Wafer labeling method and device
By automatically labeling wafer images with tags and annotation boxes in a visual interface, combined with edge detection, the problem of cumbersome and error-prone wafer annotation in existing technologies is solved, achieving efficient and accurate wafer annotation.
Patent Information
- Application Number
- CN202510884000.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-27
- Publication Date
- 2025-11-07
AI Technical Summary
Existing wafer marking methods are cumbersome, inefficient, and prone to errors, affecting detection accuracy.
The visual interface displays a set of labels associated with the wafer image. Clicking on a chip automatically adds labels and bounding boxes, and edge detection is used to generate accurate bounding boxes.
It improves the accuracy and efficiency of wafer labeling, reduces manual operations, and enhances the detection results of automated optical inspection models.
Smart Images
Figure CN120909480A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of wafer quality inspection, and provides a wafer labeling method and device. BACKGROUND
[0002] In wafer production, defect detection is crucial. With the rapid development of deep learning technology, wafer defect detection cannot be separated from the labeling of dies on the wafer.
[0003] Currently, when labeling a wafer, a labeler needs to identify the type of a die, mark the edge of the die by dragging a mouse, and then label the type of the die (such as a normal die, a champion die, an abnormal die of a certain defect, etc.). This labeling method is relatively cumbersome and inefficient, and there is a certain error in dragging a mouse, thereby affecting the accuracy of subsequent wafer detection. SUMMARY
[0004] Embodiments of the present application provide a wafer labeling method and device for improving the efficiency and accuracy of wafer labeling.
[0005] In a first aspect, embodiments of the present application provide a wafer labeling method, comprising:
[0006] In response to a labeling operation on an imported wafer image dataset, displaying a plurality of wafer images contained in the wafer image dataset and a label set used by the wafer image dataset in a labeling interface, the label set being determined according to a wafer model corresponding to the wafer image dataset;
[0007] In response to an image selection operation triggered on the wafer image, adding a selected target wafer image to a labeling area of the labeling page, the target wafer image being any wafer image in the wafer image dataset;
[0008] In response to a label selection operation triggered on the label set, presenting a selected target label as a labeling state in the labeling interface, the target label being any label in the label set;
[0009] In response to a click operation triggered on a die belonging to the target label on the target wafer image, displaying a labeling box surrounding the die, and displaying the target label in a set region of the die.
[0010] The beneficial effects of the above technical solutions are: through the visual marking page, a target wafer image needing to be marked is selected from the imported wafer image data set, because the core particle structures of different wafer models are different, the defects existing are also different, therefore, while the target wafer image is presented, a label set associated with the target wafer image is also presented, and then the core particles belonging to the selected target label in the target wafer image are sequentially marked in units of labels, in the marking process, only the core particle needs to be clicked, and the marking box and the target label of the core particle can be automatically marked, without manual frame selection, so that the precision and efficiency of wafer marking are effectively improved.
[0011] Optionally, in response to the click operation triggered for the core particle belonging to the target label on the target wafer image, a marking box surrounding the core particle is displayed, and the method comprises the following steps.
[0012] In response to the click operation triggered for the core particle belonging to the target label on the target wafer image, the click coordinates of the click position are acquired, and the marking image surrounding the core particle in the target wafer image is determined in combination with the marking box size preset for the target wafer image.
[0013] Edge detection is performed on the marking image to obtain a plurality of detection boxes.
[0014] The marking box surrounding the core particle is generated and displayed according to the plurality of detection boxes and the marking box size.
[0015] The beneficial effects of the above technical solutions are: in the marking process, after the core particle belonging to the target label on the target wafer image is clicked, the marking image surrounding the core particle is determined through the click coordinates and the marking box size, and the plurality of detection boxes obtained through edge detection in the marking image are compared with the marking box size respectively, so that the marking box surrounding the core particle is quickly generated, manual frame selection is not needed, and the marking difficulty and workload are reduced.
[0016] Optionally, the generating and displaying of the marking box surrounding the core particle according to the plurality of detection boxes and the marking box size comprises the following steps.
[0017] The plurality of detection boxes are traversed, and each time the traversal is performed: according to the center coordinates and the size of the detection box currently traversed, in combination with the click coordinates, when the detection box currently traversed contains the click position, the detection box currently traversed is taken as a candidate detection box.
[0018] According to the size of each candidate detection box, the area of the corresponding detection box is determined, and according to the marking box size, the area of the marking box is determined.
[0019] determine a target detection frame from the at least one candidate detection frame according to area differences between areas of the at least one detection frame and the area of the labeled frame, the target detection frame representing the outline of the core pellet;
[0020] generate and display a labeled frame surrounding the core pellet according to the target detection frame.
[0021] The technical scheme has the beneficial effects that: first, at least one candidate detection frame containing the click position is filtered from the multiple detection frames, so that the edge outline irrelevant to the core pellet is filtered out, and the labeling efficiency is improved; further, the target detection frame representing the click core pellet outline is obtained according to the area differences between areas of the at least one detection frame and the area of the labeled frame, and then the labeled frame surrounding the core pellet is generated, so that the accuracy of the labeled frame is improved through coarse detection and fine detection of the detection frame.
[0022] Optionally, the generating and displaying of the labeled frame surrounding the core pellet according to the target detection frame comprises:
[0023] determining the labeled frame surrounding the core pellet and displaying the labeled frame by taking the center coordinates of the target detection frame as the center point of the labeled frame and combining the size of the labeled frame.
[0024] The technical scheme has the beneficial effects that: since the target detection frame is the outline of the core pellet and the labeled frame surrounds the edge frame of the core pellet, the center of the target detection frame can be taken as the center of the labeled frame when the size of the labeled frame is determined, so that the position of the labeled frame is accurately determined.
[0025] Optionally, the obtaining of the click coordinates of the click position and the determination of the labeled image surrounding the core pellet in the target wafer image according to the pre-set size of the labeled frame for the target wafer image comprises:
[0026] determining the origin coordinates of the labeled image according to the click coordinates and the size of the labeled frame;
[0027] determining the region image surrounding the core pellet in the target wafer image according to the origin coordinates and a preset multiple of the size of the labeled frame, the preset multiple being greater than 1;
[0028] performing gray-scale processing on the region image to generate a labeled image.
[0029] The technical scheme has the beneficial effects that: the labeled image containing the smallest region of the complete core pellet is generated through the click coordinates and the size of the labeled frame, so that the completeness of the clicked core pellet in the labeled image is ensured, the workload of subsequent edge detection is effectively reduced, and the real-time performance of automatic labeling is ensured.
[0030] Optionally, the method further comprises:
[0031] in response to a selection operation of a mark style in the toolbar of the mark interface, the selected target mark style is changed to a selected state;
[0032] the mark box surrounding the core particle is displayed in response to the click operation triggered by the core particle belonging to the target label on the target wafer image, and the mark box surrounding the core particle is displayed in the target mark style.
[0033] the mark box surrounding the core particle is displayed in response to the click operation triggered by the core particle belonging to the target label on the target wafer image, and the mark box surrounding the core particle is displayed in the target mark style.
[0034] The beneficial effects of the above technical solutions are: by selecting a mark style, the mark box is adapted to the wafer model, thereby ensuring the accuracy of the mark.
[0035] Optionally, in response to the image selection operation triggered by the wafer image, the method further comprises:
[0036] in response to the setting operation triggered by the wafer model, a mark setting interface is presented, and the mark setting interface is used to set a label set associated with each wafer model and a mark box size.
[0037] The beneficial effects of the above technical solutions are: since the core particle structures on different wafer models are different, and the core particle defects are different, the label set corresponding to each wafer model and the mark box size can be pre-set through a visual interface.
[0038] Optionally, the method further comprises:
[0039] in response to the import operation triggered by the wafer image data set, a data import interface is presented, and the data import interface is used to set a wafer model corresponding to the wafer image data set.
[0040] The beneficial effects of the above technical solutions are: when importing the wafer image data set, the wafer model is set, so that the label set associated with the wafer image and the mark box size can be obtained in the subsequent mark process, and automatic mark of the core particle is realized.
[0041] In a second aspect, the embodiments of the present application provide an electronic device, comprising a processor, a memory, a display screen and a communication interface, the communication interface, the display screen, the memory and the processor are connected through a bus;
[0042] the communication interface is used to transceive data;
[0043] the display screen is used to display a mark interface;
[0044] the memory stores a computer program, and the processor executes the following operations according to the computer program:
[0045] in response to a labeling operation on the imported wafer image dataset, displaying, in a labeling interface, a plurality of wafer images included in the wafer image dataset and a label set used by the wafer image dataset, the label set being determined according to a wafer model corresponding to the wafer image dataset;
[0046] in response to an image selection operation triggered on the wafer image, adding a selected target wafer image to a labeling area of the labeling page, the target wafer image being any wafer image in the wafer image dataset;
[0047] in response to a label selection operation triggered on the label set, presenting, in the labeling interface, a selected target label as a labeling state, the target label being any label in the label set;
[0048] in response to a click operation triggered on a die belonging to the target label on the target wafer image, displaying a labeling box surrounding the die, and displaying the target label in a set area of the die.
[0049] Optionally, the processor, in response to the click operation triggered on the die belonging to the target label on the target wafer image, displays the labeling box surrounding the die, and the specific operation is:
[0050] in response to the click operation triggered on the die belonging to the target label on the target wafer image, obtaining a click coordinate of the click position, and combining a labeling box size pre-set for the target wafer image to determine a labeling image surrounding the die in the target wafer image;
[0051] performing edge detection on the labeling image to obtain a plurality of detection boxes;
[0052] generating and displaying the labeling box surrounding the die according to the plurality of detection boxes and the labeling box size.
[0053] a program, and the program performs the following operations:
[0054] Optionally, the processor generates and displays the labeling box surrounding the die according to the plurality of detection boxes and the labeling box size, and the specific operation is:
[0055] iterating through the plurality of detection boxes, and each iteration performs: according to a center coordinate and a size of a currently iterated detection box, combining the click coordinate to determine that the currently iterated detection box contains the click position, and taking the currently iterated detection box as a candidate detection box;
[0056] determine a corresponding detection box area according to the size of each of the at least one candidate detection box, and determine a label box area according to the size of the label box;
[0057] determine a target detection box from the at least one candidate detection box according to area differences between the at least one detection box area and the label box area, the target detection box representing the outline of the core particle;
[0058] generate and display a label box surrounding the core particle according to the target detection box.
[0059] Optionally, the processor generates and displays a label box surrounding the core particle according to the target detection box, and the specific operation is as follows:
[0060] determine a label box surrounding the core particle by taking the center coordinates of the target detection box as the center point of the label box and combining the size of the label box, and display the label box.
[0061] Optionally, in response to the label selection operation triggered for the label set, the processor further performs:
[0062] in response to the selection operation of the label style in the toolbar of the label interface, change the selected target label style to a selected state;
[0063] in response to the click operation triggered by the pair of core particles belonging to the target label on the target wafer image, display a label box surrounding the core particles, and the specific operation is as follows:
[0064] in response to the click operation triggered by the pair of core particles belonging to the target label on the target wafer image, display a label box surrounding the core particles in the target label style.
[0065] Optionally, in response to the image selection operation triggered for the wafer image, the processor further performs:
[0066] in response to the setting operation triggered for the wafer model, present a label setting interface, and the label setting interface is used to set the label set and the label box size associated with each wafer model.
[0067] Optionally, the processor further performs:
[0068] in response to the import operation triggered for the wafer image dataset, present a data import interface, and the data import interface is used to set the wafer model corresponding to the wafer image dataset.
[0069] Optionally, the processor obtains the click coordinates of the click position, and determines the label image surrounding the core particle in the target wafer image by combining the label box size set in advance for the target wafer image, and the specific operation is as follows:
[0070] determining an origin point coordinate of the labeled image according to the click coordinate and the size of the label frame;
[0071] determining a region image surrounding the core particle in the target wafer image according to the origin point coordinate and a preset multiple of the size of the label frame, the preset multiple being greater than 1;
[0072] performing a grayscale processing on the region image to generate a labeled image.
[0073] In a third aspect, an embodiment of the present application provides a computer readable storage medium, which stores computer executable instructions, and the computer executable instructions, when executed, can implement the steps of any of the wafer labeling methods.
[0074] The technical effects brought by any of the implementation manners of the third aspect to the third aspect can refer to the technical effects brought by the corresponding implementation manners of the first aspect, which will not be described here. BRIEF DESCRIPTION OF DRAWINGS
[0075] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.
[0076] Figure 1 The labeling setting interface schematic diagram provided by the embodiment of the present application;
[0077] Figure 2A And Figure 2B The data import interface schematic diagram provided by the embodiment of the present application;
[0078] Figure 3 The flowchart schematic diagram of the wafer labeling method provided by the embodiment of the present application;
[0079] Figure 4A The wafer image dataset labeling interface schematic diagram provided by the embodiment of the present application;
[0080] Figure 4B The wafer image selection interface schematic diagram provided by the embodiment of the present application;
[0081] Figure 4C The label selection interface schematic diagram provided by the embodiment of the present application;
[0082] Figure 4D The core particle labeling interface schematic diagram provided by the embodiment of the present application;
[0083] Figure 5 A label frame display flowchart provided for an embodiment of the present application;
[0084] Figure 6 A labeled image diagram provided for an embodiment of the present application;
[0085] Figure 7 An edge detection result diagram of a labeled image provided for an embodiment of the present application;
[0086] Figure 8 A core particle labeling result diagram provided for an embodiment of the present application;
[0087] Figure 9 A wafer labeling system architecture diagram provided for an embodiment of the present application;
[0088] Figure 10 A structure diagram of an electronic device provided for an embodiment of the present application. DETAILED DESCRIPTION
[0089] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions of the present application will be described below in connection with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments described in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.
[0090] Based on the exemplary embodiments shown in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application. In addition, although the disclosure in the present application is introduced according to one or more examples, it should be understood that each aspect of the disclosure can also constitute a complete technical solution independently.
[0091] It should be noted that the brief description of the terms in the present application is only for the convenience of understanding the subsequently described embodiments, and is not intended to limit the embodiments of the present application. Unless otherwise specified, these terms should be understood according to their ordinary and general meanings.
[0092] The terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar or similar objects or entities, and do not necessarily mean a specific order or sequence, unless otherwise indicated. It should be understood that the terms used in this way can be interchanged under appropriate circumstances, for example, those other than the order given in the embodiment illustration or description of the present application can be implemented.
[0093] Furthermore, the terms "comprise", "comprising", "have", "having", "include", "including" and "contains", "containing", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a product or article that comprises a list of components does not include only those components but can include other components not expressly listed or inherent to such product or article.
[0094] The term "module" used in the present application refers to any known or later developed hardware, software, firmware, artificial intelligence, fuzzy logic, or combination of hardware and / or software code capable of performing the functions associated with that element.
[0095] In wafer quality inspection, the automatic optical detection model based on deep learning needs a large number of labeled samples for supervised training. Due to the non-uniform wafer models, different chip particle structures, and different chip particle defects, in addition to labeling the chip particles with labels during labeling, the chip particle edges need to be marked out by dragging the mouse to circle the chip particles. This labeling method is not only complex and time-consuming, but also has labeling errors, which affects the detection results of the automatic optical detection model.
[0096] In view of this, the wafer labeling method provided in the embodiments of the present application realizes automatic labeling of wafers through a visual page. In the labeling process, due to the different chip particle structures of different wafer models, the existing defects are also different. Therefore, when a selected target wafer image to be labeled is displayed in the labeling interface in the imported wafer image data set, a label set associated with the target wafer image is also displayed. The label set contains all chip particle types of the wafer model corresponding to the target wafer image. In this way, a target label can be selected in units of labels, and after clicking the chip particle belonging to the target label on the target wafer image, the chip particle contour is automatically extracted through edge detection to realize automatic labeling. The labeling result not only contains the label to which each chip particle belongs, but also contains the labeling frame surrounding each chip particle, without the need for manual framing, thereby effectively improving the accuracy and efficiency of wafer labeling.
[0097] Since the structures and defects of chip particles under different wafer models are different, the label set corresponding to each wafer model and the labeling frame size can be pre-set through a visual interface before wafer labeling, so as to realize automatic labeling of wafers.
[0098] In one embodiment, the configuration process includes: in response to a setting operation triggered for a wafer model, presenting a labeling setting interface, wherein the labeling setting interface is used to set the label set associated with each wafer model and the labeling frame size.
[0099] As Figure 1As shown, the annotation setting interface provided by the embodiment of the present application mainly includes two parts of basic information and annotation setting. Since the size of the chiplet of each wafer model is fixed, the user can input the information of the chiplet under the current wafer model in the basic information, such as the size of the chiplet and the distance of the chiplet, so as to identify the chiplet at the clicked position in subsequent annotation. After setting the chiplet information, the user can also set the information of the annotation box surrounding the chiplet in the basic information, such as the size of the annotation box and the distance of the annotation box. Usually, the size of the annotation box is greater than the size of the chiplet, so as to automatically annotate the annotation box on the clicked chiplet in subsequent annotation, thereby training the automatic optical detection model to automatically identify the chiplet in the wafer image. Since the types covered by the chiplets of different wafer models are different, in the annotation setting, the user can input the label set of the chiplet under the current wafer model, including but not limited to background (Background), normal chiplet (Chiplet), champion chiplet (FMark, ZMark and HMark, etc.), abnormal chiplet (Abnormal Chiplet) such as dirt, scratch, twin, and other chiplets (Other), etc. The annotation boxes of different labels can be distinguished by different colors. Except for the background label, the sizes of the annotation boxes of the remaining labels are consistent.
[0100] Optionally, in the basic information, the user can also set the size of the cutting box. Usually, the size of the cutting box is greater than the size of the annotation box, so that after the automatic optical detection model automatically identifies the chiplet in the wafer image, the chiplet image is cropped according to the size of the cutting box.
[0101] Optionally, in the basic information, the user can also set the name, version and associated model of the current wafer model and other information.
[0102] The data source of wafer annotation can be obtained from the wafer production line. The wafer images of each wafer model are collected by a camera, and the wafer image data set of each wafer under the model is imported into the automatic annotation tool implementing the method of the embodiment of the present application for wafer annotation.
[0103] In an embodiment, considering that the number of chiplets on the wafer is large and the size of the chiplet is small, when collecting the wafer image of a certain model, the local wafer image can be collected from multiple angles. These local wafer images can finally be spliced into a complete wafer image.
[0104] In an embodiment, the import process of the wafer image data set includes: in response to an import operation triggered for the wafer image data set, a data import interface is presented, wherein the data import interface is used to set the wafer model corresponding to the wafer image data set.
[0105] As Figure 2A and Figure 2BAs shown, the data import interface includes a data creation sub-interface and a data import sub-interface, and in the data creation sub-interface, Figure 2A In the data creation sub-interface, the user inputs the name, data type, wafer model, and other information of the imported wafer image data set, wherein the wafer models corresponding to the wafer images in the same wafer image data set are the same, so that when the wafer image data set is subsequently annotated, the wafer model corresponding to the wafer image data set can be obtained, and then the size and distance of the core particles in the wafer image, the size and distance of the annotation box, and the label set of the core particles can be obtained according to the wafer model. After the wafer image data set is created, the data import sub-interface shown in Figure 2B As shown, the data import sub-interface, the user can set the path and import mode of the wafer image data set, and after the import is completed, the wafer image data set can be annotated.
[0106] After the above preprocessing operation, the wafer is annotated, as shown in Figure 3 The flowchart of the wafer annotation method provided by the embodiment of the present application mainly includes the following steps:
[0107] S301: In response to the annotation operation on the imported wafer image data set, the plurality of wafer images contained in the wafer image data set and the label set used by the wafer image data set are displayed in the annotation interface.
[0108] When importing the wafer data set, the wafer model corresponding to each wafer image data set is specified, and the label set of the core particles of each wafer model is pre-set, so that the label set used for annotation of the wafer image data set can be determined according to the wafer model corresponding to the wafer image data set, and the plurality of wafer images contained in the wafer image data set are displayed in the annotation interface.
[0109] As shown in Figure 4A The user can select a wafer image data set that needs to be annotated from the imported wafer image data sets through the "filter" box in the annotation interface. After the wafer image data set to be annotated is determined, the image display area 401 of the annotation interface will present the plurality of wafer images contained in the wafer image data set, and at the same time, the label display area 402 of the annotation interface will present the label set used by the wafer image data set.
[0110] In an embodiment, since the information of the annotation box of the core particles of each wafer model is fixed, after the wafer image data set to be annotated is determined, the annotation box information corresponding to the wafer image data set to be annotated can also be displayed in the annotation interface.
[0111] As shown in Figure 4A After the wafer image data set to be annotated is determined, the functional area 403 of the annotation interface will present the size and distance of the annotation box of the core particles of the wafer model.
[0112] S302: In response to an image selection operation triggered for a wafer image, add the selected target wafer image to the annotation area of the annotation page.
[0113] The target wafer image is any wafer image from the labeled wafer image dataset.
[0114] like Figure 4B As shown, when a user selects the wafer image “IMAGE1_0094.jpg” as the target wafer image for current annotation in the image display area 401, the target wafer image will be displayed in the annotation area 404 of the annotation interface.
[0115] S303: In response to a label selection operation triggered for a set of labels, the selected target label is displayed as a label in the labeling interface.
[0116] The annotation interface contains a set of labels used in the annotated wafer image dataset. For a target label selected by the user, the selected target label is presented as an annotation in the annotation interface. The target label can be any label in the label set, so that the core under each label can be annotated in the target wafer image on a label-by-label basis.
[0117] like Figure 4C As shown, after selecting a target label, the border of the target label becomes highlighted in the label display area 402 of the annotation interface, and the target label is also displayed in the function area 403 of the annotation interface.
[0118] S304: In response to a click operation triggered on a die belonging to a target label on a target wafer image, display a bounding box surrounding the die and display the target label within a defined area of the die.
[0119] Since the chip structure differs between different wafer models, after determining the wafer image dataset to be annotated, the user can select an annotation style (e.g., the shape of the annotation box) in the toolbar of the annotation interface according to the wafer model. After selecting the annotation style, the selected target annotation style will be turned on.
[0120] For example, if you select "rectangle" to annotate a target wafer image within the designated area, the "rectangle" option in the toolbar will be selected, and the outline of the focal point will be displayed as a rectangle. Figure 4D As shown by the dashed line in the image.
[0121] The embodiment of the present application needs to open the "automatic labeling" option in the function area 403 of the labeling interface when automatically labeling the wafer image. In the embodiment of the present application, the timing of opening the "automatic labeling" option is not limited, and it only needs to be opened before the clicking operation, such as being opened before the labeling operation, being opened after the label selection, and the like. It can also be opened by default.
[0122] After the "automatic labeling" is opened, the user can click any chiplet region of the chiplet belonging to the target label in the target wafer image, so that the target labeling style can display the labeling box surrounding the chiplet, and the target label in the set region of the chiplet, thereby realizing the automatic labeling of the chiplet. In addition, by selecting the labeling style, the labeling box is adapted to the wafer model, thereby ensuring the labeling accuracy.
[0123] As shown in Figure 4D , after the user clicks the chiplet of the "Chiplet" label, a rectangular labeling box is displayed around the chiplet, and the "Chiplet" label is displayed in the upper left corner region of the chiplet.
[0124] In one embodiment, after labeling each chiplet, the user can save the labeling result of the chiplet by clicking the "save" option in the toolbar.
[0125] It should be noted that after the "automatic labeling" option is closed, the embodiment of the present application can support manual labeling.
[0126] In the embodiment of the present application, a target wafer image needing labeling is selected from the imported wafer image data set through a visual labeling page. Since the chiplet structures of different wafer models are different, the defects are also different. Therefore, when the target wafer image is presented, the label set associated with the target wafer image is also presented, and then the chiplets belonging to the selected target label in the target wafer image are labeled in sequence in units of labels. During the labeling process, only the next chiplet needs to be clicked, and the labeling box and the target label of the chiplet can be automatically labeled, without manual framing, thereby effectively improving the precision and efficiency of wafer labeling.
[0127] In one embodiment, the labeling box of the chiplet is determined by an edge detection method.
[0128] As shown in Figure 5 , it is a labeling box display flowchart, mainly including the following steps:
[0129] S3041: In response to the clicking operation triggered by the chiplet belonging to the target label on the target wafer image, the clicking coordinates of the clicking position are obtained, and the labeling box surrounding the chiplet in the target wafer image is determined in combination with the labeling box size set in advance for the target wafer image.
[0130] wherein the size of the annotation image is greater than the size of the annotation box to ensure the integrity of the clicked-on core pellet.
[0131] In an embodiment, the determining process of the annotation image comprises: determining the origin coordinates of the annotation image according to the click coordinates and the size of the annotation box; determining the region image in the target wafer image that encloses the core pellet according to the origin coordinates and a preset multiple of the size of the annotation box, the preset multiple being greater than 1; and performing grayscale processing on the region image to generate the annotation image.
[0132] Optionally, the annotation image containing the smallest region of the complete core pellet can be obtained through the canvas function. By intercepting the annotation image corresponding to the smallest region containing the complete core pellet, the computing power for automatic annotation is saved, the annotation efficiency is improved, and the real-time performance of annotation is ensured.
[0133] As shown in Figure 6 , assuming that the click coordinates of the click position are (px, py), the size of the annotation box is width w and height h, and the origin coordinates of the annotation image are (x1, y1), wherein x1 = px-w and y1 = py-h, when the preset multiple is 4, the width of the annotation image is 2w and the height is 2h, and in combination with the origin coordinates (x1, y1), the remaining three corner points (x2, y2), (x3, y3) and (x4, y4) can be obtained, thereby determining the annotation image containing the clicked-on core pellet.
[0134] It should be noted that Figure 6 only the generation process of the annotation image is described, and in actual annotation, the solid line frame formed by the four corner points of the annotation image is invisible.
[0135] S3042: performing edge detection on the annotation image to obtain a plurality of detection boxes.
[0136] In an embodiment, the edge detection process comprises: converting the annotation image into a grayscale image to increase the contrast between the core pellet and the background, and performing binaryzation processing on the grayscale image to obtain a binaryzation image highlighting the contour edges, and then using a Sobel operator to perform edge detection on the binaryzation image to obtain a plurality of detection boxes.
[0137] As shown in Figure 7 , it is the edge detection result of the annotation image, which contains not only the edge contour of the clicked-on core pellet but also other edge contours.
[0138] S3043: generating and displaying an annotation box enclosing the core pellet according to the plurality of detection boxes and the size of the annotation box.
[0139] For the multiple detection boxes detected by the edge, a target detection box closest to the area of the clicked core particle can be screened out, and then the size of the labeling box is combined to determine the position of the final labeling box. Specifically, the generation process of the labeling box is as follows:
[0140] S3043_1: Traverse multiple detection boxes, and perform the following each time: according to the center coordinates and size of the current traversed detection box, and in combination with the click coordinates, determine whether the current traversed detection box contains the click position, and if so, take the current traversed detection box as a candidate detection box.
[0141] The size of the detection box includes the width and height of the detection box. The coverage of the current traversed detection box can be determined by the center coordinates and size of the current traversed detection box. When the click coordinates are contained in the coverage, it can be determined that the detection box contains the click position, so that the detection box is taken as a candidate detection box.
[0142] S3043_2: According to the size of each of the at least one candidate detection box, determine the area of the corresponding detection box, and according to the size of the labeling box, determine the area of the labeling box.
[0143] Since the size of the labeling box of the same wafer model is fixed, the area of the labeling box is fixed.
[0144] S3043_3: According to the area difference between the area of at least one detection box and the area of the labeling box, determine a target detection box from the at least one candidate detection box.
[0145] The smaller the area difference between the area of the labeling box and the area of the labeling box, the closer the detection box and the labeling box, and the more the detection box can represent the outline of the clicked core particle. Therefore, the candidate detection box with the smallest area difference can be taken as the target detection box to represent the outline of the clicked core particle.
[0146] In the automatic labeling process, for the multiple detection boxes detected by the edge, at least one candidate detection box containing the click position is filtered out from the multiple detection boxes, so that the edge contour irrelevant to the labeled core particle is filtered out, the labeling efficiency is improved, and further, according to the area difference between the area of at least one detection box and the area of the labeling box, a target detection box representing the outline of the clicked core particle is obtained, and then a labeling box surrounding the core particle is generated, realizing coarse detection and fine detection of the detection box, and effectively improving the accuracy of the labeling box.
[0147] S3043_4: According to the target detection box, generate a labeling box surrounding the core particle and display.
[0148] After the target detection frame representing the contour of the clicked core particle is determined, since the target detection frame is the contour of the core particle and the labeling frame is the position of the core particle in the image, the center coordinates of the target detection frame are taken as the center point of the labeling frame, and the labeling frame size is combined to accurately determine the position of the labeling frame and display the labeling frame surrounding the core particle.
[0149] As shown in Figure 8 , it is a schematic diagram of the display effect of the final labeling frame.
[0150] In the labeling process, after the core particle belonging to the target label on the target wafer image is clicked, the labeling image surrounding the core particle is determined through the clicked coordinates and the labeling frame size, and the multiple detection frames obtained through edge detection in the labeling image are compared with the labeling frame size respectively, so that the labeling frame surrounding the core particle is quickly determined without manual frame selection, and the labeling difficulty and workload are reduced.
[0151] In an embodiment, in order to improve the wafer labeling efficiency, the BS architecture can be used to detect the edge of the core particle through multi-threading.
[0152] Referring to Figure 9 , the wafer labeling system architecture provided by the embodiment of the present application mainly includes three modules of wafer model setting, wafer image data set import and quick labeling. In the wafer model setting module, the information of core particles, labeling frames and cutting frames is set according to the wafer model type, and the label set associated with the wafer model is set. In the wafer image data set import module, the corresponding wafer model is associated with the imported wafer image data set. In the quick labeling module, by entering the labeling page, the labeling frame size and label type corresponding to the wafer model associated with the labeled wafer image data set are obtained, a target label is selected for the labeled target wafer image, the core particle needing to be labeled with the target label is detected, the labeling image containing the smallest area of the core particle is cut out by clicking the core particle area, the edge rectangle representing the contour of the clicked core particle is screened out, and the drawing and display of the labeling frame are performed based on the edge rectangle, and the target label to which the clicked core particle belongs in the labeling frame is displayed.
[0153] The wafer labeling method provided by the embodiment of the present application only needs to detect the type of the core particle, and the work of edge recognition and labeling of the core particle is saved, so that the difficulty and workload of wafer labeling are reduced, and the efficiency and accuracy of wafer labeling are greatly improved, and the labeling efficiency is improved by more than 35%.
[0154] Those skilled in the art can understand that the various aspects of the application can be implemented as a system, a method or a program product. Therefore, the various aspects of the application can be embodied as a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation combining hardware and software aspects, which can be collectively referred to as "circuitry", "module" or "system" here.
[0155] Based on the same technical concept, the electronic device provided by the embodiment of the application can be a notebook computer, a desktop computer, a tablet, a smart phone or other terminal device, can implement the steps of the wafer marking method, and can achieve the same technical effects.
[0156] Referring to Figure 10 The electronic device includes a processor 1001, a memory 1002, a display screen 1003 and a communication interface 1004, the communication interface 1004, the display screen 1003, the memory 1002 and the processor 1001 are connected through a bus 1005;
[0157] The communication interface 1004 is used for transceiving data;
[0158] The display screen 1003 is used for displaying an interactive page;
[0159] The memory 1002 stores a computer program, and the processor 1001 executes Figure 3 the steps of the wafer marking method according to the computer program.
[0160] It should be noted that Figure 10 is only an example, and the electronic device can also include a memory, a power supply, a mouse, a crystal oscillator and other conventional devices, which are not shown.
[0161] In the embodiments of the present application, the memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system and programs required for running instant messaging functions, etc.; and the data storage area can store various instant messaging information and operation instruction sets, etc. The memory can be a volatile memory such as a random-access memory (RAM), or a non-volatile memory such as a read-only memory, a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD), or any other medium capable of carrying or storing desired computer programs in the form of instructions or data structures and accessible by a computer, but is not limited thereto. The memory can be a combination of the above memories. The processor can include one or more central processing units (CPUs), GPUs or digital processing units, etc.
[0162] The specific connection medium between the communication interface, the memory and the processor is not limited in the embodiments of the present application. In the embodiments of the present application, the bus between the communication interface, the memory and the processor is described with a thick line in the figure, and the connection mode between other components is only schematically described and is not limited. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of description, only one thick line is described in the figure, but only one bus or only one type of bus is not described.
[0163] The embodiments of the present application also provide a computer-readable storage medium for storing some instructions, which can complete the steps of any wafer method in the foregoing embodiments when executed.
[0164] The embodiments of the present application also provide a computer program product for storing a computer program, which is used for executing the steps of any wafer method in the foregoing embodiments.
[0165] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can adopt a completely hardware embodiment, a completely software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can adopt a computer program product in the form of being implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program codes.
[0166] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks. Figure 1 one or more flow or blocks.
[0167] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure 1 one or more flow or blocks. Figure 1 one or more flow or blocks.
[0168] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks. Figure 1 one or more flow or blocks.
[0169] Obviously, numerous modifications and variations of the present application are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.
Claims
1. A wafer marking method characterized by comprising: The method comprises: in response to a labeling operation on an imported wafer image dataset, displaying a plurality of wafer images contained in the wafer image dataset and a label set used by the wafer image dataset in a labeling interface, the label set being determined according to a wafer model corresponding to the wafer image dataset; in response to an image selection operation triggered on the wafer image, adding a selected target wafer image to a labeling area of the labeling page, the target wafer image being any wafer image in the wafer image dataset; in response to a label selection operation triggered on the label set, presenting a selected target label as a labeling state in the labeling interface, the target label being any label in the label set; in response to a click operation triggered on a die belonging to the target label on the target wafer image, displaying a labeling box surrounding the die, and displaying the target label in a set region of the die.
2. The method of claim 1, wherein, in response to a click operation triggered on a die belonging to the target label on the target wafer image, displaying a labeling box surrounding the die, comprising: in response to the click operation triggered on the die belonging to the target label on the target wafer image, obtaining click coordinates of the click position, and combining a labeling box size preset for the target wafer image to determine a labeling image surrounding the die in the target wafer image; performing edge detection on the labeling image to obtain a plurality of detection boxes; generating and displaying the labeling box surrounding the die according to the plurality of detection boxes and the labeling box size.
3. The method of claim 2, wherein, the generating and displaying the labeling box surrounding the die according to the plurality of detection boxes and the labeling box size, comprising: traversing the plurality of detection boxes, and each time performing: when a current traversed detection box contains the click position according to the center coordinates and size of the current traversed detection box combined with the click coordinates, taking the current traversed detection box as a candidate detection box; determining a detection box area of each candidate detection box according to the size of the candidate detection box, and determining a labeling box area according to the labeling box size; determining a target detection box from the at least one candidate detection box according to the area difference between the at least one detection box area and the labeling box area, the target detection box representing the outline of the die; generating and displaying the labeling box surrounding the die according to the target detection box.
4. The method of claim 3, wherein, the generating and displaying the labeling box surrounding the die according to the target detection box, comprising: taking the center coordinates of the target detection box as the center point of the labeling box, and combining the labeling box size to determine and display the labeling box surrounding the die.
5. The method of any one of claims 2-4, wherein, the obtaining click coordinates of the click position, and combining a labeling box size preset for the target wafer image to determine a labeling image surrounding the die in the target wafer image, comprising: determining the origin coordinates of the labeling image according to the click coordinates and the labeling box size; According to the origin coordinates and a preset multiple of the bounding box size, a region image surrounding the core particle in the target wafer image is determined, the preset multiple being greater than 1; The region image is subjected to grayscale processing to generate a labeling image.
6. The method of any one of claims 1-4, wherein, In response to a label selection operation triggered for the label set, the method further comprises: In response to a selection operation of a labeling style in the toolbar of the labeling interface, the selected target labeling style is changed to a selected state; In response to the click operation triggered for the core particle belonging to the target label on the target wafer image, a labeling box surrounding the core particle is displayed, comprising: In response to the click operation triggered for the core particle belonging to the target label on the target wafer image, a labeling box surrounding the core particle is displayed in the target labeling style.
7. The method of any one of claims 1-4, wherein, In response to an image selection operation triggered for the wafer image, the method further comprises: In response to a setting operation triggered for the wafer model, a labeling setting interface is presented, and the labeling setting interface is used to set the label set and the bounding box size associated with each wafer model.
8. The method of claim 7, wherein, The method further comprises: In response to an import operation triggered for the wafer image dataset, a data import interface is presented, and the data import interface is used to set the wafer model corresponding to the wafer image dataset.
9. An electronic device, comprising: The device comprises a processor, a memory, a display screen and a communication interface, and the communication interface, the display screen, the memory and the processor are connected through a bus; The communication interface is used for transmitting and receiving data; The display screen is used for displaying a labeling interface; The memory stores a computer program, and the processor executes the following operations according to the computer program: In response to a labeling operation for the imported wafer image dataset, a plurality of wafer images included in the wafer image dataset and a label set used by the wafer image dataset are displayed in the labeling interface, and the label set is determined according to the wafer model corresponding to the wafer image dataset; In response to an image selection operation triggered for the wafer image, a selected target wafer image is added to a labeling area of the labeling page, and the target wafer image is any wafer image in the wafer image dataset; In response to a label selection operation triggered for the label set, a selected target label is presented in a labeling state in the labeling interface, and the target label is any label in the label set; In response to a click operation triggered for the core particle belonging to the target label on the target wafer image, a labeling box surrounding the core particle is displayed, and the target label is displayed in a set region of the core particle.
10. The electronic device of claim 9, wherein, The processor displays a labeling box surrounding the core particle in response to the click operation triggered for the core particle belonging to the target label on the target wafer image, and the specific operation is: In response to the click operation triggered for the core particle belonging to the target label on the target wafer image, a click coordinate of the click position is obtained, and a labeling image surrounding the core particle in the target wafer image is determined in combination with a pre-set bounding box size for the target wafer image; The labeling image is subjected to edge detection to obtain a plurality of detection boxes; According to the plurality of detection boxes and the size of the labeled box, a labeled box surrounding the core particle is generated and displayed.