Intelligent identification method of plate rolling machine fault based on machine vision

Abstract

The present invention relates to the field of graphic recognition technology, and specifically to a method for intelligent identification of plate rolling machine faults based on machine vision. The method uses an image acquisition device to acquire images of plates processed by the plate rolling machine. The range of the center and radius of the Hough circle detection is determined according to the scene, and then the Hough circle transform is used to detect whether a circle exists. The Hough circle detection result is used to determine whether the plate rolling machine has a fault. The present invention uses the Hough circle detection method to determine the state of the plate processed by the plate rolling machine to determine whether the plate rolling machine has a fault. In the traditional Hough circle detection algorithm, each point in the image needs to be used as the center point of the circle, and the image length is used as the radius. The traversal is repeated over and over again, and the calculation cost is very high. The present invention determines the range of the center and radius by analyzing the scene, narrows the range of the Hough circle detection, and improves the calculation efficiency.

Description

Technical Field

[0001] The invention relates to the technical field of graphic recognition, and in particular to a plate rolling machine fault intelligent recognition method based on machine vision. Background Art

[0002] Plate rolling machine is a kind of equipment that uses working rollers to bend and shape sheet materials. It can be used to form parts of different shapes such as cylindrical parts and conical parts. It is a very important processing equipment. Its principle is to use external forces such as hydraulic pressure and mechanical force to move the working rollers, so as to bend or roll the sheet materials. With the rapid development of my country's economic construction, the demand for plate rolling machines is increasing, and the technical requirements are getting higher and higher. If the plate rolling machine fails when rolling the sheet, such as the rolling force or angle is incorrect, the sheet will not become a closed circle as required or there will be edges and corners around it. Therefore, it can be judged whether the plate rolling machine has a fault according to the state of the sheet after curling.

[0003] The existing judgment method can use machine vision to identify the circular area in the cross-sectional image of the plate after the plate rolling machine processes the plate, and judge whether a fault occurs based on the parameters of the circular area. However, the existing method is only applicable when it is determined that there is a circle in the image. If it is impossible to determine whether the image contains a circle, then the method is inaccurate. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a method for intelligent identification of plate rolling machine faults based on machine vision. The technical solutions adopted are as follows:

[0005] The present invention proposes a plate rolling machine fault intelligent identification method based on machine vision, the method comprising:

[0006] Obtain a cross-sectional image of the plate after being processed by the plate rolling machine; obtain edge information of the cross-sectional image, and obtain an edge image; the edge image includes an inner edge and an outer edge of the plate;

[0007] Determine four first representative points of the inner edge and four second representative points of the outer edge according to the maximum and minimum horizontal coordinates and the maximum and minimum vertical coordinates of the edge points on the inner edge and the outer edge; the first representative points constitute a first rectangular area and the second representative points constitute a second rectangular area;

[0008] Determine the thickness of the plate according to the sizes of the first rectangular area and the second rectangular area; take each pixel point in the second rectangular area as the initial center point, and make a circle with the thickness of the plate as the radius to obtain a test circle; take each pixel point in the second rectangular area as the starting point, make a ray along the direction of the starting point and an edge point, if the ray contains two pixel points, then the ray is considered to be a test ray, and obtain the first distance and the second distance from the starting point to the two pixel points on the test ray respectively; obtain the distance difference between the first distance and the second distance, obtain the similarity between the distance difference and the thickness of the plate, and obtain the position fluctuation factor of the pixel point corresponding to the starting point according to all similarities corresponding to the starting point of the test ray;

[0009] If there is no edge point in the test circle, the pixel point corresponding to the initial center point is the first candidate center point; the center confidence is obtained according to the position fluctuation factor of the first candidate center point; the second candidate center point is selected according to the center confidence; the two radii of the second candidate center point are determined according to the first representative point and the second representative point, and a circle is made to obtain a reference circle, and a detection circle is selected according to the number of edge points on the reference circle. If there is no detection circle, it means that the plate rolling machine is faulty.

[0010] Furthermore, the first representative points forming a first rectangular area include:

[0011] Select the four pixel points with the largest and smallest horizontal coordinates and the largest and smallest vertical coordinates from the coordinates of all edge points. According to the coordinates of these four pixel points, draw a horizontal straight line through the two pixel points with the largest and smallest horizontal coordinates, and draw a vertical straight line through the two pixel points with the largest and smallest vertical coordinates. These four straight lines constitute the first rectangular area.

[0012] Furthermore, the method for obtaining the second representative point includes:

[0013] Draw a vertical line for the pixel points with the largest and smallest horizontal coordinates respectively, and extract the two pixel points closest to the two points at the point where the vertical line intersects the edge point. Draw a horizontal line through the two pixel points with the largest and smallest vertical coordinates, and extract the two pixel points closest to the two points in the horizontal line to obtain the second representative point.

[0014] Furthermore, determining the thickness of the plate according to the sizes of the first rectangular area and the second rectangular area includes:

[0015]

[0016] Where u is the thickness of the plate, (a 1 ,b 1 ) is the length and width of the first rectangular area, (a 2 ,b 2 ) are the length and width of the second rectangular area.

[0017] Furthermore, the distance difference between the first distance and the second distance is obtained, and the similarity between the distance difference and the plate thickness is obtained, and the position fluctuation factor of the pixel point corresponding to the starting point is obtained according to all similarities of the starting point of the corresponding test ray, including:

[0018]

[0019] Among them, s v is the position fluctuation factor of the vth starting point, (i v , j v ) is the position coordinate of the vth starting point, (i1 z ,j1 z ) is the coordinate of the pixel point closest to the starting point on the z-th test ray, (i2 z ,j2 z ) is the coordinate of the pixel point on the z-th test ray that is the second closest to the starting point, t is the number of test rays corresponding to the v-th starting point, and u is the thickness of the plate.

[0020] Further, obtaining the circle center confidence according to the position fluctuation factor of the first candidate circle center point includes:

[0021]

[0022] Among them, τ v is the confidence of the center of the vth pixel, s v is the position fluctuation factor of the vth pixel, and e is a natural constant.

[0023] The present invention has the following beneficial effects:

[0024] The embodiment of the present invention uses the Hough circle detection method to determine the state of the plate after being processed by the plate rolling machine to determine whether the plate rolling machine has a fault. In the traditional Hough circle detection algorithm, each point in the image needs to be used as the center point of the circle, and the image length is used as the radius, and the calculation cost is very high. The embodiment of the present invention determines the range of the center point and the radius by analyzing the scene, thereby narrowing the range of the Hough circle detection and improving the calculation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0026] Figure 1A flow chart of a method for intelligently identifying plate rolling machine faults based on machine vision provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a plate rolling machine fault intelligent identification method based on machine vision proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0028] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0029] The specific scheme of the intelligent identification method of plate rolling machine fault based on machine vision provided by the present invention is described in detail below with reference to the accompanying drawings.

[0030] See also Figure 1 , which shows a flow chart of a method for intelligently identifying plate rolling machine faults based on machine vision provided by an embodiment of the present invention, the method comprising:

[0031] Step S1: obtaining a cross-sectional image of a plate after being processed by a plate rolling machine; obtaining edge information of the cross-sectional image, and obtaining an edge image; the edge image includes an inner edge and an outer edge of the plate.

[0032] The plate rolling machine is a machine that rolls the plate into a cylindrical object. After the plate is rolled, the processed plate is transported to the material presentation area through the transport channel, and the image acquisition device is used to collect the cross-sectional image of the processed plate at this time. In order to facilitate the subsequent image calculation in the embodiment of the present invention, the cross-sectional image is grayed to obtain a grayscale image of the processed plate.

[0033] In an embodiment of the present invention, a cylindrical plate is obtained after being rolled by a plate rolling machine, and the collected image is an image of its cross section. If the plate rolling machine fails, the rolling roller of the plate rolling machine cannot move in the preset angle direction, which may cause the processed plate to fail to form a cylinder or have an uneven surface. This is reflected in the cross-sectional image as gaps in the circumference of the cross-sectional circle in the image, a lack of fit, straight lines on the circumference, or changes in the curling angle of the plate.

[0034] Based on the above problems, an embodiment of the present invention uses Hough circle transform to determine whether the above situation occurs by detecting circular edges. Before using the Hough circle transform, the grayscale image must first be converted into an edge image through edge detection. Since the Hough circle detection is sensitive to noise, it is necessary to use median filtering to filter the image to achieve denoising.

[0035] After edge detection obtains edge points, each edge pixel in the rectangular coordinate system can be placed on a circle composed of multiple center points and radii. Therefore, each edge pixel can correspond to a three-dimensional curve composed of radius and center point. A threshold is set. When the number of curve intersections is greater than the threshold, it is considered that there is a circle.

[0036] The existing improvement on the amount of calculation is to calculate the gradient based on the circle, and the gradient calculated from multiple directions is directed to a certain point, which may be the center point of the circle. However, this method is only used when it is determined that there is a circle in the image. If it is not possible to determine whether the image contains a circle, then using this method is inaccurate. Therefore, the embodiment of the present invention improves this and determines the position of the center point and the size of the radius according to the scene of the embodiment of the present invention.

[0037] Step S2: Determine four first representative points of the inner edge and four second representative points of the outer edge according to the maximum and minimum horizontal coordinates and the maximum and minimum vertical coordinates of the edge points on the inner edge and the outer edge; the first representative points constitute a first rectangular area and the second representative points constitute a second rectangular area.

[0038] Because the plate has thickness, there are outer edges and inner edges in the edge image. The four first representative points of the inner edge and the four second representative points of the outer edge are determined according to the maximum and minimum horizontal coordinates and the maximum and minimum vertical coordinates of the edge points on the inner edge and the outer edge, respectively, including:

[0039] According to edge detection, the coordinates of each edge pixel point can be obtained. The four pixel points with the largest and smallest horizontal coordinates and the largest and smallest vertical coordinates are selected from the coordinates of all edge pixel points. According to the coordinates of these four pixel points, a horizontal straight line is drawn through the two pixel points with the largest and smallest horizontal coordinates, and a vertical straight line is drawn through the two pixel points with the largest and smallest vertical coordinates. These four straight lines form a rectangular area. The center pixel point must be inside all edge points. Therefore, the center pixel point must be in the first rectangular area.

[0040] In the embodiment of the present invention, the horizontal coordinate and the vertical coordinate refer to the number of rows and columns in the image, that is, the position information of the image. For example, the coordinate (1,1) represents the first row and the first column, and (n,1) represents the first column of the nth row.

[0041] Because the cylinder in the embodiment of the present invention has thickness, there will be another edge point if the 4 pixel points found in the above steps are moved inward. Therefore, for the pixel points with the largest horizontal coordinate and the smallest horizontal coordinate, a vertical line is made for the two points respectively, and the two pixel points closest to the two points are extracted from the point where the vertical line intersects with the edge point. Similarly, a horizontal line is made through the two pixel points with the largest vertical coordinate and the smallest vertical coordinate, and the two pixel points closest to the two points are also extracted from the horizontal line, thereby obtaining 4 new pixel points. These four pixel points also obtain a rectangular area in the above manner, which is smaller than the previous one and is considered to be the second rectangular area, and the center of the circle is also in this area.

[0042] Through the above steps, the approximate area of ​​the center of the circle, i.e., the second rectangular area, is obtained. Then, the confidence level of each pixel in the area can be calculated.

[0043] Step S3: Determine the thickness of the plate according to the sizes of the first rectangular area and the second rectangular area; take each pixel point in the second rectangular area as the initial center point, and make a circle with the thickness of the plate as the radius to obtain a test circle; take each pixel point in the second rectangular area as the starting point, and make a ray along the direction of the starting point and an edge point. If the ray contains two pixel points, the ray is considered to be a test ray, and the first distance and the second distance from the starting point to the two pixel points on the test ray are obtained respectively; obtain the distance difference between the first distance and the second distance, obtain the similarity between the distance difference and the thickness of the plate, and obtain the position fluctuation factor of the pixel point corresponding to the starting point according to all similarities corresponding to the starting point of the test ray.

[0044] Assume that the length and width of the first rectangular area are a 1 ,b 1 The length and width of the second rectangular area are a 2 ,b 2 , and the length and width of the two corresponding subtraction and then divided by 2 can be regarded as the thickness of the plate, calculated as follows:

[0045]

[0046] Where u is the thickness of the plate, (a 1 ,b 1 ) is the length and width of the first rectangular area, (a 2 ,b 2 ) are the length and width of the second rectangular area.

[0047] Each pixel point in the second rectangular area is used as the initial center point, and the thickness of the plate is used as the radius to make a circle to obtain a test circle. First, in the second rectangular area, all the pixels in the area are given a range. Since the center point of the circle cannot be around the edge point, a circle is obtained with each pixel point in the area as the center and u as the radius. The degree of edge separation of the point is determined based on whether there is an edge point in the circle.

[0048]

[0049] where d v It indicates the degree of edge separation of the vth pixel. If d v =1, then it means the separation is low. If d v =0, it means the degree of separation is high.

[0050] Take each pixel point in the second rectangular area as the starting point, draw a ray along the direction of the starting point and an edge point, if the ray contains two pixel points, the ray is considered to be a test ray, and obtain the first distance and the second distance from the starting point to the two pixel points on the test ray respectively; obtain the distance difference between the first distance and the second distance, obtain the similarity between the distance difference and the plate thickness, and obtain the position fluctuation factor of the pixel point corresponding to the starting point according to all similarities corresponding to the starting point of the test ray, which specifically includes:

[0051]

[0052] Among them, s v is the position fluctuation factor of the vth starting point, (i v , j v ) is the position coordinate of the vth starting point, (i1 z ,j1 z ) is the coordinate of the pixel point closest to the starting point on the z-th test ray, (j2 z ,j2 z ) is the coordinate of the pixel point on the z-th test ray that is the second closest to the starting point, t is the number of test rays corresponding to the v-th starting point, and u is the thickness of the plate.

[0053] Step S4: If there is no edge point in the test circle, the pixel point corresponding to the initial center point is the first candidate center point; the center confidence is obtained according to the position fluctuation factor of the first candidate center point; the second candidate center point is selected according to the center confidence; the two radii of the second candidate center point are determined according to the first representative point and the second representative point, and Hough circle detection is performed to obtain a reference circle, and a detection circle is selected according to the number of edge points on the reference circle. If there is no detection circle, it means that the plate rolling machine is faulty.

[0054] That is, select d vThe point with a value of 1 is used as the first candidate center point, and the center confidence is obtained according to the position fluctuation factor of the first candidate center point, which specifically includes:

[0055]

[0056] Among them, τ v is the confidence of the center of the vth pixel, s v is the position fluctuation factor of the vth pixel, and e is a natural constant.

[0057] in the formula The purpose is to make the final result fall within the range of 0-1, with the minimum numerator being 1 and the maximum being e, and the larger the numerator, the higher the confidence level. v It represents the confidence of the center of the circle. The value range of this value is 0-1. The larger the value, the better the confidence of the center of the circle. For this scheme, we choose τ v The pixel points with a confidence level greater than 0.6 are taken as the range of the circle center, that is, the second candidate circle center point for the organizing committee with a circle center confidence level greater than 0.6.

[0058] After obtaining the second candidate center point, we need to get the radius of the Hough circle detection. In the second rectangular area, a point is obtained by connecting the diagonals. The length of the line from this point to the longest of the four diagonals is selected as the maximum radius of the Hough circle detection, and u is used as the minimum radius. In this way, the center point range and radius range of the Hough circle transformation are obtained, and multiple reference circles are obtained.

[0059] The detection circle is screened out according to the number of edge points on the reference circle, and the threshold is set to 80%. That is, when 80% of the pixels on the circumference are edge points, the circumference is considered to be the detected reference circle.

[0060] If two circles are obtained after image detection, it is considered that the plate rolling machine has no faults. Otherwise, it is considered that the plate rolling machine has faults, causing the rolled plate to be unable to maintain a good round shape and is an unqualified product.

[0061] In summary, the embodiment of the present invention uses an image acquisition device to capture images of plates after being processed by a plate rolling machine. The range of the center and radius of the Hough circle detection is determined according to the scene, and then the Hough circle transform is used to detect whether a circle exists. Determine whether the plate rolling machine has a fault based on the Hough circle detection result. The embodiment of the present invention uses the Hough circle detection method to determine the state of the plate after being processed by the plate rolling machine to determine whether the plate rolling machine has a fault. In the traditional Hough circle detection algorithm, each point in the image is required to be used as the center point of the circle, and the image length is used as the radius, and the traversal is repeated over and over again, and the computational cost is very high. The embodiment of the present invention determines the range of the center and radius of the circle by analyzing the scene, narrows the range of the Hough circle detection, and improves the computational efficiency.

[0062] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0063] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0064] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for intelligent identification of plate rolling machine faults based on machine vision, characterized in that: The method comprises: Obtain a cross-sectional image of the plate after being processed by the plate rolling machine; obtain edge information of the cross-sectional image, and obtain an edge image; the edge image includes an inner edge and an outer edge of the plate; Determine four first representative points of the inner edge and four second representative points of the outer edge according to the maximum and minimum horizontal coordinates and the maximum and minimum vertical coordinates of the edge points on the inner edge and the outer edge; the first representative points constitute a first rectangular area, and the second representative points constitute a second rectangular area; Determine the thickness of the plate according to the sizes of the first rectangular area and the second rectangular area; take each pixel point in the second rectangular area as the initial center point, and make a circle with the thickness of the plate as the radius to obtain a test circle; take each pixel point in the second rectangular area as the starting point, make a ray along the direction of the starting point and an edge point, if the ray contains two pixel points, then the ray is considered to be a test ray, and obtain the first distance and the second distance from the starting point to the two pixel points on the test ray respectively; obtain the distance difference between the first distance and the second distance, obtain the similarity between the distance difference and the thickness of the plate, and obtain the position fluctuation factor of the pixel point corresponding to the starting point according to all similarities corresponding to the starting point of the test ray; If there is no edge point in the test circle, the pixel point corresponding to the initial center point is the first candidate center point; the center confidence is obtained according to the position fluctuation factor of the first candidate center point; the second candidate center point is selected according to the center confidence; the intersection of the diagonals of the second rectangular area is determined, and the longest length from the intersection to the four corners of the second rectangular area is used as the maximum radius of the Hough circle detection, the plate thickness is used as the minimum radius of the Hough circle detection, and the second candidate center point is used as the center point of the Hough circle detection to perform Hough circle detection, and obtain a reference circle, select a detection circle according to the number of edge points on the reference circle, and judge whether the plate bending machine has a fault according to the number of detection circles; The method of selecting a detection circle according to the number of edge points on the reference circle and judging whether the plate rolling machine fails according to the number of detection circles includes: setting a threshold value of 80%, and when 80% of the pixel points on the circumference of the reference circle are edge points, the circumference is the detected reference circle; if two detection circles are obtained, it is considered that the plate rolling machine does not fail, otherwise, it is considered that the plate rolling machine fails; Determining the thickness of the plate according to the sizes of the first rectangular area and the second rectangular area comprises: Where u is the thickness of the plate, (a1, b1) is the length and width of the first rectangular area, and (a2, b2) is the length and width of the second rectangular area; The method of obtaining the distance difference between the first distance and the second distance, obtaining the similarity between the distance difference and the plate thickness, and obtaining the position fluctuation factors of the pixel point corresponding to the starting point according to all similarities of the starting point of the corresponding test ray include: Among them, s v is the position fluctuation factor of the vth starting point, (i v , j v ) is the position coordinate of the vth starting point, (i1 z ,j1 z ) is the coordinate of the pixel point closest to the starting point on the z-th test ray, (i2 z ,j2 z ) is the coordinate of the pixel point on the zth test ray that is the second closest to the starting point, t is the number of test rays corresponding to the vth starting point, and u is the thickness of the plate; The step of obtaining the circle center confidence according to the position fluctuation factor of the first circle center point to be selected comprises: Among them, τ v is the confidence of the center of the vth pixel, s v is the position fluctuation factor of the vth pixel, and e is a natural constant; The first representative points forming the first rectangular area include: Select four pixel points with the largest and smallest horizontal coordinates and the largest and smallest vertical coordinates from the coordinates of all edge points. According to the coordinates of these four pixel points, draw a horizontal straight line through the two pixel points with the largest and smallest horizontal coordinates, and draw a vertical straight line through the two pixel points with the largest and smallest vertical coordinates. These four straight lines constitute the first rectangular area. The method for obtaining the second representative point includes: Draw a vertical line for the pixel points with the largest and smallest horizontal coordinates respectively, and extract the two pixel points closest to the two points at the point where the vertical line intersects the edge point. Draw a horizontal line through the two pixel points with the largest and smallest vertical coordinates, and extract the two pixel points closest to the two points in the horizontal line to obtain the second representative point.

Images