A 3D detection method for strip sealing rings

Abstract

The application provides a kind of strip-shaped sealing ring glue strip 3D detection method in the technical field of glue strip detection, comprising: step S1, 3D depth image is converted to 2D gray image to obtain detection starting and ending point P start And P end ; Step S2, set detection pixel length L * , detection direction D, detection accuracy, based on L * , detection accuracy setting first sliding window W1 to obtain cross-sectional image by segmenting 3D depth image;Step S3, the derivative value is obtained by derivation to cross-sectional image, based on derivative value record point set P List1 To determine glue peak;Step S4, set second sliding window W2 to traverse glue peak, to record point set P List2 And determine glue valley;Step S5, set third sliding window W3 to carry out segmented integration to glue strip cross-sectional curve, obtain glue strip cross-sectional area;Step S6, based on the linear function g (x) of glue valley connecting line detects convex cover, glue break, adhesion;Step S7, control W2 in glue valley sliding to detect position degree, boundary information.The application has the advantages that: greatly improve the accuracy of glue strip detection.

Description

Technical Field

[0001] The present invention relates to the field of rubber strip detection technology, and in particular to a 3D detection method for a strip-shaped sealing ring rubber strip. Background Art

[0002] During the gluing process of the sealing ring, the rubber strip may break, or parameters such as the width, height, and cross-sectional area of ​​the rubber strip may not meet the standards. Therefore, the rubber strip needs to be tested. Traditionally, there are three methods for testing the rubber strip:

[0003] Method 1: Detection is performed manually with the naked eye. However, this method is subjective and cannot objectively control the detection parameters. It has a certain error rate and requires the inspectors to enter the equipment for inspection. This not only greatly consumes manpower but also poses a safety hazard.

[0004] Method 2: Detection using an area array camera. However, due to the three-dimensional structure of the sealing ring rubber strip, this method is limited by the imaging results of the area array camera and cannot detect parameters such as the rubber height and cross-sectional area. In addition, the area array camera is affected by factors such as the environment, rubber strip color, and lens distortion, and cannot meet the requirements of high-precision detection.

[0005] Method 3: Detection is performed using a 3D camera. Due to different detection methods, the detection results of the rubber strip can only reflect the overall average value, and cannot reflect the detection parameters of a small part of the rubber strip. It is also impossible to make timely and accurate judgments on problems such as rubber wire adhesion after rubber breakage and boss coverage.

[0006] Therefore, how to provide a 3D detection method for strip sealing ring rubber strips to improve the accuracy of rubber strip detection has become a technical problem that needs to be solved urgently. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide a 3D detection method for strip sealing ring rubber strips to improve the accuracy of rubber strip detection.

[0008] The present invention is implemented as follows: a 3D detection method for a strip sealing ring rubber strip comprises the following steps:

[0009] Step S1: acquiring a 3D depth image of the strip-shaped sealing ring through a 3D camera, and converting the 3D depth image into a 2D grayscale image;

[0010] Step S2: perform feature recognition on the 2D grayscale image to obtain the pixel coordinates P of the detection starting point of the strip seal rubber strip. start and the detection end point pixel coordinate P end ;

[0011] Step S3: Set the detection pixel length L *, detection direction D and detection accuracy, based on the detection pixel length L * And the detection accuracy sets the first-level sliding window W1, based on the detection starting point pixel coordinate P start , Detection endpoint pixel coordinates P end And the detection direction D determines the starting detection position and detection direction of the first-level sliding window W1 in the 3D depth image;

[0012] Step S4: segmenting the 3D depth image using the primary sliding window W1 to obtain a plurality of cross-sectional images;

[0013] Step S5: Derivative the cross-sectional images in turn to obtain derivative values, and record the point set P based on the derivative values. List1 , based on the point set P List1 Determine the gel peak;

[0014] Step S6: Set a secondary sliding window W2, and traverse the glue peak based on the secondary sliding window W2 to record the point set P List2 ;

[0015] Step S7: Based on the point set P List2 Identify the gum valley;

[0016] Step S8: set a three-level sliding window W3 at the glue peak, perform piecewise integral calculation on the cross-sectional curve of the rubber strip within the three-level sliding window W3, and the upper limit is the curve function f(x i ), the lower limit of which is a linear function g(x) of the line connecting the rubber valleys on both sides of the rubber strip, and the cross-sectional area of ​​the rubber strip is obtained;

[0017] Step S9: Detecting boss coverage, adhesive breakage, and adhesion based on the slope of the linear function g(x);

[0018] Step S10: placing the secondary sliding window W2 at the glue valley position, and controlling the secondary sliding window W2 to slide in the unilateral direction corresponding to the glue valley based on a linear detection method to detect the glue strip position and boundary information.

[0019] Furthermore, the step S1 is specifically as follows:

[0020] A 16-bit 3D depth image of the strip-shaped sealing ring is acquired through a 3D camera, and the depth data of the 3D depth image is linearly mapped to obtain grayscale data, which is then compressed into an 8-bit 2D grayscale image.

[0021] Furthermore, the step S5 is specifically as follows:

[0022] Derivative values ​​are obtained by sequentially performing derivative calculations on each cross-sectional image, and a derivative threshold value A is set. The coordinates whose derivative values ​​are less than |A| are salient point coordinates, and the coordinates of each salient point are recorded in the point set P. List1 ;|A| is the absolute value of A;

[0023] Set a rubber strip height range, based on the rubber strip height range from the point set P List1 The coordinates that match the conditions are marked as glue peaks.

[0024] Furthermore, the step S6 is specifically as follows:

[0025] Set a secondary sliding window W2, traverse the decentralized neighborhood δ of the glue peak based on the secondary sliding window W2, and record the coordinates of the concave points whose derivative values ​​are less than |A| into the point set P List2 .

[0026] Furthermore, the step S7 is specifically as follows:

[0027] Set an angle threshold range and enumerate the point set P List2 Coordinate P in i , find the coordinates whose line angles meet the angle threshold range, and select the coordinates that meet min(|P i -P peak |) coordinates as the valley; P peak Indicates gel peak.

[0028] The advantages of the present invention are:

[0029] The 3D depth image of the strip seal is acquired by a 3D camera and converted into a 2D grayscale image. Feature recognition is performed on the 2D grayscale image to obtain the pixel coordinates P of the detection starting point of the strip seal. start and the detection end point pixel coordinate P end ; Then set the detection pixel length L * , detection direction D and detection accuracy, based on L * And the detection accuracy sets the first-level sliding window W1, based on P start 、P end And D determines the starting detection position and detection direction of W1 in the 3D depth image, and divides the 3D depth image into several cross-sectional images through W1. The derivative values ​​of each cross-sectional image are derived in turn, and the point set P is recorded based on the derivative values. List1 , based on the point set P List1 Determine the glue peak; traverse the glue peak based on the set secondary sliding window W2 to record the point set P List2 , based on the point set P List2Determine the glue valley; set a three-level sliding window W3 at the glue peak, perform piecewise integral calculation on the glue strip cross-section curve within W3, and the upper limit is the curve function f(x i ), the lower limit of which is a linear function g(x) of the line connecting the rubber valleys on both sides of the rubber strip, and the cross-sectional area of ​​the rubber strip is calculated. Finally, the boss coverage, glue breakage and adhesion are detected based on the slope of g(x). The secondary sliding window W2 is placed at the position of the rubber valley, and W2 is controlled to slide in the unilateral direction corresponding to the rubber valley based on the linear detection method to detect the position of the rubber strip and the boundary information. That is, based on the 3D depth image, combined with the three-level sliding window, segmented integration and other technical means, a comprehensive detection of the rubber strip cross-sectional area, boss coverage, glue breakage, adhesion, rubber strip position and boundary information of the rubber strip is achieved, which ultimately greatly improves the accuracy of rubber strip detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0031] Figure 1 The present invention is a flow chart of a 3D detection method for a strip sealing ring rubber strip. DETAILED DESCRIPTION

[0032] The technical solution in the embodiments of the present application has the following overall idea: converting the 3D depth image into a 2D grayscale image, and based on the 3D depth image and the 2D grayscale image, combining three-level sliding windows, segmented integration and other technical means, performing a comprehensive detection of the rubber strip's cross-sectional area, boss coverage, rubber breakage, adhesion, rubber strip position and boundary information to improve the accuracy of rubber strip detection.

[0033] Please refer to Figure 1 As shown, a preferred embodiment of a 3D detection method for a strip-shaped sealing ring rubber strip of the present invention includes the following steps:

[0034] Step S1, obtaining a 3D depth image (RangeImage) of the strip sealing ring through a 3D camera, and converting the 3D depth image into a 2D grayscale image (GreyImage);

[0035] Step S2: perform feature recognition on the 2D grayscale image to obtain the pixel coordinates P of the detection starting point of the strip seal rubber strip. start and the detection end point pixel coordinate P end ;

[0036] Step S3: Set the detection pixel length L * , detection direction D and detection accuracy, based on the detection pixel length L * And the detection accuracy sets the first-level sliding window W1, that is, calculate and set the pixel size of W1 to M*×N*, based on the detection starting point pixel coordinate Pstart , Detection endpoint pixel coordinates P end And the detection direction D determines the starting detection position and detection direction of the first-level sliding window W1 in the 3D depth image;

[0037] The search position coordinates of W1 are (P start , sliding offset), the initial value of the sliding offset is 0;

[0038] M*≤x coordinate range of 3D depth image;

[0039] N*=(L * *D) / actual detection length L;

[0040] Step S4: segmenting the 3D depth image using the primary sliding window W1 to obtain a plurality of cross-sectional images;

[0041] The segmented content is the image within the first-level sliding window W1, and the mean filter is performed on the segmented cross-sectional image using the mean value of N×1 size (N rows and 1 column), so that the image within the first-level sliding window W1 is converted from a 3D array to a 2D array, and the axial mean filter is implemented, and the discrete data points in the 2D array are sorted by step size h. i (x i ) segmented, and perform cubic spline interpolation, while taking the second-order derivative value at the given interpolation point as the unknown variable, solve the three-moment equations to obtain the curve function f(x i ) and the cross-sectional image (CutImage) of cross-sectional binarization.

[0042] By step size h i (x i ) The purpose of segmentation is to avoid the Runge-Kutta phenomenon in the fitted polynomial.

[0043] The formulas for the cubic spline curve expression f(x) and its differential expressions f'(x) and f″(x) are:

[0044] f(x)=a i +b i (xx i )+c i (xx i ) 2 +d i (xx i ) 3

[0045] f'(x)=b i +2c i (xx i )+3d i (xx i )2 ①

[0046] f″(x)=2c i +6d i (xx i ) ②

[0047] Its cubic spline interpolation expression is as follows:

[0048] i=0,1,2…n-2 ③

[0049] f i (x i )=y i ④

[0050] f i (x i+1 )=y i+1 ⑤

[0051] The step size expression is:

[0052] h i (x i )=x i+1 -x i ⑥

[0053] Let m i =f″ i (x i )=2c i , using the second-order derivative value at the given interpolation point as the unknown variable to solve the three-moment equations, that is, from the above equations ①②③④⑤⑥, we can get the value of m i The linear equations for :

[0054] a i =y i

[0055]

[0056]

[0057]

[0058]

[0059] According to ⑦, the free boundary conditions are set and the general solution of the non-homogeneous linear equations is obtained.

[0060] About a i 、b i 、c i d i The general solution vector Its general solution expression is as follows:

[0061]

[0062]

[0063]

[0064]

[0065]

[0066] Substitute the coefficient value of ζ in the general solution into the cubic spline curve expression f(x i ), the cubic spline curve function expression f(x) of the cross section of multiple rubber strips is obtained.

[0067] Step S5: Derivative the cross-sectional images in turn to obtain derivative values, and record the point set P based on the derivative values. List1 , based on the point set P List1 Determine the gel peak (P peak );

[0068] Step S6: Set a secondary sliding window W2, and traverse the glue peak based on the secondary sliding window W2 to record the point set P List2 ;

[0069] W1 search position coordinate = gel peak coordinate ± δ;

[0070] Step S7: Based on the point set P List2 Determine the glue valley (P valley );

[0071] Step S8: set a three-level sliding window W3 at the glue peak, perform piecewise integral calculation on the cross-sectional curve of the rubber strip within the three-level sliding window W3, and the upper limit is the curve function f(x i ), the lower limit is the linear function g(x) of the line connecting the rubber valleys on both sides of the rubber strip, and then the cross-sectional area of ​​the rubber strip is obtained, that is, a, b are the step lengths h of the corresponding paragraphs within the W3 size range i (x i )index;

[0072] The size of the three-level sliding window W3 is variable, and size fitting is performed at the glue peak;

[0073] W3 size = the difference in x-coordinates between the two rubber valleys * the length of the perpendicular segment of the line connecting the rubber peak to the rubber valley;

[0074] Step S9: Based on the slope of the linear function g(x), combined with the slope threshold and the coordinates of each point, perform detection of boss coverage, adhesive breakage, and adhesion;

[0075] Step S10: placing the secondary sliding window W2 at the glue valley position, and controlling the secondary sliding window W2 to slide in the unilateral direction corresponding to the glue valley based on a linear detection method to detect the glue strip position and boundary information;

[0076] Step S11, offset the first-level sliding window W1 in the detection direction D by the size of W1, and repeatedly detect the rubber strip cross-sectional area, boss coverage, rubber breakage, adhesion, rubber strip position and boundary information until the detection of the entire 3D depth image is completed.

[0077] The step S1 is specifically as follows:

[0078] The 16-bit 3D depth image of the strip seal is acquired by a 3D camera, and the depth data (Value Deep ) to perform linear mapping to obtain grayscale data (Value Grey ), compressing the grayscale data into an 8-bit 2D grayscale image.

[0079] The depth data is converted to grayscale data to meet the following conditions: Grey =αValue Deep , α is a constant.

[0080] The step S5 is specifically as follows:

[0081] Derivative values ​​are obtained by sequentially performing derivative calculations on each cross-sectional image, and a derivative threshold value A is set. The coordinates whose derivative values ​​are less than |A| are salient point coordinates, and the coordinates of each salient point are recorded in the point set P. List1 ;|A| is the absolute value of A;

[0082] Set a rubber strip height range, based on the rubber strip height range from the point set P List1 The coordinates that match the conditions are marked as glue peaks.

[0083] The step S6 is specifically as follows:

[0084] Set a secondary sliding window W2, traverse the decentralized neighborhood δ of the glue peak based on the secondary sliding window W2, and record the coordinates of the concave points whose derivative values ​​are less than |A| into the point set P List2 .

[0085] The step S7 is specifically as follows:

[0086] Set an angle threshold range and enumerate the point set P List2 Coordinate P in i , find the coordinates whose line angles meet the angle threshold range, and select the coordinates that meet min(|P i -P peakThe coordinates of the glue peak are taken as the glue valley; P peak The coordinates of the glue peak are taken as the glue valley; P

[0087] In summary, the advantages of the present application are:

[0088] The 3D depth image of the strip-shaped sealing ring glue strip is acquired by the 3D camera and converted into a 2D gray image, feature recognition is performed on the 2D gray image, the detection starting point pixel coordinates P start and the detection ending point pixel coordinates P end of the strip-shaped sealing ring glue strip are acquired; then the detection pixel length L * , the detection direction D and the detection accuracy are set, the first-level sliding window W1 is set based on L * and the detection accuracy, the starting detection position and the detection direction of W1 in the 3D depth image are determined based on P start , P end and D, the 3D depth image is segmented by W1 to obtain a plurality of cross-sectional images, the derivative values are acquired by sequentially performing derivation on each cross-sectional image, the point set P List1 is recorded based on the derivative values, the glue peak is determined based on the point set P List1 , the point set P List2 is recorded based on the traversal of the glue peak by the set second-level sliding window W2, the glue valley is determined based on the point set P List2 , the third-level sliding window W3 is set at the glue peak, the sectional integral calculation is performed on the glue strip cross-sectional curve in W3, the upper limit is the curve function f(x i of the glue strip cross-sectional curve, and the lower limit is the first linear function g(x) of the connecting line of the glue valleys on both sides of the glue strip, the glue strip cross-sectional area is calculated, and finally the convex cover, the broken glue and the adhesion are detected based on the slope of g(x), the second-level sliding window W2 is placed at the glue valley position, W2 is controlled to slide in the corresponding single side direction of the glue valley based on the linear detection method, so as to detect the glue strip position degree and the boundary information, that is, based on the 3D depth image, the third-level sliding window, the sectional integral and other technical means are combined, the glue strip cross-sectional area, the convex cover, the broken glue, the adhesion, the glue strip position degree and the boundary information of the glue strip are comprehensively detected, and finally the accuracy of the glue strip detection is greatly improved.

[0089] Although the specific embodiments of the present application are described above, those skilled in the art should understand that the specific examples described are only illustrative, and are not intended to limit the scope of the present application, and equivalent modifications and changes made by those skilled in the art in accordance with the spirit of the present application should be covered within the scope of the claims of the present application.

Claims

1. A 3D detection method for strip sealing rings, characterized by: The steps include: Step S1: acquiring a 3D depth image of the strip-shaped sealing ring through a 3D camera, and converting the 3D depth image into a 2D grayscale image; Step S2: perform feature recognition on the 2D grayscale image to obtain the pixel coordinates P of the detection starting point of the strip seal rubber strip. start and the detection end point pixel coordinate P end ; Step S3: Set the detection pixel length L * , detection direction D and detection accuracy, based on the detection pixel length L * And the detection accuracy sets the first-level sliding window W1, based on the detection starting point pixel coordinate P start , Detection endpoint pixel coordinates P end And the detection direction D determines the starting detection position and detection direction of the first-level sliding window W1 in the 3D depth image; Step S4: segmenting the 3D depth image using the primary sliding window W1 to obtain a plurality of cross-sectional images; Step S5: Derivative the cross-sectional images in turn to obtain derivative values, and record the point set P based on the derivative values. List1 , based on the point set P List1 Determine the gel peak; Step S6: Set a secondary sliding window W2, and traverse the glue peak based on the secondary sliding window W2 to record the point set P List2 ; Step S7: Based on the point set P List2 Identify the gum valley; Step S8: set a three-level sliding window W3 at the glue peak, perform piecewise integral calculation on the cross-sectional curve of the rubber strip within the three-level sliding window W3, and the upper limit is the curve function f(x i ), the lower limit of which is a linear function g(x) of the line connecting the rubber valleys on both sides of the rubber strip, and the cross-sectional area of ​​the rubber strip is obtained; Step S9: Detecting boss coverage, adhesive breakage, and adhesion based on the slope of the linear function g(x); Step S10: placing the secondary sliding window W2 at the glue valley position, and controlling the secondary sliding window W2 to slide in the unilateral direction corresponding to the glue valley based on a linear detection method to detect the glue strip position and boundary information.

2. The 3D detection method for a strip seal according to claim 1, characterized in that: The step S1 is specifically as follows: A 16-bit 3D depth image of the strip-shaped sealing ring is acquired through a 3D camera, and the depth data of the 3D depth image is linearly mapped to obtain grayscale data, which is then compressed into an 8-bit 2D grayscale image.

3. The 3D detection method for a strip seal ring according to claim 1, characterized in that: The step S5 is specifically as follows: Derivative values ​​are obtained by sequentially performing derivative calculations on each cross-sectional image, and a derivative threshold value A is set. The coordinates whose derivative values ​​are less than |A| are salient point coordinates, and the coordinates of each salient point are recorded in the point set P. List1 ;|A| is the absolute value of A; Set a rubber strip height range, based on the rubber strip height range from the point set P List1 The coordinates that match the conditions are marked as glue peaks.

4. The 3D detection method for a strip seal ring according to claim 3, characterized in that: The step S6 is specifically as follows: Set a secondary sliding window W2, traverse the decentralized neighborhood δ of the glue peak based on the secondary sliding window W2, and record the coordinates of the concave points whose derivative values ​​are less than |A| into the point set P List2 .

5. The 3D detection method for a strip seal ring according to claim 1, characterized in that: The step S7 is specifically as follows: Set an angle threshold range and enumerate the point set P List2 Coordinate P in i , find the coordinates whose line angles meet the angle threshold range, and select the coordinates that meet min(|P i -P peak |) coordinates as the valley; P peak Indicates gel peak.

Images