Tray rotation angle identification method

Abstract

The invention discloses a tray rotation angle identification method, which can improve the success rate of automatic wafer combination of crystals and wafers in the wafer combination process of a multi-wafer semiconductor silicon carbide crystal growth furnace by performing automatic processing of a computer program on an image and eliminating identification interference through fuzzy, denoising, corrosion expansion, Hough circle center identification and Hough straight line identification parameters. And errors caused by human intervention are reduced.

Description

Technical Field

[0001] The invention relates to the technical field of silicon carbide crystal growth, in particular to a method for identifying a tray rotation angle. Background Art

[0002] Due to the large-scale application of crystal materials in the prior art, the demand for silicon carbide crystal equipment is increasing, especially. During the growth process of silicon carbide, silicon carbide needs to be placed in a special tray for operation, so the accuracy of the wafer and tray directly affects the growth quality. If there is an error in the combination of silicon carbide and the tray, correction is required. In the prior art, one way is to manually correct the combination of the tray and silicon carbide, but this method requires manual intervention, and there is a waste of manpower and human errors after the equipment exceeds a certain scale. Another way is to use an intelligent camera to identify the characteristic area of ​​the tray and return the angle to the rotating mechanism, but this method is affected by the increase of sediment in the equipment during the production process, which affects the success rate and accuracy of the recognition. Summary of the invention

[0003] Purpose of the invention: The present invention provides a method for identifying the rotation angle of a tray, so as to solve the technical problem of how to automatically align a silicon carbide crystal with the tray during the crystal growth process.

[0004] Technical solution: To solve the above problems, the present invention can adopt the following technical solution.

[0005] A method for identifying a rotation angle of a pallet comprises the following steps:

[0006] S1, collecting an image of the pallet through a camera, the image being a circular surface of the pallet;

[0007] S2, blur the image to reduce noise;

[0008] S3, performing morphological corrosion processing on the image, using a structural element to scan the image, and judging the relationship between the area covered by the structural element and the foreground pixel. If the structural element is completely contained in the foreground pixel area, the pixel position corresponding to the center of the structural element is marked as the foreground pixel in the corroded image; if the structural element is not completely contained in the foreground pixel area, the center pixel position is marked as the background pixel;

[0009] S4, performing morphological dilation processing on the image, when the structure element and the foreground pixel have an intersection, the pixel corresponding to the center of the structure element is marked as the foreground pixel, and the structure element moves on the image. As long as there is a white pixel in the area covered by the structure element, the pixel corresponding to the center position of the structure element is set to white in the dilated image; when the area covered by the structure element is all black pixels, the center position pixel is set to black in the dilated image;

[0010] S5, using gradient Hough circle transform to extract the center of the inner circle of the image after the above processing as the feature center;

[0011] S6. Based on all the above processing results and the processed image, taking the detected feature center as the center and setting the ratio of the circle radius and size, planning the distance range from the feature center where the corresponding straight edge appears, using Hough line detection to identify the required straight edge;

[0012] S7, establishing a rectangular coordinate system with the characteristic circle center in the result of S5 as the origin;

[0013] S8, after filtering the straight edges, the slope of the filtered straight edges in the rectangular coordinate system is calculated. If the slope error exceeds the specified error, the recognition result is considered invalid. If the slope errors of the recognized straight lines are all within the range, the samples with the largest and smallest deviations are removed, and the average slope is calculated.

[0014] S9. Cover the landing points of all identified straight lines in the four directions of up, down, left and right of the characteristic circle, convert the slope of the straight line into an absolute angle coordinate with the positive direction of the Y-axis as 0° and the clockwise rotation as a positive angle, and return the absolute center angle of the angle between the vertical line of the straight edge and the positive direction of the Y-axis as the angle at which the pallet needs to rotate.

[0015] Furthermore, in S1, the camera sets a variety of shooting parameter solutions including exposure time, contrast, and sensitivity. If the feature center of the image after the above processing fails to be extracted using the gradient Hough circle transform in S5, the process returns to S1 and changes to the next set of shooting parameter solutions to collect the tray image.

[0016] Furthermore, in S3, morphological corrosion processing is performed on the image to remove objects smaller than the structural element in the image.

[0017] Furthermore, in S4, morphological dilation processing is performed on the image to highlight the center of the circle and the straight edge features in the image.

[0018] Further, if no feature center is detected or multiple feature centers are detected in S5, it is considered that the extraction of the inner circle center of the image after the above processing by using the gradient Hough circle transform fails.

[0019] Furthermore, the error threshold specified in S8 is 1°.

[0020] Beneficial effects: Compared with the prior art, the present invention can improve the success rate of identifying the rotation angle between the silicon carbide crystal tray and the tray during the crystal growth process by automatically processing the image with a computer program, thereby reducing errors caused by human intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a flow chart of the present invention.

[0022] Figure 2 This is an operation interface diagram of a computer program formed by the algorithm in the present invention. DETAILED DESCRIPTION

[0023] like Figure 1 As shown, the present invention provides a method for identifying a rotation angle of a pallet, characterized in that it comprises the following steps:

[0024] S1. Capture an image of the tray through a camera, where the image is a circular surface of the tray; the camera sets a variety of shooting parameter schemes including exposure time, contrast, and sensitivity.

[0025] S2. Use Gaussian blur to blur the image and reduce noise.

[0026] Gaussian blur image processing is a commonly used blur processing method in the prior art, and its principle is:

[0027] This is achieved by calculating the weighted average of each pixel in the image and its surrounding pixels. The weights of the surrounding pixels are assigned using normal distribution. The closer the pixel is to the center point, the greater the weight, and the farther the pixel is, the smaller the weight. This weight distribution method makes the image visually blurred.

[0028] Implementation steps:

[0029] Select a blur radius: The blur radius determines the range of pixels considered when calculating the average.

[0030] Calculate weights: Calculate the weight of each surrounding pixel based on the probability density function of the two-dimensional normal distribution.

[0031] Apply weighted averaging: multiply the value of the center pixel by the values ​​of the surrounding pixels, sum them, and finally divide by the sum of all weights to get a weighted average.

[0032] S3. Perform morphological corrosion processing on the image, use a structural element to scan the image, and determine the relationship between the area covered by the structural element and the foreground pixel. If the structural element is completely contained in the foreground pixel area, the pixel position corresponding to the center of the structural element is marked as the foreground pixel in the corroded image; if the structural element is not completely contained in the foreground pixel area, the center pixel position is marked as the background pixel.

[0033] Mathematical formula:

[0034]

[0035] A is the processed image, B is the structural element, and (B)z means B is translated to point z.

[0036] This step is used to remove objects (such as burrs and small protrusions) that are smaller than the structural element in the image.

[0037] S4. Perform morphological dilation on the image. When the structure element intersects with the foreground pixel, mark the pixel corresponding to the center of the structure element as the foreground pixel. The structure element moves on the image. As long as there is a white pixel in the area covered by the structure element, the pixel corresponding to the center position of the structure element is set to white in the expanded image. When the area covered by the structure element is all black pixels, the center position pixel is set to black in the expanded image.

[0038] Mathematical formula:

[0039]

[0040] Among them: A is the processed image, B is the structural element, is the reflection of B, express Translate to point z.

[0041] This step is used to highlight the center of the circle and the straight edge features.

[0042] S5. Use gradient Hough circle transform to extract the center of the inner circle of the image after the above processing as the feature center; if the detection fails (the feature center is not detected or multiple feature centers are detected), return to S1, take different optical parameters according to the optical experience data stored in S1 for shooting, and then enter step S2 to repeat the process.

[0043] The principle of gradient Hough circle transform is:

[0044] 1. Perform a Canny edge detection on the original image to obtain a binary image of edge detection.

[0045] 2. Execute the Sobel operator once on the original image to calculate the neighborhood gradient values ​​of all pixels.

[0046] 3. Initialize the circle center space N(a,b), and set all N(a,b) = 0. If the center of the circle is required to be in the image, the range of a and b values ​​corresponds to the width and height of the image respectively, and N(a,b) means there are a*b in total.

[0047] 4. Traverse all non-zero pixels in the Canny edge binary image, draw a line along the gradient direction (vertical direction of the tangent, obtained by the vertical gradient and horizontal gradient calculated by the Sobel operator) with a fixed search radius, and set N(a,b)+=1 for all points (a,b) in the accumulators through which the line segment passes.

[0048] 5. Statistically sort N(a,b) to get the possible center of the circle (the larger N(a,b) is, the more likely it is the center of the circle).

[0049] Specific methods:

[0050] CircleSegment[]circles=Cv2.HoughCircles(

[0051] graycircle,

[0052] HoughModes.Gradient, / / Use gradient method to perform Hough circle transform

[0053] dp:1, / / resolution factor

[0054] minDist:300, / / minimum distance between circles

[0055] param1:300, / / High threshold for Canny edge detection

[0056] param2:30, / / Threshold of the center accumulator

[0057] minRadius:200, / / minimum radius

[0058] maxRadius:220 / / Maximum radius (if 0, it will be automatically adjusted according to the image size)

[0059] ).

[0060] S6. Based on all the above processing results and the processed image, the detected feature center is taken as the center, and the ratio of the circle radius and size is set to plan the distance range of the feature center where the corresponding straight edge appears, and the required straight edge (such as Figure 2 The blue lines in the figure are the identified straight edges).

[0061] S7. Establish a rectangular coordinate system with the center of the characteristic circle obtained in the result of S5 as the origin.

[0062] S8. After filtering the straight edges, the slope of the filtered straight edges in the rectangular coordinate system is calculated. If the slope error exceeds the specified error (the error threshold is 1°), the recognition result is considered invalid. If the slope errors of the identified straight lines are all within the range, the samples with the largest and smallest deviations are removed, and the average slope is calculated.

[0063] S9. Cover the landing points of all identified straight lines in the four directions of up, down, left and right of the characteristic circle, convert the slope of the straight line into an absolute angle coordinate with the positive direction of the Y-axis as 0° and the clockwise rotation as a positive angle, return the absolute center angle of the angle between the vertical line of the straight edge and the positive direction of the Y-axis as the angle that the pallet needs to rotate, and then send the angle to the rotating mechanism through TCP protocol communication to control the pallet to perform regression rotation.

Claims

1. A method for identifying a rotation angle of a pallet, characterized in that: The following steps are involved: S1, collecting an image of the pallet through a camera, the image being a circular surface of the pallet; S2, blur the image to reduce noise; S3, performing morphological corrosion processing on the image, using a structural element to scan the image, and judging the relationship between the area covered by the structural element and the foreground pixel. If the structural element is completely contained in the foreground pixel area, the pixel position corresponding to the center of the structural element is marked as the foreground pixel in the corroded image; if the structural element is not completely contained in the foreground pixel area, the center pixel position is marked as the background pixel; S4, performing morphological dilation processing on the image, when the structure element and the foreground pixel have an intersection, the pixel corresponding to the center of the structure element is marked as the foreground pixel, and the structure element moves on the image. As long as there is a white pixel in the area covered by the structure element, the pixel corresponding to the center position of the structure element is set to white in the dilated image; when the area covered by the structure element is all black pixels, the center position pixel is set to black in the dilated image; S5, using gradient Hough circle transform to extract the center of the inner circle of the image after the above processing as the feature center; S6. Based on all the above processing results and the processed image, taking the detected feature center as the center and setting the ratio of the circle radius and size, planning the distance range from the feature center where the corresponding straight edge appears, using Hough line detection to identify the required straight edge; S7, establishing a rectangular coordinate system with the characteristic circle center in the result of S5 as the origin; S8, after filtering the straight edges, the slope of the filtered straight edges in the rectangular coordinate system is calculated. If the slope error exceeds the specified error, the recognition result is considered invalid. If the slope errors of the recognized straight lines are all within the range, the samples with the largest and smallest deviations are removed, and the average slope is calculated. S9. Cover the landing points of all identified straight lines in the four directions of up, down, left and right of the characteristic circle, convert the slope of the straight line into an absolute angle coordinate with the positive direction of the Y-axis as 0° and the clockwise rotation as a positive angle, and return the absolute center angle of the angle between the vertical line of the straight edge and the positive direction of the Y-axis as the angle at which the pallet needs to rotate.

2. The method for identifying the rotation angle of a pallet according to claim 1, characterized in that: In S1, the camera sets a variety of shooting parameter solutions including exposure time, contrast, and sensitivity. If the feature center of the image after the above processing fails to be extracted using the gradient Hough circle transform in S5, then return to S1 to change the next set of shooting parameter solutions to collect the tray image.

3. The method for identifying the rotation angle of a pallet according to claim 2, characterized in that: In S3, morphological corrosion processing is performed on the image to remove objects smaller than the structural element in the image.

4. The method for identifying the rotation angle of a pallet according to claim 3, characterized in that: In S4, morphological dilation processing is performed on the image to highlight the center of the circle and the straight edge features in the image.

5. The method for identifying the rotation angle of a pallet as claimed in claim 2, characterized in that: If no feature center is detected or multiple feature centers are detected in S5, it is considered that the extraction of the inner circle center of the image after the above processing by using the gradient Hough circle transform fails.

6. The method for identifying the rotation angle of a pallet according to claim 5, characterized in that: The error threshold specified in S8 is 1°.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images