CN117911419A - Method and device for detecting steel rotation angle enhancement of medium plate, medium and equipment - Google Patents

Abstract

The application relates to the field of rolling image analysis, and discloses a method and a device for detecting steel-turning angle enhancement of a medium plate, a medium and equipment. The method comprises the following steps: shooting a billet image by a target camera, and carrying out distortion correction on the billet image according to camera parameters; sequentially performing defogging treatment and contrast enhancement treatment on the billet image; based on the gray scale of each pixel point in the billet image, a foreground area and a background area are identified, and holes in the foreground area are filled; performing image morphology opening operation on the billet image; determining a target communication area based on the communication relation of the foreground area, and fitting the edge contour of the target communication area; fitting a circumscribed minimum rectangle corresponding to the target communication area according to the edge profile, and determining the angle of the billet according to the offset angle of the rectangle relative to the preset reference direction. The method solves the problems that the image of the existing method has the defects of blurring, noise and the like, and a stable angle detection result cannot be obtained.

Description

Method, device, medium and equipment for enhanced detection of steel turning angle of medium and thick plate

Technical Field

The present application relates to the field of rolling image analysis, and in particular to a method and device, medium, and equipment for enhanced detection of steel turning angles of medium and thick plates.

Background Art

According to the current national standard GB709--2006 "Dimensions, Shapes, Weights and Permissible Deviations of Hot-rolled Steel Plates and Strips", hot-rolled steel plates can be classified by three methods: classification of thickness deviation, classification of thickness upgrades, and size range of hot-rolled steel plates. Conventionally, steel plates can be divided into medium-thick plates, thick plates, and extra-thick plates according to thickness. Usually, steel plates with a thickness of 4~20mm are called medium-thick plates. In the rolling production of medium-thick plates, steel transfer operation is an important link. Today, my country's steel mills have built modern medium-thick plate production lines with sophisticated equipment, complete processes, and high automation. Especially in the rolling area, except for steel transfer operation, all other operations have been automatically controlled. Steel transfer operation has become the only bottleneck for fully automatic control of the medium-thick plate rolling area.

The detection algorithm of the billet angle in the steel turning process is a prerequisite for automatic control of steel turning. The traditional manual steel turning operation mainly detects the billet visually, which is too dependent on the operator's work experience and has low efficiency and accuracy. With the development of intelligent technology, machine vision recognition technology has been widely used in the industrial field. It has the advantages of fast processing speed, high detection accuracy, and easy integration. It has been widely used in steelmaking, continuous casting, and rolling processes in the domestic steel industry.

However, in the actual production process, due to the temperature changes of the billet, the water vapor caused by the high-pressure water dephosphorization and the roller cooling system, and the uneven grayscale distribution on the billet surface, the collected images inevitably have defects such as blur and noise, which seriously affect the image quality and make it impossible to obtain stable angle detection results.

Summary of the invention

In view of this, the present application provides a method and device, medium, and equipment for enhanced detection of the angle of steel turning of medium and thick plates, which solves the problem that the existing methods have defects such as blur and noise in the images and cannot obtain stable angle detection results.

According to one aspect of the present application, a method for enhanced detection of steel turning angle of medium and thick plates is provided, comprising:

In the steel transfer process of medium and thick plate rolling, a target camera is used to capture a steel billet image, and distortion correction is performed on the steel billet image according to camera parameters of the target camera;

The billet image after distortion correction is subjected to defogging and contrast enhancement processing in sequence;

Based on the grayscale of each pixel in the steel billet image after contrast enhancement, identify the foreground area and the background area, and fill the holes in the foreground area;

Perform image morphological opening operation on the billet image after hole filling;

Determine a target connected area based on the connectivity relationship of the foreground area in the steel billet image after the opening operation, and fit the edge contour of the target connected area;

The minimum circumscribed rectangle corresponding to the target connected area is fitted according to the edge contour, and the angle of the steel billet is determined according to the offset angle of the rectangle relative to a preset reference direction.

Optionally, the defogging and contrast enhancement processing are sequentially performed on the distortion-corrected billet image, comprising:

Determine an atmospheric light component corresponding to the distortion-corrected steel billet image, determine the perspective of each pixel in the distortion-corrected steel billet image according to the atmospheric light component, and perform defogging on the distortion-corrected steel billet image according to the atmospheric light component and the perspective;

The gray value of each pixel in the dehazed billet image is stretched by a linear transformation method to obtain a contrast-enhanced billet image.

Optionally, the identifying of the foreground area and the background area based on the grayscale of each pixel in the steel billet image after contrast enhancement processing, and filling the holes in the foreground area includes:

Calculating the average grayscale value of each local window in the steel billet image after contrast enhancement processing, and respectively calculating the deviation between the grayscale value of each pixel to be identified in the steel billet image after contrast enhancement processing and the average grayscale value;

If the deviation is greater than the deviation threshold, the pixel to be identified is determined to be a foreground pixel; otherwise, the pixel to be identified is determined to be a background pixel;

Determine the foreground area according to the foreground pixels, and scan the background pixels in the foreground area;

Holes in the foreground area are determined based on the scanning result, and the holes are filled.

Optionally, determining a target connected area based on the connectivity relationship of the foreground area in the steel billet image after the opening operation, and fitting an edge contour of the target connected area, comprises:

Traversing the pixels in the foreground area, and determining the at least one connected area based on the adjacency relationship of each pixel;

According to the area and rectangularity of each of the connected regions, determining a connected region as the target connected region;

The edge curve of the target connected area is fitted using a three-point interpolation method to obtain a contour point set corresponding to the target connected area.

Optionally, the step of fitting a minimum circumscribed rectangle corresponding to the target connected area according to the edge contour includes:

Preprocessing the contour point set, and performing initial fitting on the preprocessed contour point set using a least squares fitting method to obtain a fitting rectangle;

Residuals from each contour point in the preprocessed contour point set to the fitting rectangle are calculated respectively, and the residuals are weighted;

The weighted residual is refitted to obtain a new fitting rectangle, and the process returns to the step of respectively calculating the residual from each contour point in the preprocessed contour point set to the fitting rectangle, until the preset termination iteration condition is reached, and the current fitting rectangle is determined to be the circumscribed minimum rectangle.

Optionally, the weight corresponding to the residual is negatively correlated with the residual.

Optionally, before performing distortion correction on the steel billet image according to the camera parameters of the target camera, the method further comprises:

The target camera is used to take a plurality of calibration plate images at different angles, and the target camera is calibrated based on the calibration plate images to obtain the camera parameters, wherein the camera parameters include intrinsic parameters, extrinsic parameters and distortion parameters.

According to another aspect of the present application, a device for detecting the enhanced steel turning angle of a medium and thick plate is provided, the device comprising:

An image shooting module, used to shoot a steel billet image through a target camera in the steel transfer link of medium and thick plate rolling, and perform distortion correction on the steel billet image according to the camera parameters of the target camera;

An image processing module is used to sequentially perform defogging and contrast enhancement processing on the steel billet image after distortion correction; and, based on the grayscale of each pixel in the steel billet image after contrast enhancement, identify the foreground area and the background area, and fill the holes in the foreground area; and, perform image morphological opening operation on the steel billet image after the hole filling; and, based on the connectivity relationship of the foreground area in the steel billet image after the opening operation, determine the target connected area, and fit the edge contour of the target connected area;

The angle detection module is used to fit the circumscribed minimum rectangle corresponding to the target connected area according to the edge contour, and determine the angle of the steel billet according to the offset angle of the rectangle relative to a preset reference direction.

Optionally, the image processing module is used to:

Determine an atmospheric light component corresponding to the distortion-corrected steel billet image, determine the perspective of each pixel in the distortion-corrected steel billet image according to the atmospheric light component, and perform defogging on the distortion-corrected steel billet image according to the atmospheric light component and the perspective;

The gray value of each pixel in the dehazed billet image is stretched by a linear transformation method to obtain a contrast-enhanced billet image.

Optionally, the image processing module is used to:

Calculating the average grayscale value of each local window in the steel billet image after contrast enhancement processing, and respectively calculating the deviation between the grayscale value of each pixel to be identified in the steel billet image after contrast enhancement processing and the average grayscale value;

If the deviation is greater than the deviation threshold, the pixel to be identified is determined to be a foreground pixel; otherwise, the pixel to be identified is determined to be a background pixel;

Determine the foreground area according to the foreground pixels, and scan the background pixels in the foreground area;

Holes in the foreground area are determined based on the scanning result, and the holes are filled.

Optionally, the image processing module is used to:

Traversing the pixels in the foreground area, and determining the at least one connected area based on the adjacency relationship of each pixel;

According to the area and rectangularity of each of the connected regions, determining a connected region as the target connected region;

The edge curve of the target connected area is fitted using a three-point interpolation method to obtain a contour point set corresponding to the target connected area.

Optionally, the angle detection module is used to:

Preprocessing the contour point set, and performing initial fitting on the preprocessed contour point set using a least squares fitting method to obtain a fitting rectangle;

Residuals from each contour point in the preprocessed contour point set to the fitting rectangle are calculated respectively, and the residuals are weighted;

The weighted residual is refitted to obtain a new fitting rectangle, and the process returns to the step of respectively calculating the residual from each contour point in the preprocessed contour point set to the fitting rectangle, until the preset termination iteration condition is reached, and the current fitting rectangle is determined to be the circumscribed minimum rectangle.

Optionally, the weight corresponding to the residual is negatively correlated with the residual.

Optionally, the device further includes a camera parameter acquisition module, which is used to:

The target camera is used to take a plurality of calibration plate images at different angles, and the target camera is calibrated based on the calibration plate images to obtain the camera parameters, wherein the camera parameters include intrinsic parameters, extrinsic parameters and distortion parameters.

According to another aspect of the present application, a medium is provided, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the above-mentioned method for enhanced detection of steel turning angles of medium and thick plates is implemented.

According to another aspect of the present application, a device is provided, including a storage medium and a processor, wherein the storage medium stores a computer program, and the processor implements the above-mentioned method for enhanced detection of steel turning angles of medium and thick plates when executing the computer program.

With the help of the above-mentioned technical scheme, this application is based on machine vision technology to collect real-time images of steel billets, and based on camera calibration and distortion correction, image defogging, grayscale linear transformation, dynamic threshold segmentation, filling, opening operation, feature screening, sub-pixel edge detection and external minimum rectangle fitting, develops a steel billet angle detection enhancement algorithm for the steel turning process based on machine vision technology. It can realize real-time extraction of the angles of the steel billets before and after the machine, can adapt to complex environmental changes in the production process, and provide real-time and stable angle detection feedback values for the automatic steel turning control system.

The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

BRIEF DESCRIPTION OF THE DRAWINGS

The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

FIG1 is a schematic diagram showing a flow chart of a method for enhanced detection of steel turning angle of medium and thick plates provided in an embodiment of the present application;

FIG2 is a schematic diagram showing a flow chart of another method for enhancing detection of steel turning angle of medium and thick plates provided in an embodiment of the present application;

FIG3 shows a schematic flow chart of another method for enhanced detection of steel turning angle of medium and thick plates provided in an embodiment of the present application;

FIG4 is a schematic diagram showing a flow chart of another method for enhanced detection of steel turning angle of medium and thick plates provided in an embodiment of the present application;

FIG5 shows a schematic flow chart of a fifth method for enhanced detection of steel turning angles of medium and thick plates provided in an embodiment of the present application;

FIG6 shows a schematic flow chart of a sixth method for enhanced detection of steel turning angles of medium and thick plates provided in an embodiment of the present application;

FIG. 7 shows a schematic flow chart of a seventh method for enhanced detection of steel turning angles of medium and thick plates provided in an embodiment of the present application;

FIG8 shows a schematic diagram of a dynamic threshold segmentation provided in an embodiment of the present application;

FIG9 shows a filling schematic diagram provided in an embodiment of the present application;

FIG10 shows a schematic diagram of an opening operation provided by an embodiment of the present application;

FIG11 shows a schematic diagram of sub-pixel edge detection provided by an embodiment of the present application;

FIG12 shows a schematic diagram of an external minimum matrix fitting provided in an embodiment of the present application;

FIG13 shows a schematic diagram of a real-time angle of a steel billet provided in an embodiment of the present application;

FIG. 14 shows a structural block diagram of a device for enhanced detection of steel turning angles of medium and thick plates provided in an embodiment of the present application.

DETAILED DESCRIPTION

The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

In this embodiment, a method for enhancing the detection of the steel turning angle of a medium and thick plate is provided, specifically a method for enhancing the detection of the steel turning angle of a medium and thick plate, as shown in FIG1 , the method comprises:

Step 101 , in the steel transfer process of medium and thick plate rolling, a target camera is used to capture a steel billet image, and distortion correction is performed on the steel billet image according to camera parameters of the target camera.

Specifically, rolling refers to the process of passing metal through a pair of rollers and forming it by rolling, and the steel transfer link is an important link in the rolling process, which refers to the rotation of the metal plate between different rolling directions and positions. The medium and thick plate steel transfer angle enhancement detection method provided in the embodiment of the present application is used to analyze the steel billet image collected by the target camera in real time during the steel transfer link of medium and thick plate rolling to obtain the steel billet angle, so as to maintain the correct rolling direction and position in the subsequent rolling process, and realize the automatic control of the medium and thick plate steel transfer operation.

In this step, since the position of the target camera may not be parallel to the billet, the billet image it captures is distorted, which affects the accurate measurement of the billet angle. Therefore, it is necessary to first recalculate the pixel coordinates in the billet image based on camera parameters such as distortion parameters, and perform distortion correction on the image to eliminate the impact of distortion.

Step 102, performing defogging and contrast enhancement processing on the distortion-corrected billet image in sequence.

In this step, since atmospheric scattering will cause the billet image to blur and reduce the image quality, the billet image after distortion correction can be defogged to make the image clearer and the details more specific, thereby obtaining a more accurate angle detection result. For example, the details of the image can be restored by analyzing the degree of haze in the image. Specifically, based on the atmospheric light model, the effect of haze on the brightness and contrast of the billet image is utilized, and the haze is removed by transmittance estimation. After defogging, contrast enhancement processing can also be performed. By increasing the contrast of the billet image, the position with high brightness is brighter and the position with low brightness is darker, so as to facilitate the segmentation of the position of the billet.

Step 103, based on the grayscale of each pixel in the steel billet image after contrast enhancement processing, identify the foreground area and the background area, and fill the holes in the foreground area.

In this step, according to the grayscale of each pixel in the steel billet image after contrast enhancement, that is, according to the brightness of different positions in the steel billet image after contrast enhancement, the foreground area with higher brightness and the background area with lower brightness are identified in the image, and the position corresponding to the steel billet is in the foreground area. In addition, since there may be some pixels with lower brightness in the foreground area, the holes in the foreground area are further filled to obtain a more complete foreground area.

Step 104, performing an image morphological opening operation on the steel billet image after the holes are filled.

In this step, the steel billet image after contrast enhancement processing, wherein the opening operation is an operation of first corrosion operation and then expansion operation. Corrosion is to slide the rectangular structure kernel over the image, and compare the pixels corresponding to the binary image with the pixels of the structural element, and the intersection obtained is the image pixel after corrosion. Through the corrosion operation, the independent steel billet elements are interconnected and separated from the background elements, and the small noise objects in the background area are eliminated. Dilation is to slide the rectangular structure kernel over the image, compare the pixels of the binary image with the pixels of its corresponding structural element, and the union obtained after the comparison is completed is the image pixel after expansion. Dilation is used to fill the image pixel gaps left by excessive corrosion in the corrosion operation. The opening operation can remove some isolated, small points and rough edge lines, eliminate burrs and narrow connections, and retain the original area of most areas. The calculation formula of the opening operation is as follows: In the formula, Indicates that A is opened using the structural element B. It means A is corroded by B. express Inflated by B.

Step 105, determining a target connected area based on the connectivity relationship of the foreground area in the steel billet image after the opening operation, and fitting the edge contour of the target connected area.

In this step, considering that there may be two foreground areas connected in the diagonal direction, these two discrete foreground areas should be connected in series. Therefore, in the billet image after the opening operation, the connected area is scanned based on the neighborhood of each pixel point in the foreground area to determine the target connected area corresponding to the billet, and the edge contour of the target connected area is fitted to obtain the contour of the billet.

Step 106 , fitting a circumscribed minimum rectangle corresponding to the target connected area according to the edge contour, and determining the angle of the steel billet according to an offset angle of the rectangle relative to a preset reference direction.

In this step, since the target connected area obtained in the previous step may be an irregular shape, and the steel billet is usually rectangular, the smallest rectangular shape that can contain the target connected area is obtained by fitting the circumscribed minimum rectangle. At this time, the offset angle of the rectangle relative to the preset reference direction is the steel billet angle.

This embodiment is based on machine vision technology to collect real-time images of steel billets, and has developed a steel billet angle detection enhancement algorithm for the steel turning process based on machine vision technology based on camera calibration and distortion correction, image defogging, grayscale linear transformation, dynamic threshold segmentation, filling, opening operation, feature screening, sub-pixel edge detection and external minimum rectangle fitting. The algorithm can realize real-time extraction of the angles of the steel billets before and after the machine, can adapt to complex environmental changes in the production process, and provide real-time and stable angle detection feedback values for the automatic steel turning control system.

Further, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, another method for enhancing the detection of the steel turning angle of medium and thick plates is provided. The method further defines the content of "defogging and contrast enhancement processing are sequentially performed on the steel billet image after distortion correction", as shown in Figure 2, including the following steps:

Step 201, determining the atmospheric light component corresponding to the distortion-corrected billet image, determining the perspective of each pixel in the distortion-corrected billet image according to the atmospheric light component, and performing defogging on the distortion-corrected billet image according to the atmospheric light component and the perspective;

Step 202, stretching the gray value of each pixel in the steel billet image after defogging by a linear transformation method to obtain the steel billet image after contrast enhancement.

In this embodiment, the atmospheric scattering model is first used for defogging. In the field of machine vision, the atmospheric scattering model can be used for image defogging. The formula is: .in, represents a foggy image, represents a haze-free image, is the transmittance, and A represents the global atmospheric light value.

In the specific application process, for each pixel in the distortion-corrected billet image, the smallest pixel value in the three RGB channels is first selected as the grayscale value, and then the real image is subjected to rectangular morphological corrosion to generate a dark channel image. Among them, image morphological operation refers to a collection of a series of image processing operations based on shape, with four basic operations: corrosion, expansion, opening operation, and closing operation. Among them, corrosion is to remove some burrs and details of the image. Corrosion can generally be used to eliminate noise, segment independent image elements, etc., which is essentially a spatial filter.

It is understandable that in fog-free non-pure white areas, one of the three RGB channels usually has a very low pixel value, close to 0, while in foggy images, the dark channel area is much greater than 0 and appears gray. Therefore, in the generated dark channel image, the foggy area can be determined according to the brightness of each pixel. Based on this, after generating the dark channel image, the brightest 0.1% of pixels are extracted according to the brightness in the dark channel image, and the value of the point with the highest brightness at the corresponding position in the original foggy image is found as the atmospheric light value, that is, the atmospheric light component A.

According to the atmospheric light component, the perspective of each pixel in the original image is estimated. The formula for calculating the transmittance is: .

In the formula, ω is taken as 0.95. represents a window centered at pixel x.

When the transmittance t is too small, J will be too large. Therefore, it is necessary to set a lower limit t ₀ for t, and its value is 0.1. season , the final haze-free image It can be expressed as the following formula:

Based on the above formula, the steel billet image can be dehazed according to the perspective and atmospheric light components to restore the dehazed image.

After dehazing, the grayscale value of each pixel can be changed by linear transformation to improve the image contrast. The linear transformation formula is as follows: Mult +Add. In the formula, Represents the grayscale value of the billet image before grayscale stretching operation. represents the grayscale value of the steel billet image after grayscale stretching, Mult represents the multiplication factor, and Add represents the addend factor. It can be understood that the grayscale value refers to the color depth in a black and white image. The larger the grayscale value, the brighter the pixel. Therefore, without changing the size relationship between the grayscale values of the pixels, widening the gap between the grayscale values can improve the contrast of the image. By selecting appropriate parameters, this embodiment can increase the contrast of the steel billet image, making the background darker and the steel billet brighter.

This embodiment performs dehazing and contrast enhancement processing on the billet image, thereby improving the clarity of the image and making the brightness difference between the billet part and the background part larger, making it easier to segment the billet part in the image, and solving the problem of reduced detection accuracy of the billet angle due to unclear image.

Further, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, another method for enhancing the detection of the steel turning angle of medium and thick plates is provided. The method further defines the content of "based on the grayscale of each pixel point in the steel billet image after contrast enhancement processing, identifying the foreground area and the background area, and filling the holes in the foreground area", as shown in Figure 3, including the following steps:

Step 301, calculating the average gray value of each local window in the steel billet image after contrast enhancement processing, and respectively calculating the deviation between the gray value of each pixel to be identified in the steel billet image after contrast enhancement processing and the average gray value;

Step 302: if the deviation is greater than the deviation threshold, the pixel to be identified is determined to be a foreground pixel; otherwise, the pixel to be identified is determined to be a background pixel;

Step 303, determining a foreground area according to the foreground pixels, and scanning background pixels in the foreground area;

Step 304: determine holes in the foreground area based on the scanning result and fill the holes.

In this embodiment, the foreground and background are identified for the steel billet image after contrast enhancement. Since the steel billet is blocked by water vapor and the surface is bright and dark, it is impossible to find a suitable threshold for the entire image to segment the foreground and background. However, in the local area, there is a relatively obvious contrast between the foreground and the background. Based on this, this embodiment uses a dynamic threshold operator to segment the region of interest, i.e., the foreground area, according to the background grayscale value of the local area.

In the specific application process, the image is first smoothed using the mean filter to calculate the average gray value of the current window. This value is used as the background estimation value of the local area. The billet image after contrast enhancement and the image after smoothing are compared pixel by pixel, and the threshold is dynamically determined by the deviation between the two. According to the dynamic threshold, the area with a gray value higher than the gray value of the smoothed image is segmented. Specifically, if the deviation is greater than the deviation threshold, the pixel to be identified is determined to be a foreground pixel, otherwise the pixel to be identified is determined to be a background pixel, and the foreground area can be determined based on the foreground pixel. For example, the gray value of the pixel in the original image is g0, and in its corresponding local range, the gray value of the image after mean filtering and smoothing is gt, and offset is the deviation threshold, where the deviation threshold represents the acceptable range of the gray value deviation between the original image and the threshold image. The area that satisfies the condition of g0 ≥ gt + Offset can be selected as the area of interest, that is, the foreground area.

After obtaining the foreground area, fill the background pixels in the foreground area to avoid holes in the foreground area. Specifically, starting from the source area input by the user, scan the pixels in the foreground area in a specific order, such as from top to bottom and from left to right. When a background pixel is scanned, all background pixels connected to it are found by searching the connected domain from that point. All pixels in the searched connected domain are marked as the filling area. Continue to scan the next unvisited pixel and repeat the above steps until the entire foreground area is scanned and the foreground area after dynamic threshold segmentation is filled.

This embodiment sequentially performs the operations of taking the foreground (edge), threshold segmenting the edge, removing the background connected area, and filling the holes, so as to segment the area to which the billet may belong in the billet image, and then further processes the area in subsequent steps and detects the angle of the billet.

Further, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, another method for enhancing the detection of the steel turning angle of medium and thick plates is provided. The method further defines the content of "determining the target connected area based on the connectivity relationship of the foreground area in the steel billet image after the opening operation, and fitting the edge contour of the target connected area", as shown in Figure 4, including the following steps:

Step 401, traverse the pixels in the foreground area, and determine at least one connected area based on the adjacency relationship of each pixel;

Step 402, determining a connected region as a target connected region according to the area and rectangularity of each connected region;

Step 403: Fit the edge curve of the target connected area using a three-point interpolation method to obtain a contour point set corresponding to the target connected area.

In this embodiment, the target connected area where the steel billet is located is first further screened out in the foreground area. Specifically, the area is first screened according to the area and rectangularity of the area, and the segmented area is traversed using the 8-neighborhood adjacency relationship, and pixels with the same connectivity relationship are assigned to the same connected area. For each connected area, the pixel area is calculated by counting the number of pixels in the area. The saved connected areas are sorted by pixel area, and a target connected area is selected from the connected areas according to the pixel area as the steel billet feature image. In addition, the pixel area and the rectangularity of the area can also be considered at the same time to obtain the final target connected area.

Then, the target connected area is subjected to curve fitting of the edge contour. Specifically, for the pixel-level edge points , use three-point interpolation to fit the edge curve, take three interpolation points in the X direction , , , the corresponding function value is , , Similarly, take three interpolation points in the Y direction , , , the corresponding function value is , , Substitute the interpolation points selected in the X and Y directions into the interpolation formula to obtain the quadratic interpolation function in the X direction. and quadratic interpolation function in the Y direction , the expression is as follows:

The derivative of the above quadratic curve is 0, which is the sub-pixel coordinate in the X and Y directions. =0, get the coordinates ,make =0, get the coordinates The specific expression is as follows:

After obtaining all the sub-pixel points at the edge position and their coordinates, the sub-pixel edge contour of the image can be obtained by connecting the points.

This embodiment accurately finds the position of the steel billet in the foreground area based on the connectivity relationship of each pixel point, and uses sub-pixel edge detection to perform a more detailed analysis of the image on the basis of the pixel level to obtain more accurate edge position information, so that subsequent image processing tasks can be more accurate and reliable.

Further, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, another method for enhanced detection of steel turning angle of medium and thick plates is provided. The method further defines the content of "fitting the circumscribed minimum rectangle corresponding to the target connected area according to the edge contour", as shown in Figure 5, including the following steps:

Step 501, preprocessing the contour point set, and performing initial fitting on the preprocessed contour point set using a least squares fitting method to obtain a fitting rectangle;

Step 502, respectively calculating the residual from each contour point in the preprocessed contour point set to the fitting rectangle, and weighting the residual;

Step 503, refitting is performed based on the weighted residual to obtain a new fitting rectangle, and the process returns to the step of respectively calculating the residual from each contour point in the preprocessed contour point set to the fitting rectangle, until the preset termination iteration condition is reached, and the current fitting rectangle is determined to be the circumscribed minimum rectangle.

In this embodiment, the circumscribed minimum rectangle corresponding to the target connected area is fitted, and then the billet angle is obtained according to the inclination angle of the rectangle. Specifically, the circumscribed minimum rectangle can be fitted according to the following steps: the input contour point set is preprocessed. Including data processing techniques such as sorting data and removing outliers, the purpose is to reduce the influence of outliers on the fitting results; the preprocessed contour point set is initially fitted using the traditional least squares fitting method; the residual from each contour point to the initial fitting rectangle is calculated, and the residual represents the difference between each point and the fitting rectangle; the Tukey weighting function is applied to weight the residual; according to the weighted residual, the rectangle is refitted to obtain a more robust fitting result. This process requires multiple iterations to further optimize the fitting results. In each iteration, the residual is recalculated according to the weighting function, and the rectangle is refitted. Until the final circumscribed minimum rectangle is obtained, the angle between the major axis of the rectangle and the x-axis is the billet angle.

The minimum circumscribed rectangle obtained by fitting the target connected area in this embodiment has certain robustness and stability, and has a certain tolerance for changes in the shape, rotation or partial occlusion of the target connected area. Compared with the outline of the target connected area, the minimum circumscribed rectangle representation is more stable and is not easily disturbed by noise, incomplete edges or shape changes. In addition, the shape of the minimum circumscribed rectangle is more regular, so the position and shape of the billet can be clearly identified, which is more conducive to obtaining the billet angle.

Further, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, another method for enhancing the detection of the steel turning angle of a medium and thick plate is provided. The method further defines that before the step of "correcting the distortion of the steel billet image according to the camera parameters of the target camera", it also includes the content of "obtaining the camera parameters", as shown in FIG6, including the following steps:

Step 601: Use a target camera to take multiple calibration plate images at different angles, and calibrate the target camera based on the calibration plate images to obtain camera parameters, wherein the camera parameters include intrinsic parameters, extrinsic parameters, and distortion parameters.

In this embodiment, a CCD (Charge Coupled Device) camera installed obliquely above the roller takes 20 calibration plate images at different angles, and the camera is calibrated based on the calibration plate images to calculate the camera parameters, such as the camera internal parameters, external parameters and distortion parameters, and then the camera parameters can be used to correct the distortion of the billet image. The calibration plate can be a dot calibration plate or a checkerboard calibration plate, etc. During the calibration process, the Zhang Zhengyou calibration method can be used.

FIG. 7 shows a method for enhanced detection of steel turning angle of medium and thick plates provided by another embodiment of the present application. As shown in the figure, the method comprises the following steps:

Step 701: camera calibration and distortion correction.

In this step, the Zhang Zhengyou calibration method is used to determine the camera parameters and correct the distortion, and the dot calibration plate is used as the calibration object. The dot calibration plate consists of a 7×7 dot array and a square frame with cut corners. The calibration plate size is 800mm×800mm, each dot has a diameter of 50mm, and the distance between adjacent dots is 100mm. Each dot and the frame are used as feature reference points. The Zhang Zhengyou calibration method is used to compare the feature point data in 20 pictures to correct the internal and external parameters of the camera.

The calibration results are shown in Table 1-3 below:

Table 1 Camera internal parameters

Pixel width/μm Pixel height/μm Focal length/mm Center x coordinate/Pixel Center y coordinate/Pixel Image width/Pixel Pixel 8.29862 8.3 11.6503 359.379 235.82 659 494

Table 2 Camera external parameters

X-direction translation/mm Y direction translation/mm Z direction translation/mm X-axis rotation Y direction rotation/° Z direction rotation/° 6.60547 1.93844 425.423 3.55931 354.219 180.613

Table 3 Camera distortion coefficients

K1(1/m2) K2(1/m4) K3(1/m6) P1(1/m2) P2(1/m2) 0.2093 -0.7269 0.2962 0.1316 0.0632

Step 702: image dehazing processing.

In this step, the dark channel image is obtained, the perspective and atmospheric light components are calculated, and the atmospheric scattering model is substituted to perform dehazing on the original image.

Step 703: grayscale linear transformation.

In this step, the multiplication factor Mult is selected as 5 and the addend factor Add is selected as -150, so that the grayscale values of pixels below 80 can be reduced to 0 and the grayscale values of pixels above 80 can be converted to 250, thereby greatly improving the contrast of the billet image.

Step 704: dynamic threshold segmentation.

In this step, the mean filter mask width is set to 230 and the height is set to 230. The offset value in dynamic threshold segmentation is 12, and the extraction mode is "light mode", that is, the area of the original image that is brighter than the image after mean filtering by 12 gray levels is extracted. The image processed by dynamic threshold segmentation is shown in Figure 8.

Step 705, filling.

In this step, the holes in the segmented area are filled to make the area more continuous and complete, which is convenient for subsequent analysis and processing. The image after filling is shown in Figure 9.

Step 706, start operation.

The width and height of the rectangular structure kernel in the opening operation are set to 20 and 20, respectively. The image processed by the opening operation is shown in Figure 10.

Step 707, feature screening.

In this step, the steel billet is screened according to its basic area range and rectangularity. The minimum area of the steel billet is set to 18000 and the maximum area is set to 400000 according to the actual production situation. The minimum rectangularity is 0.8 and the maximum is 1.0. The area is screened according to the above conditions, and the screened connected domains are sorted by pixel area. The connected region with the largest pixel area is selected as the steel billet feature image.

Step 708: sub-pixel edge detection.

In this step, the sub-pixel coordinates are calculated using the polynomial interpolation method, and the image processed by the sub-pixel edge detection algorithm is shown in FIG11 .

Step 709: fitting the minimum circumscribed rectangle.

The least squares method finds a straight line by fitting data points. Its basic principle is to find the minimum sum of squares of the distances between all points and the fitted straight line. Assume that the expression of the straight line is: , then the sum of square errors is:

In some cases where the edge noise is relatively large, the data contains a large number of noise points, and the traditional least squares method cannot obtain the optimal result. Therefore, the present invention adopts the weighted least squares method for fitting. The straight line fitting with the weight function takes into account the distance from different points to the straight line, reduces the fitting error, and can improve the fitting accuracy. The weight function based on Tukey weighting function is defined as:

In the formula, represents the distance from a point to a line, It is the clipping coefficient, which indicates the distance. Its function is to judge the outliers.

when When , the value of the weighting function is 0, indicating that larger residuals are given smaller weights. When , the value of the weighting function changes with the distance in the interval (0,1). The smaller the distance from the point to the line, the larger the value of the weighting function, which means that a smaller residual is given a larger weight. The standard deviation is adaptively set according to a one-dimensional Gaussian distribution.

Under the condition of weighted least squares method, the expression for minimizing the residual error is:

When the residual is minimum, the parameters of the straight line equation are obtained.

Fitting based on Tukey weight function can eliminate the influence of outliers and obtain more accurate fitting results. The image processed by the circumscribed minimum rectangle fitting algorithm is shown in Figure 12. The angle between the major axis of the rectangle and the x-axis (-89.45°) is the real-time angle of the billet, as shown in Figure 13.

It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

Further, as a specific implementation of the above-mentioned medium and thick plate steel turning angle enhancement detection method, the embodiment of the present application provides a medium and thick plate steel turning angle enhancement detection device, as shown in FIG14, the device includes: an image capture module, an image processing module and an angle detection module, wherein:

An image capture module is used to capture the image of the steel billet through a target camera in the steel transfer process of medium and thick plate rolling, and to perform distortion correction on the steel billet image according to the camera parameters of the target camera;

An image processing module is used to sequentially perform defogging and contrast enhancement processing on the billet image after distortion correction; and, based on the grayscale of each pixel in the billet image after contrast enhancement, identify the foreground area and the background area, and fill the holes in the foreground area; and, perform image morphological opening operation on the billet image after the hole filling; and, based on the connectivity relationship of the foreground area in the billet image after the opening operation, determine the target connected area, and fit the edge contour of the target connected area;

The angle detection module is used to fit the circumscribed minimum rectangle corresponding to the target connected area according to the edge contour, and determine the angle of the steel billet according to the offset angle of the rectangle relative to the preset reference direction.

Optionally, the image processing module is used to:

Determine the atmospheric light component corresponding to the distortion-corrected billet image, determine the perspective of each pixel in the distortion-corrected billet image according to the atmospheric light component, and perform defogging on the distortion-corrected billet image according to the atmospheric light component and the perspective;

The gray value of each pixel in the dehazed billet image is stretched by a linear transformation method to obtain a contrast-enhanced billet image.

Optionally, the image processing module is used to:

Calculate the average gray value of each local window in the steel billet image after contrast enhancement processing, and respectively calculate the deviation between the gray value of each pixel to be identified in the steel billet image after contrast enhancement processing and the average gray value;

If the deviation is greater than the deviation threshold, the pixel to be identified is determined to be a foreground pixel, otherwise the pixel to be identified is determined to be a background pixel;

Determine the foreground area according to the foreground pixels, and scan the background pixels in the foreground area;

Holes in the foreground area are determined based on the scan results and filled.

Optionally, the image processing module is used to:

Traversing the pixels in the foreground area, and determining at least one connected area based on the adjacency relationship of each pixel;

According to the area and rectangularity of each connected region, a connected region is determined as the target connected region;

The edge curve of the target connected area is fitted using the three-point interpolation method to obtain the contour point set corresponding to the target connected area.

Optionally, the angle detection module is used to:

Preprocessing the contour point set, and using the least squares fitting method to perform initial fitting on the preprocessed contour point set to obtain a fitting rectangle;

Calculate the residual from each contour point in the preprocessed contour point set to the fitting rectangle, and weight the residual;

The weighted residual is refitted to obtain a new fitting rectangle, and the process returns to the step of calculating the residual from each contour point in the preprocessed contour point set to the fitting rectangle, until the preset termination condition is reached and the current fitting rectangle is determined to be the minimum circumscribed rectangle.

Optionally, the weight corresponding to the residual is negatively correlated with the residual.

Optionally, the device further includes a camera parameter acquisition module, which is used to:

The target camera is used to take multiple calibration plate images at different angles, and the target camera is calibrated based on the calibration plate images to obtain camera parameters, where the camera parameters include intrinsic parameters, extrinsic parameters, and distortion parameters.

According to another aspect of the present application, a medium is provided, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the above-mentioned method for enhanced detection of steel turning angles of medium and thick plates is implemented.

It should be noted that for other corresponding descriptions of the functional modules involved in the medium and thick plate steel turning angle enhancement detection device provided in the embodiment of the present application, reference can be made to the corresponding descriptions in the above method and will not be repeated here.

Based on the above method, accordingly, an embodiment of the present application further provides a storage medium on which a computer program is stored, and when the program is executed by a processor, the above method for enhanced detection of steel turning angles of medium and thick plates is implemented.

Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), and includes a number of instructions for enabling an electronic device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present application.

Based on the above-mentioned method as shown in Figures 1 to 13, and the virtual device embodiment shown in Figure 14, in order to achieve the above-mentioned purpose, the embodiment of the present application also provides a device, which can be specifically a personal computer, a server, a network device, etc. The electronic device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to implement the above-mentioned method for enhanced detection of medium and thick plate turning angles as shown in Figures 1 to 13.

Optionally, the electronic device may further include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a Bluetooth interface, a WI-FI interface), etc.

Those skilled in the art will appreciate that the electronic device structure provided in this embodiment does not limit the electronic device, and may include more or fewer components, or a combination of certain components, or different component arrangements.

The storage medium may also include an operating system and a network communication module. The operating system is a program that manages and saves the hardware and software resources of the electronic device, and supports the operation of information processing programs and other software and/or programs. The network communication module is used to realize communication between the various controls inside the storage medium, and communication with other hardware and software in the physical device.

Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform, or by hardware.

Those skilled in the art will appreciate that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the units or processes in the accompanying drawings are not necessarily necessary for implementing the present application. Those skilled in the art will appreciate that the units in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the description of the implementation scenario, or can be changed accordingly and located in one or more devices different from the present implementation scenario. The units of the above-mentioned implementation scenarios can be combined into one unit, or can be further split into multiple sub-units.

The above serial numbers of this application are only for description and do not represent the advantages and disadvantages of the implementation scenarios. The above disclosure is only a few specific implementation scenarios of this application, but this application is not limited to them, and any changes that can be thought of by technicians in this field should fall within the scope of protection of this application.

Claims (10)

1. A method for enhancing the detection of steel turning angle of medium and thick plates, characterized in that the method comprises: In the steel transfer process of medium and thick plate rolling, a target camera is used to capture a steel billet image, and distortion correction is performed on the steel billet image according to camera parameters of the target camera; The billet image after distortion correction is subjected to defogging and contrast enhancement processing in sequence; Based on the grayscale of each pixel in the steel billet image after contrast enhancement processing, a foreground area and a background area are identified, and holes in the foreground area are filled; Perform image morphological opening operation on the billet image after hole filling; Determine a target connected area based on the connectivity relationship of the foreground area in the steel billet image after the opening operation, and fit the edge contour of the target connected area; The minimum circumscribed rectangle corresponding to the target connected area is fitted according to the edge contour, and the angle of the steel billet is determined according to the offset angle of the rectangle relative to a preset reference direction. 2. The method according to claim 1, characterized in that the step of sequentially performing defogging and contrast enhancement processing on the distortion-corrected billet image comprises: Determine an atmospheric light component corresponding to the distortion-corrected steel billet image, determine the perspective of each pixel in the distortion-corrected steel billet image according to the atmospheric light component, and perform defogging on the distortion-corrected steel billet image according to the atmospheric light component and the perspective; The gray value of each pixel in the dehazed billet image is stretched by a linear transformation method to obtain a contrast-enhanced billet image. 3. The method according to claim 1, characterized in that the foreground area and the background area are identified based on the grayscale of each pixel in the steel billet image after contrast enhancement processing, and the holes in the foreground area are filled, comprising: Calculating the average grayscale value of each local window in the steel billet image after contrast enhancement processing, and respectively calculating the deviation between the grayscale value of each pixel to be identified in the steel billet image after contrast enhancement processing and the average grayscale value; If the deviation is greater than the deviation threshold, the pixel to be identified is determined to be a foreground pixel; otherwise, the pixel to be identified is determined to be a background pixel; Determine the foreground area according to the foreground pixels, and scan the background pixels in the foreground area; Holes in the foreground area are determined based on the scanning result, and the holes are filled. 4. The method according to claim 1, characterized in that the step of determining the target connected area based on the connectivity relationship of the foreground area in the steel billet image after the opening operation and fitting the edge contour of the target connected area comprises: Traversing the pixels in the foreground area, and determining the at least one connected area based on the adjacency relationship of each pixel; According to the area and rectangularity of each of the connected regions, determining a connected region as the target connected region; The edge curve of the target connected area is fitted using a three-point interpolation method to obtain a contour point set corresponding to the target connected area. 5. The method according to claim 1, characterized in that the step of fitting the circumscribed minimum rectangle corresponding to the target connected area according to the edge contour comprises: Preprocessing the contour point set, and performing initial fitting on the preprocessed contour point set using a least squares fitting method to obtain a fitting rectangle; Residuals from each contour point in the preprocessed contour point set to the fitting rectangle are calculated respectively, and the residuals are weighted; The weighted residual is refitted to obtain a new fitting rectangle, and the process returns to the step of respectively calculating the residual from each contour point in the preprocessed contour point set to the fitting rectangle, until the preset termination iteration condition is reached, and the current fitting rectangle is determined to be the circumscribed minimum rectangle. 6. The method according to claim 5, characterized in that the weight corresponding to the residual is negatively correlated with the residual. 7. The method according to claim 1, characterized in that before the distortion correction of the steel billet image according to the camera parameters of the target camera is performed, it comprises: The target camera is used to take a plurality of calibration plate images at different angles, and the target camera is calibrated based on the calibration plate images to obtain the camera parameters, wherein the camera parameters include intrinsic parameters, extrinsic parameters and distortion parameters. 8. A device for detecting the enhanced steel turning angle of a medium and thick plate, characterized in that the device comprises: An image shooting module, used to shoot a steel billet image through a target camera in the steel transfer link of medium and thick plate rolling, and perform distortion correction on the steel billet image according to the camera parameters of the target camera; An image processing module is used to sequentially perform defogging and contrast enhancement processing on the steel billet image after distortion correction; and, based on the grayscale of each pixel in the steel billet image after contrast enhancement, identify the foreground area and the background area, and fill the holes in the foreground area; and, perform image morphological opening operation on the steel billet image after the hole filling; and, based on the connectivity relationship of the foreground area in the steel billet image after the opening operation, determine the target connected area, and fit the edge contour of the target connected area; The angle detection module is used to fit the circumscribed minimum rectangle corresponding to the target connected area according to the edge contour, and determine the angle of the steel billet according to the offset angle of the rectangle relative to a preset reference direction. 9. A storage medium having a program or instruction stored thereon, wherein the program or instruction, when executed by a processor, implements the method according to any one of claims 1 to 7. 10. An electronic device comprising a storage medium and a processor, wherein the storage medium stores a computer program, and the processor implements the method according to any one of claims 1 to 7 when executing the computer program.