Crystal defect identification system and method based on feature analysis
Through multi-dimensional feature analysis and synthesis of crystal image data and growth environment data, the problem of low defect recognition efficiency and accuracy in the prior art is solved, and efficient and accurate defect type recognition is achieved.
Patent Information
- Application Number
- CN202411287937.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-14
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-09-14
AI Technical Summary
The prior art cannot perform image analysis on crystal images to obtain multi-dimensional image implicit features of reaction defect types, and at the same time, it cannot be comprehensively analyzed with the crystal growth process environmental data to accurately identify defect features, resulting in low classification efficiency and accuracy.
By obtaining crystal image data and growth process environmental data, feature extraction, preliminary identification and secondary identification are performed, combined with environmental impact analysis, and output defect type identification results. Specific steps include data collection, defect feature extraction, preliminary identification, environmental impact analysis and final identification, and use multi-dimensional features and environmental data to improve identification accuracy.
The technical effect of improving the accuracy and efficiency of defect recognition is achieved, and the defect types are accurately identified by performing multi-dimensional feature analysis and environmental data synthesis on crystal images.
Smart Images

Figure CN118799325B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of feature analysis, and in particular relates to a crystal defect recognition system and method based on feature analysis. Background Art
[0002] A crystal is a solid whose atoms, molecules or ions are repeatedly arranged in a regular geometric arrangement in three-dimensional space, forming a long-range ordered structure. This ordered structure is the main feature that distinguishes crystals from amorphous solids (such as glass) and liquids. The basic structural unit in a crystal is called a unit cell, which is repeatedly arranged in space according to a certain rule to form the structure of the entire crystal. The crystal will melt at a specific temperature, which is called the melting point. For pure crystals, the melting point is fixed. Crystals usually have broad application prospects in the semiconductor and optoelectronic industries due to their unique physical and chemical properties. With the continuous advancement of technology and the optimization of manufacturing processes, the performance and application range of crystal devices are expected to be further expanded;
[0003] During the crystal growth process, growth defects may occur due to the abnormal influence of its growth environment. However, there are many kinds of existing crystal defects, which have a significant impact on the performance of crystal defect materials. For example, point defects can increase the hardness and strength of materials; dislocation movement is the main cause of plastic deformation; irregular atomic arrangement at grain boundaries may affect the corrosion behavior and mechanical properties of materials. Therefore, the study of crystal defects is of great significance to materials science and engineering. The existing technology classifies and identifies crystal defects through simple image processing, and cannot perform image analysis on crystal images to obtain multi-dimensional image implicit features that reflect the types of defects. At the same time, it cannot be comprehensively analyzed with the environmental data of the crystal growth process to accurately identify the defect characteristics, resulting in low classification efficiency and accuracy. Most of the existing technologies have the above problems;
[0004] For example, a batch processing method for defect identification of an X-ray orienter is proposed in a Chinese patent with the authorization publication number CN109034262B, which belongs to the field of single crystal material processing, and proposes an offline batch single crystal defect detection method, designs a characteristic vector of the single crystal swing curve to abstract the curve characteristics, combines the particle swarm optimization algorithm, and introduces an effective radius based on the density function, improves the traditional FCM algorithm, improves the robustness of the algorithm for initializing the cluster center, avoids falling into the local optimum, and filters the interference data well, thereby realizing clustering of batch data, and then only needs to perform defect detection on the characteristic vector of the cluster center, and the defect type of all the curves to be tested can be obtained according to the membership relationship, which includes the improved fuzzy transitive closure clustering algorithm proposed by the present invention, defines the fuzzy similarity matrix, and ensures the accuracy of the similarity calculation. The invention provides a new idea and implementation method for crystal detection technology;
[0005] The above patents all have the problems raised by this background technology. In order to solve the problems raised by this background technology, this application designs a crystal defect identification system and method based on feature analysis. Summary of the invention
[0006] In view of the deficiencies in the prior art, the present invention proposes a crystal defect identification system and method based on feature analysis.
[0007] To achieve the above object, the present invention provides the following technical solution: a crystal defect identification method based on feature analysis, which comprises the following specific steps:
[0008] Acquire crystal image data to be analyzed and environmental data during crystal growth;
[0009] Import the crystal image data to be analyzed into the feature extraction model to extract defect features;
[0010] Importing the extracted defect features into the constructed defect feature preliminary identification model to perform preliminary identification of the defect features;
[0011] Importing environmental data from the crystal growth process into the environmental impact analysis model for environmental impact analysis;
[0012] Import the environmental impact analysis results and the preliminary identification results of defect features into the defect feature secondary identification model to identify the defect types;
[0013] The obtained defect type identification result is output to the user end.
[0014] It should be noted that, as a preferred technical solution of the crystal defect identification method based on feature analysis, the specific steps of obtaining the crystal image data to be analyzed and the environmental data during the crystal growth process are:
[0015] S11, placing a crystal to be defect-analyzed in an electron microscope imaging module to obtain microscopic images of various parts of the crystal, and displaying a three-dimensional model of the crystal in an image display module;
[0016] S12. Extracting the stored environmental data of the crystal during the growth process for defect analysis through the data acquisition terminal. The environmental data during the growth process includes temperature data, humidity data and pressure data.
[0017] It should be noted that, as a preferred technical solution of the crystal defect recognition method based on feature analysis, the step of importing the crystal image data to be analyzed into the feature extraction model to extract defect features includes the following specific steps:
[0018] S21, obtaining the pixel value of each pixel point of the microscopic image of each part of the crystal, and at the same time obtaining the average pixel value of the microscopic image of the corresponding part of the normal crystal, obtaining the absolute value of the pixel difference between the pixel value of each pixel point of the microscopic image of each part of the crystal and the average pixel value of the microscopic image of the corresponding part of the normal crystal, obtaining the absolute value of the pixel difference of the corresponding pixel point, obtaining the pixel point whose absolute value of the pixel difference is greater than or equal to the pixel difference threshold and setting it as the abnormal pixel point, obtaining the pixel point corresponding to the largest absolute value of the pixel difference and setting it as the central abnormal pixel point, obtaining the regional image composed of the abnormal pixels, setting it as the defective regional image, and obtaining the contour features and average pixel value features of the defective regional image;
[0019] S22, obtaining a corresponding defect area image, taking the central abnormal pixel point as the center, dividing a number of pixel abnormal areas according to a set pixel gradient, obtaining contour features of the obtained pixel abnormal areas, and obtaining an average value of the distance between the edge of the pixel abnormal area and the central abnormal pixel point.
[0020] It should be noted that, as a preferred technical solution of the crystal defect identification method based on feature analysis, the method of importing the extracted defect features into the constructed defect feature preliminary identification model for preliminary identification of defect features includes the following specific steps:
[0021] S31, obtaining contour features and average pixel value features of the defect area image, and simultaneously obtaining contour features and average pixel value features of defects that have been judged in the past;
[0022] S32, importing the obtained contour features of the defect area image and the contour features of the determined defect into a contour similarity value calculation formula to calculate the contour similarity value, wherein the contour similarity value calculation formula between the j-th determined defect and the defect area image is: , where m() is the area of the contour in the brackets, c is the contour feature of the defect area image, and cj is the contour feature of the jth identified defect image. is the intersection of the contours, is the union of contours;
[0023] S33, substituting the obtained contour similarity value, the average pixel value feature of the defect area image and the average pixel value feature of the determined defect into the overall similarity value calculation formula to calculate the overall similarity value, wherein the overall similarity value calculation formula between the j-th determined defect and the defect area image is: , where a is the contour ratio, z is the average pixel value of the defect area image, and zj is the average pixel value of the jth determined defect.
[0024] It should be noted here that as a preferred technical solution for the crystal defect recognition method based on feature analysis, the method of importing the extracted defect features into the constructed defect feature preliminary recognition model for preliminary recognition of defect features also includes the following specific contents: S34, obtaining the contour features of each pixel abnormal area of the defect area and the average value of the distance of the edge relative to the central abnormal pixel point, and at the same time obtaining the contour features of each pixel abnormal area of the determined defect, and importing them into the pixel abnormal area contour similarity value calculation formula to calculate the pixel abnormal area contour similarity value, wherein the pixel abnormal area contour similarity value calculation formula of the jth determined defect and the defect area image is: , where n is the number of pixel abnormal regions in the defect area, ki is the contour feature of the i-th pixel abnormal region in the defect area, Lij is the contour feature of the i-th pixel abnormal region corresponding to the j-th determined defect, and Di is the average distance between the edge of the i-th pixel abnormal region in the defect area and the central abnormal pixel point;
[0025] S35, obtaining the calculated overall similarity value between the determined defect and the defective region image, and the similarity value between the determined defect and the pixel abnormal region contour of the defective region image, and substituting them into the preliminary recognition value calculation formula to calculate the preliminary recognition value of the determined defect and the defective region image, wherein the preliminary recognition value calculation formula is: , where Nj is the initial recognition value of the jth identified defect and defect area image, is the contour similarity ratio coefficient;
[0026] S36, arranging the calculated preliminary identification values in descending order, obtaining the determined defect types corresponding to the preliminary identification values arranged in the first five items and setting them as the preliminary identification defect types.
[0027] It should be noted here that, as a preferred technical solution of the crystal defect identification method based on feature analysis, the environmental data during the crystal growth process is imported into the environmental impact analysis model for environmental impact analysis, which includes the following specific contents: obtaining the environmental data during the growth process of the crystal corresponding to the defect type initially identified and the environmental data during the growth process of the crystal for defect analysis, and importing them into the environmental impact analysis value calculation formula to calculate the environmental impact analysis value, wherein the environmental impact analysis value calculation formula is: , T is the crystal growth time, V is the type of environmental data, wv is the proportion coefficient of the vth environmental data, dt is the time integral, pvt is the vth environmental data at the growth time t corresponding to the preliminary identification of the defect type, and pvtm is the vth environmental data at the growth time t during the growth process of the crystal for defect analysis.
[0028] It should be noted that, as a preferred technical solution of the crystal defect identification method based on feature analysis, the method of importing the environmental impact analysis results and the preliminary identification results of the defect features into the defect feature secondary identification model for secondary defect feature identification includes the following specific contents:
[0029] Substitute the environmental impact analysis value and preliminary identification value of the defect type obtained in the preliminary identification into the final identification value calculation formula to calculate the final identification value, where the final identification value calculation formula is: , where exp() is the power of the natural constant e, and the preliminary identified defect type corresponding to the obtained maximum final identification value is set as the defect type of the crystal to be analyzed.
[0030] A crystal defect recognition system based on feature analysis, which is implemented based on the above-mentioned crystal defect recognition method based on feature analysis, and specifically includes a data acquisition module, a defect feature extraction module, a defect feature preliminary recognition module, an environmental impact analysis module, a defect type recognition module and a defect type output module, wherein the data acquisition module is used to obtain crystal image data to be analyzed and environmental data during the crystal growth process;
[0031] The defect feature extraction module is used to import the crystal image data to be analyzed into the feature extraction model to extract the defect features;
[0032] The defect feature preliminary identification module is used to import the extracted defect features into the constructed defect feature preliminary identification model to perform preliminary identification of the defect features;
[0033] The environmental impact analysis module is used to import the environmental data during the crystal growth process into the environmental impact analysis model to perform environmental impact analysis;
[0034] The defect type identification module is used to import the environmental impact analysis results and the preliminary identification results of the defect characteristics into the defect characteristic secondary identification model to identify the defect type;
[0035] The defect type output module is used to output the obtained defect type identification result to the user end.
[0036] An electronic device comprises: a processor and a memory, wherein the memory stores a computer program that can be called by the processor;
[0037] The processor executes the above-mentioned crystal defect identification method based on feature analysis by calling the computer program stored in the memory.
[0038] A computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned crystal defect identification method based on feature analysis.
[0039] Compared with the prior art, the present invention has the following beneficial effects:
[0040] The present invention obtains crystal image data that needs to be analyzed and environmental data during the crystal growth process, imports the crystal image data that needs to be analyzed into a feature extraction model to extract defect features, imports the extracted defect features into a constructed defect feature preliminary identification model to perform preliminary identification of the defect features, imports the environmental data during the crystal growth process into an environmental impact analysis model to perform environmental impact analysis, imports the environmental impact analysis results and the preliminary identification results of the defect features into a defect feature secondary identification model to perform defect type identification, and outputs the obtained defect type identification results to a user end. The present application obtains multi-dimensional image implicit features that reflect the defect types by performing image analysis on crystal images, and at the same time comprehensively analyzes the environmental data of the crystal growth process to accurately identify the defect features, thereby achieving a technical effect of improving the accuracy and efficiency of defect identification. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Figure 1 It is a schematic diagram of the overall process of the crystal defect identification method based on feature analysis of the present invention;
[0042] Figure 2 It is a schematic flow chart of step S3 of the crystal defect identification method based on feature analysis of the present invention;
[0043] Figure 3 It is a schematic diagram of the overall framework of the crystal defect recognition system based on feature analysis of the present invention;
[0044] Figure 4 It is a data transmission schematic diagram of the present invention. DETAILED DESCRIPTION
[0045] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. The following description of at least one exemplary embodiment is actually only illustrative and is by no means intended to limit the present application and its application or use.
[0046] Example 1
[0047] To solve the technical problems raised in the background technology: the existing technology classifies and identifies crystal defects through simple image processing, and is unable to perform image analysis on crystal images to obtain multi-dimensional image implicit features that reflect the types of defects, and is also unable to perform comprehensive analysis with crystal growth process environmental data to accurately identify defect features, resulting in low classification efficiency and accuracy;
[0048] The implementation scenario of this embodiment is as follows Figure 4 As shown, specifically: a number of data acquisition terminals collect crystal image data to be analyzed and environmental data during the crystal growth process, and the data analysis terminal obtains multi-dimensional image implicit features reflecting the defect types by performing image analysis on the crystal image, and simultaneously comprehensively analyzes with the environmental data during the crystal growth process to accurately identify the defect features, and the user terminal is used to display the obtained defect type identification results;
[0049] The present invention provides a preferred embodiment: Figure 1-Figure 2 As shown, the crystal defect identification method based on feature analysis includes the following specific steps:
[0050] S1, obtaining crystal image data to be analyzed and environmental data during crystal growth;
[0051] In a specific implementation of this embodiment, the specific steps of obtaining the crystal image data to be analyzed and the environmental data during the crystal growth process are:
[0052] S11, placing a crystal to be defect-analyzed in an electron microscope imaging module to obtain microscopic images of various parts of the crystal, and displaying a three-dimensional model of the crystal in an image display module;
[0053] To analyze the defects of the crystal, use the electron microscope imaging module to obtain the microscopic image of the crystal, and display the three-dimensional model of the crystal in the image display module, which is a conventional technical means of three-dimensional reconstruction in the prior art. The specific steps are as follows:
[0054] S111, acquiring an image:
[0055] Use a scanning electron microscope (SEM) or a transmission electron microscope (TEM) to scan the surface or internal structure of the crystal; adjust the imaging mode and contrast to obtain clear microscopic images; collect images from multiple angles to obtain detailed information from different parts of the crystal;
[0056] S112, Image Processing
[0057] Image preprocessing: including denoising, contrast enhancement, edge detection, etc., to facilitate subsequent 3D modeling and analysis; Feature extraction: using image processing software to extract the geometric features of crystal defects;
[0058] S113. 3D model reconstruction
[0059] Select modeling software: Choose appropriate software for 3D modeling, such as Avizo, AMIRA, MATLAB or related libraries in Python; perform 3D reconstruction using images collected from different angles; optimize and adjust the reconstructed 3D model to ensure its accuracy;
[0060] S114. Image display module operation steps
[0061] Load model: load the reconstructed 3D model into the image display module; Adjust viewing angle: rotate and scale the model to observe the 3D structure of the crystal from different angles; Analysis: perform detailed analysis of crystal defects in the 3D model, such as measuring defect size, shape and distribution;
[0062] S12, extracting the environmental data of the crystal growth process for defect analysis stored in the data acquisition terminal, the environmental data of the growth process including temperature data, humidity data and pressure data;
[0063] S2, importing the crystal image data to be analyzed into the feature extraction model to extract defect features;
[0064] In a specific implementation of this embodiment, the crystal image data to be analyzed is imported into the feature extraction model to extract defect features, which includes the following specific steps:
[0065] S21, obtaining the pixel value of each pixel point of the microscopic image of each part of the crystal, and at the same time obtaining the average pixel value of the microscopic image of the corresponding part of the normal crystal, obtaining the absolute value of the pixel difference between the pixel value of each pixel point of the microscopic image of each part of the crystal and the average pixel value of the microscopic image of the corresponding part of the normal crystal, obtaining the absolute value of the pixel difference of the corresponding pixel point, obtaining the pixel point whose absolute value of the pixel difference is greater than or equal to the pixel difference threshold and setting it as the abnormal pixel point, obtaining the pixel point corresponding to the largest absolute value of the pixel difference and setting it as the central abnormal pixel point, obtaining the regional image composed of the abnormal pixels, setting it as the defective regional image, and obtaining the contour features and average pixel value features of the defective regional image;
[0066] S22, obtain the corresponding defect area image, take the central abnormal pixel point as the center, divide several pixel abnormal areas according to the set pixel gradient, obtain the contour features of the obtained pixel abnormal area, and obtain the average value of the distance between the edge of the pixel abnormal area and the central abnormal pixel point. It should be noted in this step that the pixel gradient here is set according to the experience of those skilled in the art. For example, with the central abnormal pixel point as the center, two pixel values are used as the pixel gradient to divide several pixel abnormal areas. The specific method is: obtain the pixel value of the central abnormal pixel point, obtain the area within two pixel values different from the pixel value of the central abnormal pixel point as one pixel abnormal area, and the area within two pixel values to four pixel values different from the pixel value of the central abnormal pixel point is set as another pixel abnormal area, and so on to obtain all pixel abnormal areas;
[0067] Through this step, the crystal image is analyzed to obtain the multi-dimensional image implicit features reflecting the defect types;
[0068] S3, importing the extracted defect features into the constructed defect feature preliminary identification model to perform preliminary identification of the defect features;
[0069] In a specific implementation of this embodiment, the extracted defect features are imported into the constructed defect feature preliminary identification model to perform preliminary identification of the defect features, which includes the following specific steps:
[0070] S31, obtaining contour features and average pixel value features of the defect area image, and simultaneously obtaining contour features and average pixel value features of defects that have been judged in the past;
[0071] S32, importing the obtained contour features of the defect area image and the contour features of the determined defect into a contour similarity value calculation formula to calculate the contour similarity value, wherein the contour similarity value calculation formula between the j-th determined defect and the defect area image is: , where m() is the area of the contour in the brackets, c is the contour feature of the defect area image, and cj is the contour feature of the jth identified defect image. is the intersection of the contours, is the union of contours;
[0072] S33, substituting the obtained contour similarity value, the average pixel value feature of the defect area image and the average pixel value feature of the determined defect into the overall similarity value calculation formula to calculate the overall similarity value, wherein the overall similarity value calculation formula between the j-th determined defect and the defect area image is: , where a is the contour ratio, z is the average pixel value of the defect area image, and zj is the average pixel value of the jth identified defect;
[0073] S34, obtaining the contour features of each abnormal pixel region of the defect area and the average value of the distance between the edge and the central abnormal pixel point, and at the same time obtaining the contour features of each abnormal pixel region of the determined defect, and importing them into the calculation formula of the contour similarity value of the abnormal pixel region to calculate the contour similarity value of the abnormal pixel region, wherein the calculation formula of the contour similarity value of the abnormal pixel region of the j-th determined defect and the defect area image is: , where n is the number of pixel abnormal regions in the defect area, ki is the contour feature of the i-th pixel abnormal region in the defect area, Lij is the contour feature of the i-th pixel abnormal region corresponding to the j-th determined defect, and Di is the average distance between the edge of the i-th pixel abnormal region in the defect area and the central abnormal pixel point;
[0074] S35, obtaining the calculated overall similarity value between the determined defect and the defective region image, and the similarity value between the determined defect and the pixel abnormal region contour of the defective region image, and substituting them into the preliminary recognition value calculation formula to calculate the preliminary recognition value of the determined defect and the defective region image, wherein the preliminary recognition value calculation formula is: , where Nj is the initial recognition value of the jth identified defect and defect area image, is the contour similarity ratio coefficient;
[0075] S36, arranging the calculated preliminary identification values in descending order, obtaining the determined defect types corresponding to the preliminary identification values arranged in the first five items and setting them as the preliminary identification defect types;
[0076] S4, importing the environmental data during the crystal growth process into the environmental impact analysis model to perform environmental impact analysis;
[0077] In a specific implementation of this embodiment, the environmental data during crystal growth is imported into the environmental impact analysis model for environmental impact analysis, including the following specific contents: obtaining environmental data during the growth of the crystal corresponding to the defect type initially identified and environmental data during the growth of the crystal for defect analysis, and importing them into the environmental impact analysis value calculation formula to calculate the environmental impact analysis value, wherein the environmental impact analysis value calculation formula is: , T is the crystal growth time, V is the type of environmental data, wv is the proportion coefficient of the vth environmental data, dt is the time integral, pvt is the vth environmental data at the growth time t corresponding to the initial identification of the defect type, and pvtm is the vth environmental data at the growth time t during the growth process of the crystal for defect analysis;
[0078] S5. Import the environmental impact analysis results and the preliminary identification results of defect characteristics into the defect characteristic secondary identification model to identify the defect type;
[0079] In a specific implementation of this embodiment, the environmental impact analysis results and the preliminary identification results of defect features are introduced into the defect feature secondary identification model for secondary identification of defect features, including the following specific contents:
[0080] Substitute the environmental impact analysis value and preliminary identification value of the defect type obtained in the preliminary identification into the final identification value calculation formula to calculate the final identification value, where the final identification value calculation formula is: , where exp() is the power of the natural constant e, and the preliminary identified defect type corresponding to the obtained maximum final identification value is set as the defect type of the crystal to be analyzed.
[0081] S6. Output the obtained defect type identification result to the user end.
[0082] It should be noted that the contour proportion coefficient, contour similarity proportion coefficient and proportion coefficient of the vth environmental data are obtained as follows: 5000 representative groups of crystal image data and environmental data during crystal growth are obtained, experts are hired to manually classify crystal defects, and the obtained crystal image data and environmental data during crystal growth are substituted into the final recognition value calculation formula to calculate the final recognition value, and the final recognition values of various types of recognition defects are obtained. The manual classification results of crystal defects and the final recognition values of various types of recognition defects are obtained and substituted into the fitting software, and the values of the contour proportion coefficient, contour similarity proportion coefficient and proportion coefficient of the vth environmental data that meet the highest judgment accuracy are output;
[0083] The advantages of this embodiment over the prior art are as follows: obtaining crystal image data that needs to be analyzed and environmental data during the crystal growth process, importing the crystal image data that needs to be analyzed into a feature extraction model to extract defect features, importing the extracted defect features into a constructed defect feature preliminary identification model to perform preliminary identification of the defect features, importing the environmental data during the crystal growth process into an environmental impact analysis model to perform environmental impact analysis, importing the environmental impact analysis results and the preliminary identification results of the defect features into a defect feature secondary identification model to perform defect type identification, and outputting the obtained defect type identification results to the user end. This application obtains multi-dimensional image implicit features that reflect the defect types by performing image analysis on crystal images, and at the same time comprehensively analyzes the environmental data of the crystal growth process to accurately identify the defect features, thereby achieving the technical effect of improving the accuracy and efficiency of defect identification.
[0084] Example 2
[0085] like Figure 3As shown, a crystal defect recognition system based on feature analysis is implemented based on the above-mentioned crystal defect recognition method based on feature analysis, which specifically includes a data acquisition module, a defect feature extraction module, a defect feature preliminary recognition module, an environmental impact analysis module, a defect type recognition module and a defect type output module, wherein the data acquisition module is used to obtain crystal image data that needs to be analyzed and environmental data during the crystal growth process;
[0086] A defect feature extraction module is used to import the crystal image data to be analyzed into the feature extraction model to extract the defect features;
[0087] A defect feature preliminary identification module is used to import the extracted defect features into the constructed defect feature preliminary identification model for preliminary identification of the defect features;
[0088] Environmental impact analysis module, used to import environmental data during crystal growth into the environmental impact analysis model for environmental impact analysis;
[0089] The defect type identification module is used to import the environmental impact analysis results and the preliminary identification results of defect characteristics into the defect characteristic secondary identification model to identify the defect type;
[0090] The defect type output module is used to output the defect type identification results to the user end; at the same time, Figure 3 The arrows in the figure represent the signal transmission connection relationship between the modules.
[0091] Example 3
[0092] This embodiment provides an electronic device, including: a processor and a memory, wherein the memory stores a computer program that can be called by the processor;
[0093] The processor executes the above-mentioned crystal defect identification method based on feature analysis by calling the computer program stored in the memory.
[0094] The electronic device may have relatively large differences due to different configurations or performances, and may include one or more processors and one or more memories, wherein the memory stores at least one computer program, and the computer program is loaded and executed by the processor to implement the crystal defect identification method based on feature analysis provided by the above method embodiment. The electronic device may also include other components for implementing the functions of the device, for example, the electronic device may also have components such as a wired or wireless network interface and an input and output interface to input and output data. This embodiment will not be described in detail here.
[0095] Example 4
[0096] This embodiment provides a computer-readable storage medium having a rewritable computer program stored thereon;
[0097] When the computer program runs on a computer device, the computer device is enabled to execute the above-mentioned crystal defect identification method based on feature analysis.
[0098] For example, the computer readable storage medium can be a read-only memory, a random access memory, a read-only CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.
[0099] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When a computer instruction or computer program is loaded or executed on a computer, a process or function according to an embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. Computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center through a wired network or / and a wireless network. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.
Claims
1. A crystal defect identification method based on feature analysis, characterized in that: It includes the following specific steps: Acquire crystal image data to be analyzed and environmental data during crystal growth; Import the crystal image data to be analyzed into the feature extraction model to extract defect features; Importing the extracted defect features into the constructed defect feature preliminary identification model to perform preliminary identification of the defect features; Importing environmental data from the crystal growth process into the environmental impact analysis model for environmental impact analysis; Import the environmental impact analysis results and the preliminary identification results of defect features into the defect feature secondary identification model to identify the defect types; The obtained defect type identification result is output to the user end; the step of importing the crystal image data to be analyzed into the feature extraction model to extract the defect features includes the following specific steps: Obtain the pixel value of each pixel point of the microscopic image of each part of the crystal, and at the same time obtain the average pixel value of the microscopic image of the corresponding part of the normal crystal, obtain the absolute value of the pixel difference between the pixel value of each pixel point of the microscopic image of each part of the crystal and the average pixel value of the microscopic image of the corresponding part of the normal crystal, obtain the absolute value of the pixel difference of the corresponding pixel point, obtain the pixel point whose absolute value of the pixel difference is greater than or equal to the pixel difference threshold and set it as an abnormal pixel point, obtain the pixel point corresponding to the largest absolute value of the pixel difference and set it as the central abnormal pixel point, obtain the regional image composed of the abnormal pixels, set it as the defective regional image, and obtain the contour characteristics and average pixel value characteristics of the defective regional image; Obtain the corresponding defect area image, divide a number of pixel abnormal areas according to the set pixel gradient with the central abnormal pixel point as the center, obtain the contour features of the obtained pixel abnormal area, and obtain the average value of the distance between the edge of the pixel abnormal area and the central abnormal pixel point; the extracting defect features are imported into the constructed defect feature preliminary recognition model for preliminary recognition of the defect features, which includes the following specific steps: Obtain the contour features and average pixel value features of the defect area image, and at the same time obtain the contour features and average pixel value features of the defects that have been judged in the past; The contour features of the acquired defect area image and the contour features of the determined defect are introduced into the contour similarity value calculation formula to calculate the contour similarity value, wherein the contour similarity value calculation formula of the jth determined defect and the defect area image is: , where m() is the area of the contour in the brackets, c is the contour feature of the defect area image, and cj is the contour feature of the jth identified defect image. is the intersection of the contours, is the union of contours; Substitute the obtained contour similarity value, the average pixel value feature of the defect area image and the average pixel value feature of the determined defect into the overall similarity value calculation formula to calculate the overall similarity value, where the overall similarity value calculation formula between the jth determined defect and the defect area image is: , where a is the contour ratio, z is the average pixel value of the defect area image, and zj is the average pixel value of the jth determined defect; the step of importing the extracted defect features into the constructed defect feature preliminary recognition model for preliminary recognition of defect features also includes the following specific contents: The contour features of each abnormal pixel area in the defect area and the average value of the distance between the edge and the central abnormal pixel point are obtained. At the same time, the contour features of each abnormal pixel area of the judged defect are obtained, and the contour similarity value of the abnormal pixel area is calculated by importing the contour similarity value calculation formula of the abnormal pixel area. Among them, the contour similarity value calculation formula of the abnormal pixel area of the j-th judged defect and the defect area image is as follows: , where n is the number of pixel abnormal regions in the defect area, ki is the contour feature of the i-th pixel abnormal region in the defect area, Lij is the contour feature of the i-th pixel abnormal region corresponding to the j-th determined defect, and Di is the average distance between the edge of the i-th pixel abnormal region in the defect area and the central abnormal pixel point; Obtain the calculated overall similarity value between the determined defect and the defective region image, and the similarity value between the determined defect and the pixel abnormal region contour of the defective region image, and substitute them into the preliminary recognition value calculation formula to calculate the preliminary recognition value of the determined defect and the defective region image, wherein the preliminary recognition value calculation formula is: , where Nj is the initial recognition value of the jth identified defect and defect area image, is the contour similarity ratio coefficient; The calculated preliminary identification values are arranged in descending order, and the determined defect types corresponding to the preliminary identification values arranged in the first five items are obtained and set as the preliminary identification defect types; the environmental data during the crystal growth process is imported into the environmental impact analysis model for environmental impact analysis, which includes the following specific contents: the environmental data during the growth process of the crystal corresponding to the preliminary identification defect type and the environmental data during the growth process of the crystal for defect analysis are obtained, and the environmental impact analysis value calculation formula is imported to calculate the environmental impact analysis value, wherein the environmental impact analysis value calculation formula is: , T is the crystal growth time, V is the type of environmental data, wv is the proportion coefficient of the vth environmental data, dt is the time integral, pvt is the vth environmental data at the growth time t corresponding to the initial identification defect type, and pvtm is the vth environmental data at the growth time t during the growth process of the crystal for defect analysis; the introduction of the environmental impact analysis results and the initial identification results of the defect characteristics into the defect characteristic secondary identification model for secondary identification of defect characteristics includes the following specific contents: Substitute the environmental impact analysis value and preliminary identification value of the defect type obtained in the preliminary identification into the final identification value calculation formula to calculate the final identification value, where the final identification value calculation formula is: , where exp() is the power of the natural constant e, and the preliminary identified defect type corresponding to the obtained maximum final identification value is set as the defect type of the crystal to be analyzed.
2. The crystal defect identification method based on feature analysis according to claim 1, characterized in that: The specific steps of obtaining the crystal image data to be analyzed and the environmental data during the crystal growth process are: Place the crystal that needs to be defect analyzed in the electron microscope imaging module to obtain microscopic images of various parts of the crystal, and display the three-dimensional model of the crystal in the image display module; The environmental data during the growth process of the crystal for defect analysis is extracted and stored through the data acquisition terminal, and the environmental data during the growth process includes temperature data, humidity data and pressure data.
3. A crystal defect recognition system based on feature analysis, which is implemented based on the crystal defect recognition method based on feature analysis as claimed in any one of claims 1 to 2, characterized in that: It specifically includes a data acquisition module, a defect feature extraction module, a defect feature preliminary identification module, an environmental impact analysis module, a defect type identification module and a defect type output module, wherein the data acquisition module is used to obtain crystal image data that needs to be analyzed and environmental data during the crystal growth process; The defect feature extraction module is used to import the crystal image data to be analyzed into the feature extraction model to extract the defect features; The defect feature preliminary identification module is used to import the extracted defect features into the constructed defect feature preliminary identification model to perform preliminary identification of the defect features; The environmental impact analysis module is used to import the environmental data during the crystal growth process into the environmental impact analysis model to perform environmental impact analysis; The defect type identification module is used to import the environmental impact analysis results and the preliminary identification results of the defect characteristics into the defect characteristic secondary identification model to identify the defect type; The defect type output module is used to output the obtained defect type identification result to the user end.
4. An electronic device comprising: A processor and a memory, wherein the memory stores a computer program that can be called by the processor; It is characterized in that the processor executes the crystal defect identification method based on feature analysis as described in any one of claims 1-2 by calling the computer program stored in the memory.
5. A computer-readable storage medium, characterized in that: Instructions are stored, and when the instructions are executed on a computer, the computer is caused to execute the crystal defect identification method based on feature analysis as described in any one of claims 1 to 2.
