Method for measuring diameter of crystal-pulling crystal bar and measuring device applying method

Abstract

The invention relates to a crystal pulling crystal bar diameter measurement method and a measurement device using the same, and solves the problems of low measurement precision, complex operation, easy influence by human factors and the like caused by manual observation or simple mechanical measurement in the prior art. Processing the target image to generate a target image grey-scale map; analyzing the grey-scale map of the target image, and finding out the contour edge of the target image; generating a contour point set of the target image from the contour edge; fitting a target circle based on the contour point set; and obtaining the seeding diameter according to the target circle. The method has the advantages that the camera arranged in the crystal pulling furnace captures a real-time image in the seeding process, the contour line of the seeding interface is extracted through the image processing technology, and accurate measurement of the seeding diameter is achieved; automatic measurement is realized, the degree of dependence on operators is reduced, the generation efficiency is improved, direct contact between the operators and dangerous environments such as high temperature and high pressure is reduced, and the safety risk is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of semiconductor material processing, and is mainly applied to the technical field of photovoltaic and semiconductor single crystal pulling, and in particular to a method for measuring the diameter of a crystal pulling rod and a measuring device using the method. Background Art

[0002] Crystal pulling refers to the process of melting high-purity silicon raw materials at high temperature and immersing single-crystal silicon seed crystals into the melt. With the control of temperature and the pulling of the seed crystals, the melt gradually solidifies around the seed crystals and grows into a complete single-crystal silicon rod. This crystal pulling method can produce high-purity, high-quality single-crystal silicon materials, which are widely used in semiconductors and solar energy.

[0003] In the process of pulling semiconductor crystals, the seed crystal and the melt are first brought into contact, so that the silicon melt at the solid-liquid interface cools and crystallizes along the seed crystal. The process of cooling and crystallizing the seed crystal is called seeding. The seeding diameter is an important process parameter that directly affects the quality and growth efficiency of the crystal. However, most existing seeding diameter measurement methods are based on manual observation or simple mechanical measurement, which have problems such as low measurement accuracy, complex operation, and susceptibility to human factors. Summary of the invention

[0004] The purpose of the present invention is to solve the above-mentioned problems existing in the prior art and to provide a method for measuring the diameter of a crystal pulling rod and a measuring device using the same. A camera is arranged in a crystal pulling furnace to capture real-time images of the crystal seeding process, and image processing and other technologies are used to extract the contour lines of the crystal seeding interface. Accurate measurement of the crystal seeding diameter is achieved through image analysis, etc.; automated measurement is achieved, the degree of dependence on operators is reduced, production efficiency is improved, direct contact between operators and dangerous environments such as high temperature and high pressure is reduced, and safety risks are reduced.

[0005] The above technical objectives of the present invention are mainly solved by the following technical solutions:

[0006] The technical solution of the first technical subject of the present invention is: a method for measuring the diameter of a crystal pulling rod, characterized in that:

[0007] Acquire target images;

[0008] Process the target image and generate a grayscale image of the target image;

[0009] Analyze the grayscale image of the target image and find the contour edge of the target image;

[0010] Generate a contour point set of the target image from the contour edge;

[0011] Fitting a target circle based on the contour point set;

[0012] The seeding diameter is obtained according to the target circle.

[0013] As a further improvement and supplement to the above technical solution, the present invention adopts the following technical measures:

[0014] Preferably, an image recognition module is trained based on a number of image samples annotated with edge information, and the target image is recognized based on the image recognition module.

[0015] Preferably, the target image is optimized based on the Canny() function in the edge detection model OpenCV to find the edge of the target image.

[0016] Preferably, the contour point set of the target image is generated based on the findContours() function in the edge detection model OpenCV.

[0017] Preferably, based on the minEnclosingTriangle() function in the edge detection model OpenCV, the area of ​​each triangle in the contour point set of the target image is calculated through the convex hull algorithm, an area set is generated, the triangle with the minimum area in the target image, that is, the minimum convex hull triangle, is found, and the area set is returned, the sizes of the minimum convex hull triangles are compared, and the contour point set of the target image corresponding to the largest minimum convex hull triangle is retained.

[0018] Preferably, based on the obtained contour point set of the target image, its circumscribed polygon is found by using a convex hull algorithm, and returned to the contour point set of the target image; based on the contour point set of the target image, the target circle is fitted by using a fitting circle model.

[0019] Preferably, the target image includes: a first side edge image corresponding to the lower edge of the crystal rod, and a second side edge image of the lower edge of the crystal rod projected onto the target molten silicon liquid surface; the target circle is a corresponding target circle fitted with the contour point sets of the first side edge image and the second side edge image respectively, and the diameters of the target circles are averaged to obtain the seeding diameter.

[0020] Preferably, the pixel distance in a predetermined direction between predetermined pixel points in the first side edge image and the second side edge image and the pixel point in the center of the field of view of the image acquisition module is obtained based on the image acquisition module, and based on the pixel distance, the radius value r of the crystal rod is calculated, wherein the predetermined pixel point and the pixel point in the center of the field of view are on the same straight line extending along the predetermined direction.

[0021] Preferably, the radius r of the target circle is: 2 =(xA) 2 +(yB) 2 ,

[0022] Let a = -2A, b = -2B, c = A2 + B2 - r 2 ,

[0023] get:

[0024] Wherein: A is the horizontal coordinate value of the center of the target circle, and B is the vertical coordinate value of the center of the target circle.

[0025] Preferably, the target image is captured by an industrial camera, and a spatial coordinate system is established based on the spatial position of the camera and the processed target image.

[0026] The technical solution of the second technical subject of the present invention is: a measuring device using the method for measuring the diameter of a crystal pulling rod, characterized in that it uses the aforementioned method for measuring the diameter of a crystal pulling rod, and comprises:

[0027] An image acquisition unit, which uses a camera to collect a target image, wherein the target image includes: an image of a seed crystal of a crystal rod, and an image of a lower edge of a target crystal rod seed crystal in contact with a target molten silicon liquid surface;

[0028] An image processing unit is used to connect with the image acquisition unit, and use an image recognition model to recognize a target image, wherein the target image includes a first side edge image corresponding to the lower edge of the crystal rod and a second side edge image projected from the lower edge of the crystal rod to the target molten silicon liquid surface; the target image is processed to generate a grayscale image of the target image;

[0029] A data analysis unit analyzes the grayscale image of the target image, finds the contour edge of the target image, and generates a contour point set of the target image from the contour edge;

[0030] A target image determination unit, used for determining a circle to be fitted in the target image, and fitting the target circle based on the contour point set;

[0031] A diameter determination subunit determines a seeding diameter based on the target circle;

[0032] A controller unit, connected to the image processor, for receiving the processing result of the image processor and controlling the shooting parameters of the image acquisition device;

[0033] A display unit, connected to the controller device, for displaying the measurement result of the seeding diameter;

[0034] A memory unit; used for connecting with the controller device and storing the measurement results.

[0035] The present invention has the beneficial effects:

[0036] 1. Improve measurement accuracy: The camera installed in the crystal pulling furnace captures the real-time image of the seeding process, and uses image processing technology to extract the contour line of the seeding interface to achieve accurate measurement of the seeding diameter. Combined with multiple measurements and calculations, the average value of the seeding diameter is obtained, which effectively reduces the error that may be caused by a single measurement and improves the overall measurement accuracy.

[0037] 2. Enhanced real-time performance: It can obtain image data of the seeding process in real time, and process and analyze it, so as to realize real-time monitoring and adjustment of the seeding diameter; it helps to timely discover and solve problems that may arise in the crystal pulling process, and ensure the stability of the production process and product quality.

[0038] 3. Simplify the operation process: Compared with the traditional manual measurement method, the present invention realizes automated measurement and reduces the dependence on operators.

[0039] 4. The device has a simple structure and is easy to operate, which reduces the frequency of operators entering the production workshop and the frequency of operating near the crystal pulling furnace, thereby reducing the difficulty of operation and labor intensity.

[0040] 5. Improve production efficiency: Automated measurement reduces human intervention and improves production efficiency; real-time monitoring and adjustment of seeding diameter helps optimize the crystal pulling process, reduce raw material waste and crystal rod scrap rate, and improve product qualification rate and production efficiency.

[0041] 6. Improve production safety: Reduce the direct contact of operators with dangerous environments such as high temperature and high pressure, reducing safety risks; precise control of crystal pulling diameter helps prevent waste of raw materials caused by excessive crystal pulling diameter or scrapping or degradation of crystal rods caused by too small crystal pulling diameter, further improving production safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a schematic flow chart of a method of the present invention.

[0043] Figure 2 It is a schematic diagram of a target image photographed by the industrial camera of the present invention.

[0044] Figure 3 yes Figure 2 Schematic diagram of the grayscale image of the target image.

[0045] Figure 4 is Figure 3 A schematic diagram of the target circle (shown by the red circle) is shown on this basis.

[0046] Figure 5 is Figure 3 Based on this, a schematic diagram of a first side edge image and a second side edge image is illustrated.

[0047] In the figure: 1. crystal rod; 2. target molten silicon liquid level; 3. target circle; 4. first side edge image; 5. second side edge image. DETAILED DESCRIPTION

[0048] The technical solution of the present invention is further specifically described below through embodiments and in conjunction with the accompanying drawings.

[0049] Example 1: Figure 1-Figure 5 As shown, the technical solution of the first technical subject of the present invention is:

[0050] A method for measuring the diameter of a crystal pulling rod:

[0051] The target image is acquired by taking images of the crystal rod 1 of the seeding and the contact portion between the lower edge of the crystal rod and the target molten silicon liquid surface 2 through an industrial camera arranged in the crystal pulling furnace;

[0052] Set ROI (region of interest), select the part that needs to be processed and identified (target image), process the acquired target image through the image processing module that interacts with the image acquisition module, remove some influencing factors, get the desired part, and then convert the image into a grayscale image by changing the color space, thereby generating a grayscale image of the target image, which can reduce the amount of original image data and facilitate less calculation in subsequent processing;

[0053] Analyze the grayscale image of the target image and use the edge detection method to find the contour edge of the target image;

[0054] Generate a contour point set of the target image from the contour edge;

[0055] Fitting a target circle 3 based on the contour point set;

[0056] The seeding diameter is obtained according to the target circle.

[0057] Then further improve and supplement the above technical solutions:

[0058] In practical applications, an image recognition module is trained based on a number of image samples annotated with edge information, and the target image is recognized based on the image recognition module.

[0059] In practical applications, the target image is optimized based on the Canny() function in the edge detection model OpenCV to find the edge of the target image. The Canny() function is a powerful edge detection tool in OpenCV. Through reasonable parameter setting and adjustment, the noise in the target image is removed, and the contact boundary between the lower edge of the target image and the molten silicon liquid surface is made more obvious, which can reduce the influence of the molten silicon liquid surface fluctuation and the unclear projection of the lower edge of the target image on the determination of the seeding diameter, and can achieve efficient edge detection effect.

[0060] That is, an image recognition module is used to identify a first side edge image 4 corresponding to the lower edge of the target crystal rod in the target image and a second side edge image 5 projected from the lower edge of the crystal rod to the target molten silicon liquid surface. The image recognition model is trained using a plurality of image samples annotated with edge information. The edge information of the target image includes a first side edge corresponding to the lower edge of the crystal rod in the image sample and a second side edge projected from the lower edge of the crystal rod to the target molten silicon liquid surface.

[0061] In practical applications, the contour point set of the target image is generated based on the findContours() function in the edge detection model OpenCV.

[0062] In practical applications, the minEnclosingTriangle() function in OpenCV based on the edge detection model calculates the area of ​​each triangle in the contour point set of the target image through the convex hull algorithm, generates an area set, finds the triangle with the smallest area in the target image, that is, the minimum convex hull triangle, and returns the area set, compares the size of the minimum convex hull triangle, and retains the contour point set of the target image corresponding to the largest minimum convex hull triangle. This operation can remove unnecessary edge lines and improve the accuracy of seeding diameter measurement.

[0063] In practical applications, based on the obtained contour point set of the target image, its circumscribed polygon is found by using a convex hull algorithm and returned to the contour point set of the target image; based on the contour point set of the target image, the target circle is fitted by using a fitting circle model.

[0064] In actual applications, the target image includes: a first side edge image corresponding to the lower edge of the crystal rod, and a second side edge image of the lower edge of the crystal rod projected onto the target molten silicon liquid surface. The target circle is a corresponding target circle fitted from the contour point sets of the first side edge image and the second side edge image respectively, and the average diameter of each target circle is taken to obtain the seeding diameter.

[0065] In practical applications, the pixel distance in a predetermined direction between a predetermined pixel point in the first side edge image and the second side edge image and a pixel point in the center of the field of view of the image acquisition module is acquired based on the image acquisition module, and based on the pixel distance, the radius value r of the crystal rod is calculated, wherein the predetermined pixel point and the pixel point in the center of the field of view are on the same straight line extending along the predetermined direction.

[0066] In the process of seeding diameter measurement, factors such as molten silicon liquid level fluctuation and unclear projection will affect the accuracy of seeding diameter. At present, the seeding diameter is calculated only by pixel distance and set coefficients, which can easily lead to inaccurate seeding diameter. Based on this, the present invention combines the spatial position of the industrial camera and the processed target image to establish a spatial coordinate system, and combines the principle of optical imaging to achieve the purpose of more accurate measurement of seeding diameter.

[0067] In practical applications, the radius r of the target circle is: 2 =(xA) 2 +(yB) 2 ,

[0068] Let a = -2A, b = -2B, c = A2 + B2 - r 2 ,

[0069] get:

[0070] Wherein: A is the horizontal coordinate value of the center of the target circle, and B is the vertical coordinate value of the center of the target circle.

[0071] In practical applications, the average value of the seeding diameter is obtained through multiple measurements and calculations to improve the measurement accuracy.

[0072] In summary, the method for measuring the diameter of a crystal pulling rod involved in the present invention includes: obtaining a target image captured by an industrial camera; identifying the target image using an image recognition model; based on the identified target image, obtaining the target image to be fitted using image processing; then fitting the required target circle through an algorithm; obtaining the radius r through a calculation formula based on the target image; and then obtaining the average value of the crystal seeding diameter through multiple measurements and calculations to improve the measurement accuracy.

[0073] Embodiment 2: The technical solution of the second technical subject of the present invention: a measuring device using the method for measuring the diameter of a crystal rod pulled, characterized in that it uses the method for measuring the diameter of a crystal rod pulled described in Embodiment 1, comprising:

[0074] The image acquisition unit uses a camera to collect a target image, wherein the target image includes: an image of a seed crystal of a crystal rod, and an image of a lower edge of a target crystal rod seed crystal in contact with a target molten silicon liquid surface.

[0075] The image processing unit is used to connect with the image acquisition unit, and use the image recognition model to recognize the target image, wherein the target image includes a first side edge image corresponding to the lower edge of the crystal rod and a second side edge image of the lower edge of the crystal rod projected onto the target molten silicon liquid surface; the target image is processed to generate a grayscale image of the target image. Processing the target image includes preprocessing and image processing, and the preprocessing includes denoising, graying, convex hull, fitting and other processing to improve the accuracy of image recognition.

[0076] The data analysis unit analyzes the grayscale image of the target image and finds the contour edge of the target image; the contour edge generates a contour point set of the target image.

[0077] The target image determination unit is used to determine a circle to be fitted in the target image and fit the target circle based on the contour point set.

[0078] A diameter determination unit, which determines a seeding diameter based on the target circle;

[0079] A controller unit, connected to the image processor, for receiving the processing result of the image processor and controlling the shooting parameters of the image acquisition device;

[0080] A display unit, connected to the controller device, for displaying the measurement result of the seeding diameter;

[0081] A memory unit; used for connecting with the controller device and storing the measurement results.

[0082] The diameter determination unit is used to calculate the radius r of the target circle:

[0083] The radius r of the target circle: r 2 =(xA) 2 +(yB) 2 ,

[0084] Let a = -2A, b = -2B, c = A2 + B2 - r 2 ,

[0085] get:

[0086] Wherein: A is the horizontal coordinate value of the center of the target circle, and B is the vertical coordinate value of the center of the target circle.

[0087] In practical applications, the crystal rod diameter measuring device also includes an image filtering subunit, which is used to perform image filtering on the target image to remove noise in the target image before the image segmentation unit uses the image segmentation model to identify the first side edge image and the second side edge image.

[0088] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. In the above embodiments, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, 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 measuring the diameter of a crystal pulling rod, characterized in that: Acquire target images; Process the target image and generate a grayscale image of the target image; Analyze the grayscale image of the target image and find the contour edge of the target image; Generate a contour point set of the target image from the contour edge; Fitting a target circle based on the contour point set; The seeding diameter is obtained according to the target circle.

2. The method for measuring the diameter of a crystal pulling rod according to claim 1, characterized in that An image recognition module is obtained by training based on a plurality of image samples annotated with edge information, and the target image is recognized based on the image recognition module.

3. The method for measuring the diameter of a crystal pulling rod according to claim 2, characterized in that The target image is optimized based on the Canny() function in the edge detection model OpenCV to find the edge of the target image.

4. The method for measuring the diameter of a crystal pulling rod according to claim 3, characterized in that The contour point set of the target image is generated based on the findContours() function in the edge detection model OpenCV.

5. The method for measuring the diameter of a crystal pulling rod according to claim 4, characterized in that Based on the edge detection model, the minEnclosingTriangle() function in OpenCV calculates the area of ​​each triangle in the contour point set of the target image through the convex hull algorithm, generates an area set, finds the triangle with the minimum area in the target image, that is, the minimum convex hull triangle, and returns the area set, compares the size of the minimum convex hull triangle, and retains the contour point set of the target image corresponding to the largest minimum convex hull triangle; Based on the obtained contour point set of the target image, the circumscribed polygon is found by using a convex hull algorithm, and the polygon is returned to the contour point set of the target image; based on the contour point set of the target image, the target circle is fitted by using a fitting circle model.

6. The method for measuring the diameter of a crystal pulling rod according to any one of claims 1 to 5, characterized in that The target image includes: a first side edge image corresponding to the lower edge of the crystal rod, and a second side edge image projected from the lower edge of the crystal rod to the target molten silicon liquid surface. The target circle is a corresponding target circle fitted with the contour point sets of the first side edge image and the second side edge image respectively, and the average diameter of each target circle is taken to obtain the seeding diameter.

7. The method for measuring the diameter of a crystal pulling rod according to claim 6, characterized in that A pixel distance in a predetermined direction between a predetermined pixel point in the first side edge image and the second side edge image and a pixel point at the center of the field of view of the image acquisition module is acquired based on the image acquisition module, and a radius value r of the crystal rod is calculated based on the pixel distance, wherein the predetermined pixel point and the pixel point at the center of the field of view are on the same straight line extending along the predetermined direction.

8. The method for measuring the diameter of a crystal pulling rod according to claim 7, characterized in that The radius r of the target circle: r 2 =(xA) 2 +(yB) 2 , Let a = -2A, b = -2B, c = A2 + B2 - r 2 , get: Wherein: A is the horizontal coordinate value of the center of the target circle, and B is the vertical coordinate value of the center of the target circle.

9. The method for measuring the diameter of a crystal pulling rod according to claim 8, characterized in that The target image is captured by an industrial camera, and a spatial coordinate system is established based on the spatial position of the camera and the processed target image.

10. A measuring device using a method for measuring the diameter of a crystal pulling rod, characterized in that The method for measuring the diameter of a crystal pulling rod according to any one of claims 1 to 9 is applied, comprising: An image acquisition unit, which uses a camera to collect a target image, wherein the target image includes: an image of a seed crystal of a crystal rod, and an image of a lower edge of a target crystal rod seed crystal in contact with a target molten silicon liquid surface; An image processing unit is used to connect with the image acquisition unit, and use an image recognition model to recognize a target image, wherein the target image includes a first side edge image corresponding to the lower edge of the crystal rod and a second side edge image projected from the lower edge of the crystal rod to the target molten silicon liquid surface; the target image is processed to generate a grayscale image of the target image; The data analysis unit analyzes the grayscale image of the target image and finds the contour edge of the target image; the contour edge generates a contour point set of the target image; A target image determination unit, used for determining a circle to be fitted in the target image, and fitting the target circle based on the contour point set; A diameter determination unit, which determines a seeding diameter based on the target circle; A controller unit, connected to the image processor, for receiving the processing result of the image processor and controlling the shooting parameters of the image acquisition device; A display unit, connected to the controller device, for displaying the measurement result of the seeding diameter; A memory unit; used for connecting with the controller device and storing the measurement results.

Images