A quantitative determination method of elliptical spot image distortion

Abstract

The application provides a quantitative determination method of an elliptical wear scar image distortion variable, comprising the following steps: step 1, segmenting the obtained four-ball friction wear scar image to obtain a wear scar segmentation image; step 2, rotating the obtained wear scar segmentation image to obtain a rotated wear scar image; step 3, constructing an elliptical wear scar image according to the obtained rotated wear scar image; step 4, calculating the matching degree between the rotated segmentation image and the elliptical template image; step 5, calculating the wear scar distortion variable according to the obtained matching degree; and step 6, determining the wear scar abnormality of the obtained four-ball friction wear scar image according to the obtained wear scar distortion variable. The application establishes an evaluation index of the wear scar distortion degree based on multiple image characteristic parameters, realizes the determination of the wear scar shape distortion variable, and changes the subjectivity and unscientific nature of the qualitative determination by a tester.

Description

Technical Field

[0001] The invention belongs to the technical field of image test data analysis and processing, in particular to a method for quantitatively determining the distortion amount of an elliptical wear spot image. Background Art

[0002] my country's petrochemical industry standards (GB-T 12583-1998 and HT 0762-2005) define the testing process for lubricating oil friction coefficients and specify the method for observing wear spot morphology. Wear spots are generally circular or elliptical in shape. Severely deformed wear spots cannot be used for lubricating oil friction coefficient testing and require a new run-in test. Therefore, rapid and accurate determination of wear spot shape is crucial for determining wear spot morphology and test validity. However, no research has yet been conducted on automated detection of wear spot deformation in four-ball friction tests. Qualitative determinations are still made by testers based on their experience, inevitably leading to subjective errors. This is neither scientific nor objective, and offers limited practical guidance. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for quantitatively determining the distortion amount of an elliptical wear spot image, which solves the above-mentioned deficiencies in the prior art.

[0004] In order to achieve the above object, the technical solution adopted in the present invention is:

[0005] The present invention provides a method for quantitatively determining the distortion amount of an elliptical wear spot image, comprising the following steps:

[0006] Step 1, segmenting the acquired four-ball friction wear spot image to obtain a wear spot segmentation image;

[0007] Step 2: rotating the obtained wear spot segmentation image to obtain a rotated wear spot image;

[0008] Step 3, constructing an elliptical wear spot image based on the obtained rotation wear spot image;

[0009] Step 4, calculating the matching degree between the rotated segmentation image and the ellipse template image;

[0010] Step 5, calculating the wear spot distortion amount according to the obtained matching degree;

[0011] Step 6: Determine the wear spot abnormality of the acquired four-ball friction wear spot image based on the obtained wear spot distortion amount.

[0012] Preferably, in step 2, the obtained wear spot segmentation image is rotated to obtain a rotated wear spot image, and the specific method is:

[0013] Obtaining the wear scar direction angle of the wear spot segmentation image;

[0014] The wear spot segmentation map is rotated according to the obtained wear scar direction angle to obtain a rotated wear spot map.

[0015] Preferably, in step 3, an elliptical wear spot image is constructed according to the obtained rotation wear spot image, and the specific method is:

[0016] Obtaining position parameters of the rotating wear spot image;

[0017] According to the obtained position parameters, the elliptical wear spot map is constructed by combining the following formula:

[0018]

[0019] Among them, x O and y O are the row coordinates and column coordinates of the centroid O of the ellipse template graph, respectively, and and a and b are the major and minor axes of the elliptical template graph, respectively, and S is the area of ​​the rotating wear spot pattern; r is the radius of the rotating wear spot pattern; (x C ,y C ) are the centroid coordinates of the rotated segmentation graph.

[0020] Preferably, the position size of the rotating wear spot image includes the centroid and area, wherein the centroid and area of ​​the rotating wear spot image are calculated respectively by the following formulas:

[0021]

[0022]

[0023] S=num(g) (3)

[0024]

[0025] Wherein, (x, y) is the coordinate of any pixel in the rotated segmentation image, 1≤x≤M and 1≤y≤N; g(x, y) represents the value of pixel (x, y) in the rotated segmentation image g; g(x, y) = 1 indicates that the pixel (x, y) is a wear spot, and g(x, y) = 0 indicates that the pixel (x, y) is not a wear spot; S is the area; r is the equivalent radius value; num(g) is the pixel statistical operator of the wear spot area in the rotated wear spot image.

[0026] Preferably, in step 4, the matching degree between the rotated segmentation image and the ellipse template image is calculated by the following formula:

[0027]

[0028] Where T is the matching degree between the wear spot and the ellipse template; ∩ is the intersection operator.

[0029] Preferably, in step 5, the wear spot distortion amount is calculated based on the obtained matching degree, and the specific method is:

[0030] Obtain the best ellipse template image according to the matching degree;

[0031] Calculate the shape rate and eccentricity corresponding to the optimal ellipse template image;

[0032] The wear spot distortion is calculated based on the obtained shape ratio and eccentricity.

[0033] Preferably, the optimal ellipse template graph is obtained according to the step length solution method combined with the following formula:

[0034]

[0035]

[0036] Among them, T * is the matching degree between the best elliptical template and the wear spot; and The centroid O of the optimal ellipse template * The row and column coordinates of a * and b * are the major and minor axes of the optimal ellipse template.

[0037] Preferably, the profile rate and eccentricity are calculated by the following formula:

[0038] Y=1-T * (9)

[0039]

[0040] Among them, Y is the shape rate; T * is the matching degree between the best elliptical template and the wear spot; e is the eccentricity; a * and b * are the major and minor axes of the optimal ellipse template.

[0041] Preferably, the wear spot distortion is calculated by the following formula:

[0042] H=α·Y+β·e

[0043] Among them, α and β are the weight of the shape rate and the eccentricity rate, respectively; H is the wear spot distortion; Y is the shape rate; and e is the eccentricity.

[0044] Preferably, in step 6, the wear spot abnormality of the acquired four-ball friction wear spot image is determined based on the obtained wear spot distortion amount, and the specific method is:

[0045] When the obtained wear spot distortion amount is less than or equal to the preset judgment threshold, it is determined that the wear spot of the obtained four-ball friction wear spot image has not been deformed; otherwise, it is determined that the wear spot of the obtained four-ball friction wear spot image has been deformed.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] The present invention provides a method for quantitatively determining the distortion of elliptical wear spot images. Wear spot images are captured using an image acquisition device or a scanning electron microscope (both with a magnification function). Image analysis and processing techniques are then used to automatically segment the wear spot area, improving segmentation speed and accuracy while significantly reducing the workload of manual segmentation. Based on the construction of an elliptical template, multiple image feature parameters describing the wear spot shape are established, enabling quantitative expression of the wear spot shape. Furthermore, an evaluation index for the degree of wear spot distortion is established based on these multiple image feature parameters, enabling determination of the wear spot shape distortion, thus eliminating the subjective and unscientific nature of qualitative determinations by testers. This method is simple, efficient, and requires no human assistance, making it highly suitable for development into automatic analysis and processing software for four-ball friction test data. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is the wear spot image f;

[0049] Figure 2 is the wear spot segmentation map R;

[0050] Figure 3 Figure g is the rotation wear spot;

[0051] Figure 4 This is the ellipse template diagram B (example). DETAILED DESCRIPTION

[0052] The present invention will be described in further detail below with reference to the accompanying drawings.

[0053] The present invention proposes a method for quantitatively determining the distortion of an elliptical wear spot image. The specific implementation steps are as follows:

[0054] Step S0: Collection of wear spot images. After the four-ball friction test, wear spots will form on the surface of the steel balls. The surface wear spots (≤1mm) are not visible to the naked eye and require special collection equipment such as a high-power microscope or a scanning electron microscope to collect the wear spot images.

[0055] In this embodiment, the collected wear spot image f is as follows: Figure 1 shown.

[0056] Step S1: Segmentation of the wear spot image. The segmentation of the wear spot image includes manual segmentation and automatic segmentation methods, and its purpose is to distinguish the wear spot from the background. Compared with the manual segmentation method, the automatic segmentation method has the advantages of high segmentation accuracy and fast speed. In this embodiment, a dual-extreme value filtering difference measurement operator is established to perform preliminary segmentation of the wear spot, and grayscale and distance dual-constrained boundary subdivision is used to accurately segment the wear spot area boundary. The algorithm includes the steps of grayscale conversion, denoising, preliminary segmentation of the wear spot area, and dual-constrained boundary fine segmentation. The algorithm has a good boundary segmentation effect and runs fast.

[0057] In this embodiment, the wear spot segmentation map R is as follows: Figure 2 shown.

[0058] Step S2: rotating the wear spot segmentation image according to the wear scar direction angle.

[0059] Determining the wear scar direction angle is not easy, and the directions of all wear scars need to be considered comprehensively. Considering the similarity of the wear scar directions in the four-ball friction test images, this embodiment uses a row convolution operation that reflects the grayscale change of adjacent rows to calculate the wear amount. Starting from the obvious wear scar, the wear scar horizontality under angle transformation is calculated. The angle corresponding to the maximum wear scar horizontality is the wear scar direction angle. This algorithm has high detection accuracy and fast processing speed. According to the wear scar direction angle, the wear spot segmentation image R is rotated 46° clockwise to obtain the rotated segmentation image g. After the rotation operation, the image size will become larger, which is M×N. M and N are the maximum row number and the maximum column number of the rotated segmentation image, i.e., the size.

[0060] In this embodiment, the wear scar direction angle is 46°, and the rotation segmentation diagram g is as follows: Figure 3 As shown, its dimensions are: M=1261 and N=1273.

[0061] Step S3: Extraction of shape parameters of wear spots. Extraction of the position and size parameters of wear spots from the rotated segmentation image g is the key to subsequent processing, which mainly includes: the position of the centroid C, the area S and the radius r. The position of the centroid C is expressed in two-dimensional coordinates of the pixel coordinate system as (x C ,y C ), where x C and y C The row and column numbers are respectively; the area refers to the number of pixels in the wear spot area, represented by S:

[0062]

[0063]

[0064] S=num(g) (3)

[0065]

[0066] Wherein, (x, y) is the coordinate of any pixel in the rotated segmentation image, 1≤x≤M and 1≤y≤N; g(x, y) represents the value of pixel (x, y) in the rotated segmentation image g, and g is a binary image; g(x, y) = 1 indicates that the pixel (x, y) is a wear spot, and g(x, y) = 0 indicates that the pixel (x, y) is not a wear spot; S is the area; r is the equivalent radius value; num(g) is the pixel statistical operator of the wear spot area in the rotated wear spot image g.

[0067] In this embodiment, the position and size parameters of the wear spot are: C =581.58,y C =580.58, S=422549 and r=366.74.

[0068] Step S4: Construction of ellipse template.

[0069] Ideally, the wear spot is elliptical. By constructing an elliptical template and comparing the fit between the two, the morphology and deviation of the wear spot can be measured.

[0070] The ellipse constructed based on the position and size parameters of the wear spot must meet four conditions at the same time:

[0071] ① The area is equal to S (obtained from step S3);

[0072] ②The two symmetry axes are horizontal and vertical respectively;

[0073] ③ The Euclidean distance between the ellipse centroid O and the wear spot centroid C in row and column coordinates is not greater than r;

[0074] ④ The major axis a and minor axis b of the ellipse. According to expert experience, the range of the major axis a satisfies: According to condition ①, we can know

[0075] When the above four conditions are met at the same time, the obtained elliptical template graph B is expressed as shown in (5):

[0076]

[0077] Among them, x O and y O are the row and column coordinates of the centroid O of the elliptical template image, respectively. The absolute distance between the centroid O of the elliptical template image and the row and column coordinates of the center C of the rotating wear spot pattern is not greater than r, which satisfies: and a and b are the long and short semi-axes of the ellipse template respectively; B is the ellipse template, the size is also M×N, and its pixel value will change with x O 、y O or a or b) changes with the value.

[0078] In this embodiment, when x o =560.58,y o =601.58, a=380.74 and b=353.26, the constructed ellipse template diagram B is as follows Figure 4 shown.

[0079] Step S5: Calculate the matching degree between the rotated segmentation image and the ellipse template image.

[0080] The matching degree is used to measure the similarity between the wear spot and the ellipse, and its calculation formula is shown in (6).

[0081]

[0082] Where T is the matching degree between the wear spot and the ellipse template; ∩ is the intersection operator, that is, when g(x, y) and B(x, y) are both 1, the pixel value is 1, otherwise it is 0.

[0083] In this embodiment, the matching degree corresponding to B in step S4 is:

[0084] Step S6: Intelligent solution of the optimal ellipse template.

[0085] When the matching degree between the wear spot and the elliptical template is the highest, the elliptical template at this time is called the best elliptical template. O ,y O The solution space for a and b is very large, and traversing to find the optimal ellipse template takes a long time. Alternatively, fast intelligent solution algorithms such as genetic algorithms and ant colonies can be used. The present invention adopts a different step size solution method, which greatly reduces the amount of calculation while ensuring the optimal solution. That is, a larger parameter change step size is first used to traverse the solution space, and then a smaller step size is given to traverse to find the optimal parameter solution.

[0086] The maximum matching degree and its corresponding optimal parameter solution expression are shown in (7-8):

[0087]

[0088]

[0089] Among them, T * is the matching degree between the best elliptical template and the wear spot; and The centroid O of the optimal ellipse template * The row and column coordinates of a * and b * are the major and minor axes of the optimal ellipse template.

[0090] In this embodiment, the matching degree between the optimal elliptical template and the wear spot is: The characteristic parameters of the optimal ellipse template are: a * =369.74 and b * =363.77.

[0091] Step S7: deformation characteristics of wear spots and their representation.

[0092] When the wear spot is deformed, it is mainly manifested as a larger deformation of the (actual) wear spot compared with the standard elliptical template, or a larger deviation between the major and minor axes. Therefore, the two indicators of irregularity and eccentricity are established to determine the deformation degree of the wear spot.

[0093] The deformity rate Y refers to the proportion of the area where the actual wear spot does not match the standard wear spot, reflecting the deformation rate of the actual wear spot. The larger the Y value, the greater the deformation rate of the actual wear spot.

[0094] The eccentricity e is used to express the deviation between the major and minor semi-axis, reflecting the difference in the measured diameter value of the actual wear spot. The deformation degree of the wear spot is directly proportional to the size of the e value.

[0095] The expressions of wear spot shape and eccentricity are shown in (9-10):

[0096] Y=1-T * (9)

[0097]

[0098] In this embodiment, the profile ratio Y=0.0344, and the eccentricity e=0.1790.

[0099] Step S8: Wear spot distortion. The wear spot distortion is used to measure the degree of wear spot distortion. When the wear spot's shape or eccentricity is large, the wear spot distortion is also large, and vice versa. The wear spot distortion is calculated as shown in (11).

[0100] H=α·Y+β·e (11)

[0101] Among them, α and β are the weight of the irregularity rate and the weight of the eccentricity, respectively, satisfying: 0≤α≤1, 0≤β≤1, α+β=1; H is the distortion amount of the wear spot.

[0102] In this embodiment, the roundness weight α and the distortion weight β are set to 0.2 and 0.8, respectively, and the wear spot distortion amount is obtained to be H=0.0730.

[0103] Step S9: Determination of abnormality of wear spots.

[0104] Furthermore, by establishing an abnormal wear spot database, analyzing the unified characteristics of the abnormal wear spot shape rate Y and eccentricity e and expert experience, the judgment threshold of abnormal wear spots is obtained, which can realize the abnormal discrimination of wear spot morphology.

[0105] When the obtained wear spot distortion amount is less than or equal to the preset judgment threshold, it is determined that the wear spot of the obtained four-ball friction wear spot image has not been deformed; otherwise, it is determined that the wear spot of the obtained four-ball friction wear spot image has been deformed.

[0106] In this embodiment, the threshold value for determining abnormal wear spots is 0.2, and the distortion of the wear spot image is 0.0730<0.2, so the wear spot image is elliptical and has not been deformed.

Claims

1. A method for quantitatively determining the distortion of an elliptical wear spot image, characterized in that: The following steps are involved: Step 1, segmenting the acquired four-ball friction wear spot image to obtain a wear spot segmentation image; Step 2: rotating the obtained wear spot segmentation image to obtain a rotated wear spot image; Step 3, constructing an elliptical template image based on the obtained rotation wear spot image; Step 4, calculating the matching degree between the rotated segmentation image and the ellipse template image; Step 5, calculating the wear spot distortion amount according to the obtained matching degree; Step 6, judging the wear spot abnormality of the acquired four-ball friction wear spot image according to the obtained wear spot distortion amount; In step 3, an elliptical template image is constructed based on the obtained rotational wear spot image. The specific method is: Obtaining position parameters of the rotating wear spot image; According to the obtained position parameters, the ellipse template graph is constructed by combining the following formula: in, and are the centroids of the ellipse template graphs The row and column coordinates of and ; and are the major and minor axes of the elliptical template graph, respectively, and , ; is the area of ​​the rotating wear spot pattern; is the radius of the rotating wear spot pattern; are the centroid coordinates of the rotated segmentation image.

2. The method for quantitatively determining the distortion of an elliptical wear spot image according to claim 1, characterized in that: In step 2, the obtained wear spot segmentation image is rotated to obtain a rotated wear spot image. The specific method is: Obtaining the wear scar direction angle of the wear spot segmentation image; The wear spot segmentation map is rotated according to the obtained wear scar direction angle to obtain a rotated wear spot map.

3. The method for quantitatively determining the distortion of an elliptical wear spot image according to claim 1, wherein: The positional dimensions of the rotating wear spot image include the centroid and area, wherein the centroid and area of ​​the rotating wear spot image are calculated by the following formulas: (1) (2) (3) (4) in, is the coordinate of any pixel in the rotated segmentation map, ; Represents the rotation segmentation map g Medium pixels The value of Represents pixels For wear spots, Represents pixels It is non-wear spot; is the area of ​​the rotating wear spot pattern; is the equivalent radius value; It is the pixel statistical operator of the wear spot area in the rotation wear spot image.

4. The method for quantitatively determining the distortion of an elliptical wear spot image according to claim 1, wherein: In step 4, the matching degree between the rotated segmentation image and the ellipse template image is calculated by the following formula: (6) in, is the matching degree between the wear spot and the elliptical template; is the intersection operator; is the rotation segmentation map; is the area of ​​the rotating wear spot pattern; It is an ellipse template diagram.

5. The method for quantitatively determining the distortion of an elliptical wear spot image according to claim 1, wherein: In step 5, the wear spot distortion is calculated based on the obtained matching degree. The specific method is: Obtain the best ellipse template image according to the matching degree; Calculate the shape rate and eccentricity corresponding to the optimal ellipse template image; The wear spot distortion is calculated based on the obtained shape ratio and eccentricity; The optimal ellipse template graph is obtained by combining the step length solution method with the following formula: (7) (8) in, is the matching degree between the best elliptical template and the wear spot; and The centroid of the best ellipse template The row and column coordinates of and are the major and minor axes of the optimal ellipse template.

6. The method for quantitatively determining the distortion of an elliptical wear spot image according to claim 5, characterized in that: The shape ratio and eccentricity are calculated by the following formula: (9) (10) in, is the heteromorphism rate; is the matching degree between the best elliptical template and the wear spot; is the eccentricity; and are the major and minor axes of the optimal ellipse template.

7. The method for quantitatively determining the distortion of an elliptical wear spot image according to claim 5, wherein: The wear spot distortion is calculated by the following formula: in, and are the shape rate weight and eccentricity weight respectively; is the wear spot distortion; is the heteromorphism rate; is the eccentricity.

8. The method for quantitatively determining the distortion of an elliptical wear spot image according to claim 1, wherein: In step 6, the wear spot abnormality of the acquired four-ball friction wear spot image is determined based on the obtained wear spot distortion amount. The specific method is: When the obtained wear spot distortion amount is less than or equal to the preset judgment threshold, it is determined that the wear spot of the obtained four-ball friction wear spot image has not been deformed; otherwise, it is determined that the wear spot of the obtained four-ball friction wear spot image has been deformed.

Images