 3.3: Randomly initialize the velocity of each particle. The velocity V i of particle i is represented by (V i,x ,V i,y ,V i,L ,V i,H ,V i,θ ). The meanings of these 5 velocity components are as follows:

[0076] V i,x is the moving velocity of the i-th particle in the abscissa component; V i,y is the moving velocity of the i-th particle in the ordinate component; V i,L is the moving velocity of the i-th particle in the L component; V i,H is the moving velocity of the i-th particle in the H component; Vi,θ is the moving speed of the i-th particle in the θ component;

[0077] Step 3.4: Initialize the optimal position P i ′ searched by each particle; Initialize the optimal position P′ searched among all particles; Initialize the maximum area S′ searched by each particle i to be 0; Initialize the maximum area S′ searched among all particles to be 0;

[0078] Step 3.5: Calculate the area S of the rectangle represented by each particle i , if the rectangle is completely inside the steel plate and the area is greater than the maximum area S′ searched by the particle i , then assign the current position P i of the particle to the optimal position P i ′ of the particle, and assign the area S i of the rectangle represented by the particle to the maximum area S′ searched by the particle i ;

[0079] Specifically, when implementing, judge whether the rectangle is completely inside the steel plate according to the following method:

[0080] Step 3.5.1: Take the center position (x i , y i ) of the rectangle as the center point O and emit rays in all directions;

[0081] Step 3.5.2: Each ray will intersect with the rectangle and the edge contour of the steel plate, and the intersection points are denoted as n1 and n2 respectively;

[0082] Step 3.5.3: Calculate the distances u1 and u2 between O and n1 and between O and n2 respectively. If u1 is less than or equal to u2 in each direction, it is considered that the rectangle is completely inside the steel plate;

[0083] Step 3.6: Find the particle with the largest rectangle area among all current particles. If the area of the rectangle corresponding to the particle is greater than the maximum area S′ searched among all particles, then assign the current position P i of the particle to the optimal position P′ searched among all particles, and assign the area S i of the rectangle represented by the particle to the maximum area S′ searched by all particles;

[0084] Step 3.7: Update the velocity of each particle:

[0085] V i k+1 = ω k V i k + c1r1(Pi ′-P i ) + c2r2(P′ - P i )

[0086] Wherein, V i k represents the velocity of particle i in the k-th iteration; ω k is the inertia weight in the k-th iteration; c1 is the individual learning weight; c2 is the global learning weight; r1, r2 are random numbers in the interval [0, 1], which are updated at each iteration;

[0087] Step 3.8: Update the position of each particle:

[0088] P i k+1 = P i k + V i k+1

[0089] Wherein, P i k represents the position of particle i in the k-th iteration;

[0090] Step 3.9: Update the inertia weight:

[0091] ω k+1 = γω k

[0092] Where: γ is the decay rate of the weight;

[0093] Step 3.10: Repeat Steps 3.5 to 3.9 until the number of iterations reaches the maximum number of iterations M. Then, the rectangle corresponding to the optimal position P′(x′, y′, L′, H′, θ′) searched among all particles is the calculated maximum inscribed rectangle abcd of the steel plate.

[0094] Step 4: Obtain the target cutting rectangle efgh according to the maximum inscribed rectangle of the steel plate obtained in Step 3 and the target cutting width of the steel plate;

[0095] Specifically, when implemented, the maximum inscribed rectangle is expressed as P′(x′, y′, L′, H′, θ′), then the target cutting rectangle is expressed as P″(x′, y′, L′, w, θ′), where w is the target cutting width, that is, the distance between the two side blades of the double-sided shearing device.

[0096] Step 5: Calculate the adjustment amount of each magnetic centering device according to the distance between the edge of the target cutting rectangle close to the driving side and the extension line of the shearing blade, and perform the adjustment. The specific content of Step 5 is as follows:

[0097] Step 5.1: Let the extension line of the shear blade near the drive side of the double-sided shearing device be l1, the side of the target cutting rectangle near the drive side be eh, and there are n magnetic centering devices M on the conveyor roller. n ;

[0098] Step 5.2: Given that the center point position of the target cutting rectangle is (x′, y′) and the inclination angle of the target cutting rectangle is θ′, then through geometric relationships, the expression for side eh is:

[0099]

[0100] The expression for the shear blade extension line l1 is:

[0101] y = h1

[0102] where h1 is the distance between the shear blade extension line l1 and the x-axis;

[0103] Step 5.3: Set the magnetic centering device M n The expression for the straight line where the moving direction is located is x = m n ; then the intersection coordinates of eh and the central axis of M n are expressed as:

[0104]

[0105] The intersection of l1 and the magnetic centering device M n on the central axis is (m n , h1), so the distance between the two intersections, that is, the adjustment amount of the magnetic centering device M n is:

[0106]

[0107] Step 6: After this round of adjustment is completed, obtain the steel plate contour, the largest inscribed rectangle, and the target cutting rectangle again, and judge whether the steel plate is adjusted at this time. If the adjustment is in place, enter the next process;

[0108] Specifically, when implementing, judge the absolute value Δh of the difference between h M1 and h1 M1 and the absolute value Δh of the difference between h M2 and h2 M2 whether it is within the allowable error range. If it is within the range, the centering adjustment is in place. h M1 is the distance from the intersection of the side of the adjusted target cutting rectangle and the magnetic centering device M1 to the position reference line, and h M2 is the distance from the intersection of the side of the adjusted target cutting rectangle and the magnetic centering device M2 to the position reference line.

[0109] Step 7: If the adjustment is not in place and the number of adjustment times for the steel plate does not exceed the maximum allowable adjustment times at this time, repeat Steps 3 to 6.

[0110] Step 8: If the adjustment is not in place and the number of adjustment times for the steel plate exceeds the maximum allowable adjustment times at this time, the system stops automatic adjustment and notifies the operator that manual adjustment is required.

[0111] The above are only the preferred embodiments of the present invention and are not intended to limit the idea of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An intelligent centering method for double-sided shearing of steel plates based on machine vision, characterized in that, Including: Step 1: Transport the steel plate to the conveyor rollers in the to-be-sheared area of the double-sided shear at the front end of the entrance of the double-sided shear device. The basic automation system of the double-sided shear sends a control signal to the intelligent centering system. Step 2: After receiving the control signal, the intelligent centering system controls the camera group located above the to-be-sheared area of the double-sided shear to take pictures to obtain the contour position coordinates of the steel plate. Step 3: Based on the contour position coordinates of the steel plate obtained in Step 2, calculate the maximum inscribed rectangle of the steel plate based on the particle swarm optimization algorithm. Step 4: Obtain the target cutting rectangle according to the maximum inscribed rectangle of the steel plate obtained in Step 3 and the target cutting width of the steel plate. Step 5: Calculate the adjustment amount of each magnetic centering device according to the distance between the edge of the target cutting rectangle near the drive side and the extended line of the shear blade, and make adjustments. Step 6: After this round of adjustment is completed, obtain the steel plate contour, the maximum inscribed rectangle, and the target cutting rectangle again, and judge whether the steel plate is adjusted in place at this time. If it is adjusted in place, enter the next process. Step 7: If it is not adjusted in place and the adjustment times of this steel plate do not exceed the maximum allowable adjustment times at this time, repeat Steps 3 to 6. Step 8: If it is not adjusted in place and the adjustment times of this steel plate exceed the maximum allowable adjustment times at this time, the system stops automatic adjustment and notifies the operator that manual adjustment is required.

2. The intelligent centering method for the double-sided shearing of steel plates based on machine vision according to claim 1, wherein, The specific content of Step 2 is as follows: Step 2.1: Stitch multiple photos taken by the camera group to obtain a complete steel plate image. Step 2.2: Use the Canny operator for edge detection to obtain the steel plate contour image.

3. The intelligent centering method for double-sided shearing of steel plates based on machine vision according to claim 2, wherein The specific content of Step 2.2 is as follows: Step 2.2.1: Perform binary segmentation on the strip steel image collected by the camera, set the gray value of the pixel points greater than the threshold to the maximum value 255, and set the gray value of the pixel points less than or equal to the threshold to the minimum value 0 to obtain the image A1 after binary segmentation. Step 2.2.2: Perform morphological operations of dilation and erosion on the image A1 to fill the holes existing in the image A1 due to image noise to obtain the image A2 after morphological processing. Step 2.2.3: Find the largest connected domain B in the image A2. The area corresponding to the connected domain B is the area where the strip steel is located in the image. Keep the gray value of the image of this connected domain unchanged, and set the gray values of other connected domains in A2 to 0. At this time, obtain the image A3 with only the connected domain B as the foreground and the background gray value of 0. Step 2.2.4: Use the Canny edge detection algorithm to perform edge detection on image A3, and extract the contour line Z at the junction of the connected domain B and the background. The position coordinates of m pixel points on Z are respectively (x1, y1), (x2, y2), ……, (x m , y m ).

4. The intelligent centering method for the double-sided shearing of steel plates based on machine vision according to claim 1, wherein, The specific content of Step 3 is as follows: Step 3.1: Establish a coordinate system. The x-axis of the coordinate system is parallel to the extended line of the shear blade of the double-sided shear device. The positive direction of the x-axis is consistent with the steel plate transportation direction, and the direction is from the to-be-sheared area of the double-sided shear to the working area of the double-sided shear; the positive direction of the y-axis is from the drive side to the operator side. Step 3.2: Randomly generate N rectangles as particles to be optimized, and the position P of particle i i is described by (x i , y i , L i , H i , θ i ), and the meanings of these five attributes are as follows: x i is the abscissa of the center position of the rectangle corresponding to the i-th particle; y i is the ordinate of the center position of the rectangle corresponding to the i-th particle; L i is the length of the rectangle corresponding to the i-th particle; H i is the width of the rectangle corresponding to the i-th particle; θ i is the inclination angle of the rectangle corresponding to the i-th particle; Step 3.3: Randomly initialize the velocity of each particle. The velocity of particle i is V i represented by (V i,x , V i,y , V i,L , V i,H , V i,θ ). The meanings of these five velocity components are as follows: V i,x is the moving speed of the i-th particle in the abscissa component; V i,y is the moving speed of the i-th particle in the ordinate component; V i,L is the moving speed of the i-th particle in the L component; V i,H is the moving speed of the i-th particle in the H component; V i,θ is the moving speed of the i-th particle in the θ component; Step 3.4: Initialize the optimal position P i ' searched by each particle; Initialize the optimal position P' searched among all particles; Initialize the maximum area S i ' searched by each particle to be 0; Initialize the maximum area S' searched among all particles to be 0; Step 3.5: Calculate the area S of the rectangle represented by each particle i , if the rectangle is completely inside the steel plate and its area is greater than the maximum area S i ′ searched by the particle, then assign the current position P i of the particle to the optimal position P i ′ of the particle, and assign the area S i of the rectangle represented by the particle to the maximum area S i ′ searched by the particle; Step 3.6: Find the particle with the largest rectangle area among all current particles. If the rectangle area corresponding to this particle is greater than the maximum area S' searched among all particles, then assign the current position P of this particle i to the optimal position P' searched among all particles, and assign the area S of the rectangle represented by this particle i to the maximum area S' searched among all these particles; Step 3.7: Update the velocity of each particle: V i k+1 = ω k V i k + c1r1(P i ′ - P i ) + c2r2(P′ - P i ) Among them, V i k represents the velocity of particle i in the k-th iteration; ω k is the inertia weight in the k-th iteration; c1 is the individual learning weight; c2 is the global learning weight; r1 and r2 are random numbers in the interval [0, 1], which are updated at each iteration; Step 3.8: Update the position of each particle: P i k+1 = P i k + V i k+1 where P i k represents the position of particle i at the k-th iteration; Step 3.9: Update the inertia weight: ω k+1 =γω k Where: γ is the attenuation rate of the weight; Step 3.10: Repeat Steps 3.5 to 3.9 until the iteration times reach the maximum iteration times M. Then, the rectangle corresponding to the optimal position P′(x′, y′, L′, H′, θ′) searched among all particles is the calculated maximum inscribed rectangle of the steel plate.

5. The intelligent centering method for double-sided shears of steel plates based on machine vision according to claim 4, wherein In step 3.5, it is determined whether the rectangle is completely inside the steel plate according to the following method: Step 3.5.1: Taking the center position (x i , y i ) of the rectangle as the center point O, emit rays in all directions; Step 3.5.2: Each ray will intersect with the rectangle and the edge contour of the steel plate, and the intersection points are denoted as n1 and n2 respectively; Step 3.5.3: Calculate the distances u1 and u2 from O to n1 and from O to n2 respectively. If u1 is less than or equal to u2 in each direction, it is considered that the rectangle is completely inside the steel plate.

6. The intelligent centering method for the double-sided shearing of steel plates based on machine vision according to claim 4, characterized in that The specific content of step 4 is as follows: The maximum inscribed rectangle is expressed as P′(x′, y′, L′, H′, θ′), and the target cutting rectangle is expressed as P″(x′, y′, L′, w, θ′), where w is the target cutting width, that is, the distance between the two side blades of the double-sided shearing device.

7. The intelligent centering method for the double-sided shears of steel plates based on machine vision according to claim 6, characterized in that, The specific content of step 5 is as follows: Step 5.1: Let the extension line of the shear blade near the drive side of the double-sided shearing device be l1, the side of the target cutting rectangle near the drive side be eh, and there are n magnetic centering devices M on the conveyor roller n ; Step 5.2: Given that the center point position of the target cutting rectangle is (x′, y′) and the inclination angle of the target cutting rectangle is θ′, the expression of side eh can be obtained through geometric relationships: The expression of the extended line l1 of the cutting blade is: y = h1 where h1 is the distance between the extended line l1 of the cutting blade and the x-axis; Step 5.3: Set the magnetic centering device M n The expression of the straight line where the moving direction is located is x = m n ; then eh and M n The intersection coordinates of the central axis of the magnetic centering device are expressed as: The intersection point of l1 and the magnetic centering device M n with the central axis is (m n , h1). Therefore, the distance between the two intersection points, that is, the adjustment amount of the magnetic centering device M n is:

Images