CN118115496A - Wafer defect detection method and device - Google Patents
Wafer defect detection method and device Download PDFInfo
- Publication number
- CN118115496A CN118115496A CN202410501637.2A CN202410501637A CN118115496A CN 118115496 A CN118115496 A CN 118115496A CN 202410501637 A CN202410501637 A CN 202410501637A CN 118115496 A CN118115496 A CN 118115496A
- Authority
- CN
- China
- Prior art keywords
- defect
- image
- wafer
- defects
- detected
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000007547 defect Effects 0.000 title claims abstract description 566
- 238000001514 detection method Methods 0.000 title claims abstract description 143
- 238000001914 filtration Methods 0.000 claims abstract description 81
- 238000012545 processing Methods 0.000 claims abstract description 44
- 238000000034 method Methods 0.000 claims abstract description 34
- 238000013145 classification model Methods 0.000 claims abstract description 32
- 238000004458 analytical method Methods 0.000 claims abstract description 23
- 238000003384 imaging method Methods 0.000 claims description 43
- 238000007781 pre-processing Methods 0.000 claims description 17
- 238000000605 extraction Methods 0.000 claims description 11
- 238000013527 convolutional neural network Methods 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 9
- 238000009826 distribution Methods 0.000 claims description 6
- 238000007689 inspection Methods 0.000 claims description 6
- 230000004807 localization Effects 0.000 claims description 4
- 235000012431 wafers Nutrition 0.000 description 123
- 102000029749 Microtubule Human genes 0.000 description 15
- 108091022875 Microtubule Proteins 0.000 description 15
- 210000004688 microtubule Anatomy 0.000 description 15
- 238000012549 training Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 9
- 238000013135 deep learning Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 7
- 238000004590 computer program Methods 0.000 description 6
- 239000013078 crystal Substances 0.000 description 5
- 239000004065 semiconductor Substances 0.000 description 5
- 238000003860 storage Methods 0.000 description 5
- 230000004927 fusion Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000002146 bilateral effect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 230000001351 cycling effect Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000013140 knowledge distillation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012634 optical imaging Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 230000010287 polarization Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- HBMJWWWQQXIZIP-UHFFFAOYSA-N silicon carbide Chemical compound [Si+]#[C-] HBMJWWWQQXIZIP-UHFFFAOYSA-N 0.000 description 1
- 229910010271 silicon carbide Inorganic materials 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30148—Semiconductor; IC; Wafer
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Quality & Reliability (AREA)
- Testing Or Measuring Of Semiconductors Or The Like (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The embodiment of the invention discloses a wafer defect detection method and device, wherein the method comprises the following steps: acquiring an image to be detected of a wafer to be detected; performing defect positioning on the image to be detected based on the defect detection positioning model to obtain the position information of at least one first defect; for each detected first defect, performing feature analysis on a local defect image corresponding to the first defect to perform first filtering processing on the first defects so as to determine at least one second defect and at least one third defect in the first defects; performing feature analysis on the third defects based on the subdivision filter model to perform second filtering processing on each third defect so as to obtain fourth defects; and carrying out defect classification on the second defect and the fourth defect based on the defect classification model so as to obtain the target defect of the wafer to be detected. By adopting the invention, the accuracy of wafer defect detection can be improved.
Description
Technical Field
The present invention relates to the field of industrial machine vision automatic detection technology, and in particular, to a wafer defect detection method, a wafer defect detection device, a wafer defect detection computer device, and a wafer defect detection computer readable medium.
Background
As technology advances, silicon carbide wafer chip design sizes have become smaller and more compact, from 2.5D/3D stacking to TSV advanced packaging, wafer surface defects have increasingly affected semiconductor device yield and performance. Therefore, detecting and classifying defects on wafers is increasingly important.
Wafers have a variety of defects, mainly including micropipes, dislocations, faults, polytype and small angle grain boundaries, among which micropipe defects are the most serious ones to affect chip performance. Wafer defect detection is critical in semiconductor manufacturing processes. Traditionally, experienced engineers have used computer-aided tools to identify and classify defect types in semiconductor wafer maps. The process of manual cycling is often time consuming and visual inspection to produce accuracy of detection inconsistencies, affecting manufacturing efficiency and semiconductor device yield. Besides the influence of imaging hardware devices such as a microscope, a light source, a wafer tray and the like, the environment light changes, the angles of wafer placement operated by a detector, the crystal face inclination and the like can have inconsistency of brightness and color of the acquired wafer image, so that various characteristic changes on imaging of defects such as micro-tube splash areas, mixed crystal areas, weak micro-tubes, weak chains and the like are caused to have larger differences, and the defect detection performance is unstable.
The industrial defect detection has high requirements on the detection rate and the over-omission rate, and a deep learning target detection and classification model is difficult to reach the detection performance requirements in many application scenes, and particularly has higher detection performance requirements in the aspect of semiconductor defect detection. When the deep learning neural network is applied to detect and classify wafer defects, the effect of detecting and classifying the wafer defects after training a large number of samples is not very good, and the detection and classifying cannot be compatible with each other. Due to the changes of the ambient light, the different angles and inclinations of the wafer, and the like, the imaging of the defects such as weak microtubules, weak chains, and the like is changed in characteristics, brightness, color, and the like, so that the deep learning target detection classification model is difficult to balance between over-detection and omission detection, and the defects are difficult to accurately classify.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a wafer defect detection method, apparatus, computer device, and computer readable medium.
In a first aspect of the present invention, there is provided a wafer defect detection method, the method comprising:
acquiring an image to be detected of a wafer to be detected, wherein the image to be detected is obtained through polarized light imaging;
performing defect positioning on the image to be detected based on a defect detection positioning model to obtain position information of at least one first defect, wherein the defect detection positioning model is a deep convolution network model;
Performing feature analysis on the local defect image corresponding to the first defects aiming at each detected first defect to perform first filtering processing on the first defects so as to determine at least one second defect and at least one third defect in the first defects, wherein the third defects are weak wafer defects;
Performing feature analysis on the third defects based on a subdivision filtering model to perform second filtering processing on each third defect so as to obtain fourth defects, wherein the subdivision filtering model is a multi-stage convolutional neural network;
And carrying out defect classification on the second defect and the fourth defect based on a defect classification model so as to obtain the target defect of the wafer to be detected.
Optionally, the step of acquiring the image to be detected of the wafer to be detected further includes: and acquiring a plurality of polarized light images of the wafer to be detected as the image to be detected based on different scanning modes through a preset imaging system, wherein the imaging system comprises an imaging device, a polarizer and a light source, and the light source parameters of the imaging system are preset imaging parameters.
Optionally, after the step of obtaining the image to be detected of the wafer to be detected, the method further includes: image preprocessing is performed on the plurality of images to be detected, wherein the image preprocessing comprises one or more of image denoising processing, image enhancement processing and image normalization processing.
Optionally, the step of performing defect positioning on the image to be detected based on the defect detection positioning model to obtain position information of at least one first defect further includes: inputting the image to be detected into the defect detection positioning model, and acquiring at least one coordinate position output by the defect detection positioning model as position information of at least one first defect in the image to be detected.
Optionally, the step of performing feature analysis on the local defect image corresponding to the first defect for each detected first defect to perform first filtering processing on the first defect to determine at least one second defect and at least one third defect in the first defects, further includes: aiming at each first defect, acquiring a local defect image corresponding to the first defect from the image to be detected according to the position information corresponding to the first defect; extracting features of the local defect image to obtain local defect features corresponding to the local defect image, wherein the local defect features comprise one or more of brightness distribution features, regional features and gray level features; and filtering the local defect characteristics based on a preset characteristic threshold to obtain at least one second defect meeting the preset characteristic threshold and at least one third defect not meeting the preset characteristic threshold, wherein the third defect is a weak wafer defect.
Optionally, the subdivision filtering model includes a plurality of cascading sub-models, each cascading sub-model includes 1 feature extraction module and 1 filtering module;
The step of performing feature analysis on the third defects based on the subdivision filter model to perform second filtering processing on each third defect to obtain fourth defects, further includes: and inputting at least one third defect into the plurality of cascading submodels to obtain a filtering result output by the last cascading submodel, and determining at least one fourth defect according to the filtering result.
Optionally, the step of classifying the second defect and the fourth defect based on the defect classification model to obtain the target defect of the wafer to be detected further includes: inputting at least one second defect and at least one fourth defect into the defect classification model to perform feature extraction and defect classification on the at least one second defect and the at least one fourth defect through the defect classification model, and obtaining the target defect of the wafer to be detected.
Optionally, the target defect includes location information of the target defect and defect type identification;
The method further comprises the steps of: and identifying the position information and the defect type identification of the target defect in the coordinate image corresponding to the wafer to be detected, and outputting a defect map of the wafer to be detected.
In a second aspect of the present invention, there is provided a wafer defect inspection apparatus, the apparatus comprising:
the image acquisition module is used for acquiring an image to be detected of the wafer to be detected, wherein the image to be detected is obtained through polarized light imaging;
The defect detection positioning module is used for carrying out defect positioning on the image to be detected based on a defect detection positioning model to obtain the position information of at least one first defect, wherein the defect detection positioning model is a deep convolution network model;
The first filtering module is used for carrying out feature analysis on the local defect image corresponding to each detected first defect so as to carry out first filtering processing on the first defects, so as to determine at least one second defect and at least one third defect in the first defects, wherein the third defects are weak wafer defects;
The second filtering module is used for carrying out feature analysis on the third defects based on a subdivision filtering model so as to carry out second filtering processing on each third defect to obtain fourth defects, wherein the subdivision filtering model is a multi-stage convolutional neural network;
and the defect classification module is used for classifying the defects of the second defect and the fourth defect based on a defect classification model so as to acquire the target defect of the wafer to be detected.
Optionally, the apparatus further includes: and the image preprocessing module is used for carrying out image preprocessing on the plurality of images to be detected, wherein the image preprocessing comprises one or more of image denoising processing, image enhancement processing and image normalization processing.
In a third aspect of the invention, there is provided a computer device comprising a processor and a memory for storing a computer program; the processor is configured to execute the steps of the wafer defect detection method according to the first aspect according to the computer program.
In a fourth aspect of the present invention, there is provided a computer readable storage medium for storing a computer program for performing the steps of the wafer defect detection method as described in the first aspect above.
The embodiment of the invention has the following beneficial effects:
After the wafer defect detection method, the device, the computer equipment and the computer readable medium are adopted, an image to be detected of the wafer to be detected is obtained, and the image to be detected is obtained through polarized light imaging; performing defect positioning on the image to be detected based on a defect detection positioning model to obtain position information of at least one first defect, wherein the defect detection positioning model is a deep convolution network model; performing feature analysis on the local defect image corresponding to the first defects aiming at each detected first defect to perform first filtering processing on the first defects so as to determine at least one second defect and at least one third defect in the first defects, wherein the third defects are weak wafer defects; performing feature analysis on the third defects based on a subdivision filtering model to perform second filtering processing on each third defect so as to obtain fourth defects, wherein the subdivision filtering model is a multi-stage convolutional neural network; and carrying out defect classification on the second defect and the fourth defect based on a defect classification model so as to obtain the target defect of the wafer to be detected. After the wafer defect detection method, the device, the computer equipment and the computer readable medium are adopted, the accuracy of defect classification is greatly enhanced, and particularly, defect classification with more defect types and less characteristic difference among defects is achieved, so that ideal detection and classification effects are achieved. Specifically, by adopting a polarized light imaging scheme, the characteristic representation of wafer defects is improved, and the defect detection of wafer weak microtubules, weak chain defects and the like is facilitated; the deep learning target detection and the deep over-detection filtering method are combined, so that the defect detection rate is improved, and the false detection is reduced; the cascade of the multistage deep classification network is adopted, so that the accuracy of defect classification is greatly enhanced, and particularly, the defect detection accuracy is obviously improved aiming at defect classification with more defect types and fewer feature differences among defects. And filtering the pseudo defects of the slight microtubule features detected by the target detection by multistage deep convolution network cascade, and entering a defect classification model of the lightweight small model cascade for step-by-step classification. Through small model combination, on one hand, the defect detection rate is improved, and false detection and over detection of defects are reduced. On the other hand, the accuracy of defect classification is greatly enhanced, and particularly, defect classification with more defect types and less feature difference among defects is achieved, so that ideal detection and classification effects are achieved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Wherein:
FIG. 1 is a flow chart of a method for detecting wafer defects according to one embodiment;
FIG. 2 is a schematic diagram of a subdivision filter model in one embodiment;
FIG. 3 is a schematic diagram of a defect classification model in one embodiment;
FIG. 4 is a schematic diagram of a micropipe defect in one embodiment;
FIG. 5 is a schematic diagram of a wafer defect inspection apparatus according to one embodiment;
FIG. 6 is a schematic diagram of a computer device for performing the wafer defect detection method according to one embodiment.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In this embodiment, a wafer defect detection method is provided, so that the detection rate of wafer defects and the accuracy of defect classification are improved.
Specifically, referring to fig. 1, a flow chart of a wafer defect detection method is shown, wherein the method includes steps S102-S110 shown in fig. 1:
Step S102: and obtaining an image to be detected of the wafer to be detected, wherein the image to be detected is obtained through polarized light imaging.
In this embodiment, the imaging system uses polarized light imaging based on the physical characteristics and optical imaging characteristics of defects such as micropipes. In order to eliminate the influence of imaging hardware devices such as a microscope, a light source, a wafer tray and the like on imaging, before the imaging system scans and collects wafer images on the whole wafer, parameters such as a camera (image pickup device), a polarizer and the light source contained in the imaging system, the optimal position of the wafer tray and the like need to be adjusted.
Specifically, a wafer with typical weak microtubes and weak chain microtube defects is taken, polarization imaging is carried out, the image is evaluated, corresponding image quality indexes are calculated, defect detection is carried out on the wafer image, and then comprehensive evaluation is carried out on the wafer defect detection result. If the wafer defect detection and classification performance is not met, the imaging parameters are adjusted again until the wafer defect detection and classification performance is met, and especially the over-omission ratio of the weak wafer defect meets the index and defect classification requirements. The imaging parameters of the imaging system are then fixed and subsequently no longer changed.
After the imaging parameters of the imaging system are adjusted, the whole wafer to be detected is scanned and imaged. The scanning sequence is based on different equipment designs, and the scanning is performed in multiple modes such as from top to bottom, from left to right, or from bottom to top, from right to left, or in multiple modes at the same time, so as to obtain images of the wafer to be detected in multiple scanning modes. In one embodiment, depending on the accuracy of wafer defect inspection requirements and camera view, a wafer may need to be scanned to image several hundred wafers to be inspected as an image to be inspected for later inspection.
It should be noted that, in this embodiment, the image to be detected may be a color image or a gray-scale image, and may be specifically set according to the requirement of the defect classification model.
Further, in the present embodiment, in order to improve the stability of the detection performance, to ensure the accuracy of the subsequent defect detection, it is also necessary to perform preprocessing on the image, such as one or more of image denoising processing, image enhancement processing, and image normalization processing.
That is, the above method further includes step S103: and carrying out image preprocessing on the image to be detected.
Specifically, in the image denoising process, in the case that the image to be detected is a color image, the image to be detected is converted to the HSI color space to be denoised, or three channels of the color image are denoised independently, respectively. The gray scale image may be directly subjected to denoising processing. The algorithm corresponding to the image denoising process can use mean filtering, median filtering, bilateral filtering or the like. Because of the defects of weak microtubules and weak chains of the wafer, the characteristic difference is smaller, the characteristic details such as edges of the image are required to be kept as much as possible during denoising, and a denoising algorithm which can keep the edge details and local contrast by adopting bilateral filtering and the like has a better denoising effect.
In the image enhancement processing, an edge enhancement processing and a luminance uniformization processing may be performed on the image to be detected. The edge enhancement processing adopts a multi-scale Gaussian filtering mode to enhance the contrast between the wafer defect edge and the background, so that more false edges are prevented from being detected. And carrying out weighted fusion on the edge image details extracted by the multi-scale Gaussian filtering to strengthen the defect weak edges and the defect details. In addition, because the operator puts the wafer and has slight inclination during imaging, uneven brightness of the image to be detected of the wafer can be caused, and detection and classification of defects such as weak microtubes, weak chains and the like of the wafer are affected, so that uniform brightness treatment of the image to be detected of the wafer to be detected is needed. The brightness homogenizing treatment can adopt histogram equalization or illumination compensation algorithm.
In the image normalization processing, the average brightness and the average gray level of imaging of different detection machines or different pairs of detection wafer images are different, and the difference of the average brightness and the average gray level can be eliminated by carrying out the image normalization processing on the image to be detected, so that the stability of the detection performance is improved.
In this embodiment, the image preprocessing may include one or more of image denoising, image enhancement, and image normalization, and the processing order may be set as needed, which is not limited in this embodiment.
Step S104: and performing defect positioning on the image to be detected based on a defect detection positioning model to obtain the position information of at least one first defect, wherein the defect detection positioning model is a deep convolution network model.
In this embodiment, the defect detection positioning model is a deep convolutional network model, and may detect and position defects of a wafer to be detected to obtain position information corresponding to each defect contained therein. Here, the defect detection localization model allows for proper overstock in order to detect all defects to avoid missed detection.
The defect detection positioning model is a pre-trained deep convolutional network model, wafer defect samples are required to be acquired in the training process, the sample data are subjected to de-duplication processing, some defect sample data are incomplete, data generation enhancement is required to be performed, so that wafer defect samples of the training deep convolutional network model are distributed as completely as possible, then the data samples are marked, supervised training learning is performed on the deep convolutional network model, and wafer defect detection is performed after the model training is completed. Self-detecting wafer defects, particularly weak micropipes and weak chain defects that need to be detected, can ensure full detection and allow the model to fit over-inspection.
In a specific embodiment, a backbone network of the deep convolution network model adopts Darknet to introduce a mixed attention mechanism and a feature pyramid structure, adopts multi-scale multi-layer feature fusion and strengthens the extraction capability of weak features of a network bottom layer.
Specifically, the image to be detected is input into the defect detection positioning model, and at least one coordinate position output by the defect detection positioning model is obtained as position information of at least one first defect in the image to be detected. At least one first defect herein is all defects detected by the defect detection localization model, which requires further filtering and classification to determine the actual defects therein.
Step S106: and performing feature analysis on the local defect image corresponding to the first defects aiming at each detected first defect to perform first filtering processing on the first defects so as to determine at least one second defect and at least one third defect in the first defects, wherein the third defects are weak wafer defects.
At least one first defect detected by the defect detection positioning model, wherein a part of the first defect can be classified by a general defect detection method (a second defect, such as a wafer defect of a mixed crystal area, an overexposed area and the like), and a part of the first defect cannot be classified by a general defect detection method (a third defect, such as a weak micropipe, a weak chain and the like, and the like).
Specifically, for each first defect, acquiring a local defect image corresponding to the first defect from the image to be detected according to the position information corresponding to the first defect; extracting features of the local defect image to obtain local defect features corresponding to the local defect image, wherein the local defect features comprise one or more of brightness distribution features, regional features and gray level features; and filtering the local defect characteristics based on a preset characteristic threshold to obtain at least one second defect meeting the preset characteristic threshold and at least one third defect not meeting the preset characteristic threshold, wherein the third defect is a weak wafer defect.
In the specific implementation, in the step, a first defect detected by a defect detection positioning model is subjected to position information corresponding to the first defect, a local defect image of the first defect is obtained from an image to be detected, brightness distribution characteristics are calculated according to the local defect image, the local defect image is subjected to binarization processing and defect segmentation, and then the regional characteristics and gray scale characteristics of the segmented characteristics are calculated. One or more of a luminance distribution feature, a region feature, and a gray scale feature are taken as the local defect feature. The detected at least one first defect may then be filtered (first filtering process) based on a preset feature threshold to determine at least one second defect that may be classified by the local defect feature and at least one third defect that may not be classified by the local defect feature. The local defect characteristics corresponding to the second defect meet a preset characteristic threshold, and the local defect characteristics of the third defect do not meet the preset characteristic threshold, so that further classification and identification are needed in an additional mode.
Step S108: and carrying out feature analysis on the third defects based on a subdivision filtering model to carry out second filtering processing on each third defect so as to obtain fourth defects, wherein the subdivision filtering model is a multi-stage convolutional neural network.
In a specific embodiment, the subdivision filter model is a multi-stage convolutional neural network, e.g., a multi-stage compact deep convolutional network, comprising a plurality of cascaded sub-models, each comprising 1 feature extraction module and 1 filter module. See in particular fig. 2. The subdivision filter model may filter similar slight micropipe or slight chain defects.
Specifically, at least one third defect is input into the plurality of cascade sub-models, and each model in cascade is sequentially processed to obtain a filtering result output by the last cascade sub-model, and at least one fourth defect is determined according to the filtering result.
The deep convolutional network of each stage of the subdivision filter model is essentially a classification network, and may take the form of a simple Darknet or residual network structure. The networks of each level can be cascaded by the same network model or by different networks, the depth of the network of each level can be deepened step by step, and the stronger the feature extraction capability is, the stronger the filtering capability is. The classification categories of each level are mainly used for subdividing the wafer defects to be filtered, and the wafer defects to be detected can be classified into one type or several types, and mainly used are wafer defects to be filtered. The training sample data of each stage of model is marked and trained by adopting the first defect detected before, and can also be migrated by training a large filtering model with strong characteristic recognition capability. After training, cascading is carried out to form a subdivision filtering model. The first-stage filtering can filter out non-wafer defects outside the wafer, characters in the wafer and the like detected by the defect detection positioning model, the second-stage filtering can filter out non-microtubule defects such as wafer edge bubbles and carbon-based in the wafer detected by the defect detection positioning model, and the third-stage filtering mainly filters out similar slight microtubule or slight chain-shaped detection defects detected by the defect detection positioning model.
Step S110: and carrying out defect classification on the second defect and the fourth defect based on a defect classification model so as to obtain the target defect of the wafer to be detected.
Specifically, at least one second defect and at least one fourth defect are input into the defect classification model, so that feature extraction and defect classification are performed on the at least one second defect and the at least one fourth defect through the defect classification model, and the target defect of the wafer to be detected is obtained. The target defect comprises position information of the target defect and defect type identification.
Further, the defect classification model is a deep convolution network model, can be a cascade of a plurality of deep convolution network models, and can also be a cascade of a plurality of different deep convolution network models. Under the condition that the defect classification model is a cascaded deep convolution network model, the network depth of each stage can be different, the depth can be deepened step by step, and the better the feature recognition and classification performance is. Referring to fig. 3, a schematic diagram of a defect classification model is shown. Training sample data of each level of deep convolution network model is trained by adopting wafer defects with good category labels, or training by using a knowledge distillation mode to obtain small classification models with higher precision of each level. And after training, cascading to form a deep convolution network model. If the same deep convolutional network classification model cascade is adopted, classification categories are the same, if different deep convolutional network classification cascades are adopted, the classification category of each stage can be different, and the defect classification is divided into categories from coarse to fine. The first class classification can accurately classify the micro-pipe defects, the mixed crystal defects, the chain defects and the like detected by the defect detection positioning model, the second class classification classifies most of the slight micro-pipe or slight chain over-detection defects from the micro-pipe defects, the chain defects and the like, and the third class classification accurately classifies the slight micro-pipe or slight chain weak wafer defects.
Further, after the defect detection and defect classification of the whole wafer to be detected are completed, outputting the position coordinates and defect type identifiers of each defect (target defect), marking corresponding defects and types on the whole wafer coordinate system (coordinate image) of the wafer to be detected, and outputting a defect MAP (namely, a visual MAP of the wafer defect) of the wafer to be detected. Further, in an alternative embodiment, the number of wafer defects per unit area of the wafer, i.e., the MPD value, may also be calculated based on the above-mentioned detection result.
By adopting the wafer defect detection method, the characteristic representation of the wafer defects is improved through a polarized light imaging scheme, and the wafer defect detection method is more beneficial to the defect detection of weak microtubules, weak chain defects and the like of the wafer. In addition, compared with the common defect detection which adopts a larger deep learning target detection model, the method for detecting the deep learning target and filtering the deep over-detection is combined in the embodiment, so that the defect detection rate is improved, and the false over-detection is reduced. Furthermore, aiming at inaccurate defect classification by adopting a depth classification network, the application adopts cascade connection of multi-stage depth classification networks, greatly enhances the accuracy of defect classification, and particularly aims at defect classification with more defect categories and less feature difference among defects, and the defect detection accuracy is improved more obviously. In order to strengthen the detection capability of the defect detection positioning model, the defect detection positioning model is subjected to proper over-detection in order to avoid slight microtubule defect omission caused by imaging difference, then pseudo defects of slight microtubule characteristics detected and inspected by targets are filtered through multistage deep convolution network cascade, and the pseudo defects are classified step by step in a defect classification model cascade of lightweight small models. Through small model combination, on one hand, the defect detection rate is improved, and false detection and over detection of defects are reduced. On the other hand, the accuracy of defect classification is greatly enhanced, and particularly, defect classification with more defect types and less feature difference among defects is achieved, so that ideal detection and classification effects are achieved.
In the application, in order to eliminate the difference of brightness and color of the collected wafer image caused by the environment, the angle of the wafer placement operated by the detection personnel, the crystal face inclination and the like as much as possible, the collected image is subjected to brightness, contrast and characteristic similarity evaluation, and then the image is subjected to image preprocessing such as brightness equalization, contrast enhancement and the like, so that the brightness, the color and the definition of the image to be detected tend to be consistent. For example, the wafer microtubule defect has a large number of defects, and the defect of weak microtubules, weak chains, etc. is difficult to balance due to the omission of the microtubule defect detection, and in a specific embodiment, please refer to the schematic diagram of the typical microtubule defect shown in fig. 4. Since the over-detection of these weak defects has no strong feature boundaries, only subtle variations in brightness and local contrast are manifested. In the application, the deep learning target detection and classification model adopts multi-level multi-scale low-layer feature fusion and attention introducing mechanism to strengthen the extraction capability of weak features of the model.
In another embodiment, as shown in fig. 5, there is provided a wafer defect detection apparatus, including:
The image acquisition module 102 is used for acquiring an image to be detected of the wafer to be detected, wherein the image to be detected is obtained through polarized light imaging;
The defect detection positioning module 104 is configured to perform defect positioning on the image to be detected based on a defect detection positioning model, and obtain position information of at least one first defect, where the defect detection positioning model is a deep convolution network model;
A first filtering module 106, configured to perform feature analysis on the local defect image corresponding to the first defect for each detected first defect to perform a first filtering process on the first defects, so as to determine at least one second defect and at least one third defect in the first defects, where the third defect is a weak wafer defect;
A second filtering module 108, configured to perform a feature analysis on the third defects based on a subdivision filtering model to perform a second filtering process on each third defect to obtain a fourth defect, where the subdivision filtering model is a multi-level convolutional neural network;
and the defect classification module 110 is configured to perform defect classification on the second defect and the fourth defect based on a defect classification model, so as to obtain the target defect of the wafer to be detected.
In an alternative embodiment, as shown in fig. 5, the apparatus further includes an image preprocessing module 103, configured to perform image preprocessing on a plurality of images to be detected, where the image preprocessing includes one or more of image denoising, image enhancement, and image normalization.
In an alternative embodiment, the image acquisition module 102 is further configured to acquire, by using a preset imaging system, based on different scanning modes, multiple polarized light images of the wafer to be detected as the image to be detected, where the imaging system includes an image capturing device, a polarizer, and a light source, and the light source parameter of the imaging system is a preset imaging parameter.
In an alternative embodiment, the defect detection positioning module 104 is further configured to input the image to be detected into the defect detection positioning model, and obtain at least one coordinate position output by the defect detection positioning model as the position information of at least one first defect in the image to be detected.
In an optional embodiment, the first filtering module 106 is further configured to obtain, for each first defect, a local defect image corresponding to the first defect in the image to be detected according to location information corresponding to the first defect; extracting features of the local defect image to obtain local defect features corresponding to the local defect image, wherein the local defect features comprise one or more of brightness distribution features, regional features and gray level features; and filtering the local defect characteristics based on a preset characteristic threshold to obtain at least one second defect meeting the preset characteristic threshold and at least one third defect not meeting the preset characteristic threshold, wherein the third defect is a weak wafer defect.
In an alternative embodiment, the subdivision filter model includes a plurality of cascading sub-models, each cascading sub-model including 1 feature extraction module and 1 filter module; the second filtering module 108 is further configured to perform a feature analysis on the third defects based on the subdivision filtering model to perform a second filtering process on each third defect to obtain a fourth defect, and further includes:
and inputting at least one third defect into the plurality of cascading submodels to obtain a filtering result output by the last cascading submodel, and determining at least one fourth defect according to the filtering result.
In an alternative embodiment, the defect classification module 110 is further configured to input at least one second defect and at least one fourth defect into the defect classification model, so as to perform feature extraction and defect classification on the at least one second defect and the at least one fourth defect through the defect classification model, and obtain the target defect of the wafer to be inspected.
In an alternative embodiment, the target defect includes location information and defect type identification of the target defect; as shown in fig. 5, the apparatus further includes a visual output module 111, configured to identify, in a coordinate image corresponding to the wafer to be detected, the location information and the defect type identifier of the target defect, and output a defect map of the wafer to be detected.
Fig. 6 is an internal structural diagram of a computer device implementing the above wafer defect detection method in one embodiment. The computer device may specifically be a terminal or a server. As shown in fig. 6, the computer device includes a processor, a memory, and a network interface connected by a system bus. The memory includes a nonvolatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system, and may also store a computer program which, when executed by a processor, causes the processor to implement the method described above. The internal memory may also have stored therein a computer program which, when executed by a processor, causes the processor to perform the method described above. It will be appreciated by those skilled in the art that the structure shown in FIG. 6 is merely a block diagram of some of the structures associated with the present inventive arrangements and is not limiting of the computer device to which the present inventive arrangements may be applied, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
After the wafer defect detection method, the device, the computer equipment and the computer readable medium are adopted, an image to be detected of the wafer to be detected is obtained, and the image to be detected is obtained through polarized light imaging; performing defect positioning on the image to be detected based on a defect detection positioning model to obtain position information of at least one first defect, wherein the defect detection positioning model is a deep convolution network model; performing feature analysis on the local defect image corresponding to the first defects aiming at each detected first defect to perform first filtering processing on the first defects so as to determine at least one second defect and at least one third defect in the first defects, wherein the third defects are weak wafer defects; performing feature analysis on the third defects based on a subdivision filtering model to perform second filtering processing on each third defect so as to obtain fourth defects, wherein the subdivision filtering model is a multi-stage convolutional neural network; and carrying out defect classification on the second defect and the fourth defect based on a defect classification model so as to obtain the target defect of the wafer to be detected. After the wafer defect detection method, the device, the computer equipment and the computer readable medium are adopted, the accuracy of defect classification is greatly enhanced, and particularly, defect classification with more defect types and less characteristic difference among defects is achieved, so that ideal detection and classification effects are achieved. Specifically, by adopting a polarized light imaging scheme, the characteristic representation of wafer defects is improved, and the defect detection of wafer weak microtubules, weak chain defects and the like is facilitated; the deep learning target detection and the deep over-detection filtering method are combined, so that the defect detection rate is improved, and the false detection is reduced; the cascade of the multistage deep classification network is adopted, so that the accuracy of defect classification is greatly enhanced, and particularly, the defect detection accuracy is obviously improved aiming at defect classification with more defect types and fewer feature differences among defects. And filtering the pseudo defects of the slight microtubule features detected by the target detection by multistage deep convolution network cascade, and entering a defect classification model of the lightweight small model cascade for step-by-step classification. Through small model combination, on one hand, the defect detection rate is improved, and false detection and over detection of defects are reduced. On the other hand, the accuracy of defect classification is greatly enhanced, and particularly, defect classification with more defect types and less feature difference among defects is achieved, so that ideal detection and classification effects are achieved.
Those skilled in the art will appreciate that all or part of the processes in the methods of the above embodiments may be implemented by a computer program for instructing relevant hardware, where the program may be stored in a non-volatile computer readable storage medium, and where the program, when executed, may include processes in the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (SYNCHLINK) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples illustrate only a few embodiments of the application and are described in detail herein without thereby limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of protection of the present application is to be determined by the appended claims.
Claims (10)
1. A method for detecting wafer defects, the method comprising:
acquiring an image to be detected of a wafer to be detected, wherein the image to be detected is obtained through polarized light imaging;
performing defect positioning on the image to be detected based on a defect detection positioning model to obtain position information of at least one first defect, wherein the defect detection positioning model is a deep convolution network model;
Performing feature analysis on the local defect image corresponding to the first defects aiming at each detected first defect to perform first filtering processing on the first defects so as to determine at least one second defect and at least one third defect in the first defects, wherein the third defects are weak wafer defects;
Performing feature analysis on the third defects based on a subdivision filtering model to perform second filtering processing on each third defect so as to obtain fourth defects, wherein the subdivision filtering model is a multi-stage convolutional neural network;
And carrying out defect classification on the second defect and the fourth defect based on a defect classification model so as to obtain the target defect of the wafer to be detected.
2. The method for inspecting a wafer defect according to claim 1, wherein the step of acquiring the image to be inspected of the wafer to be inspected further comprises:
And acquiring a plurality of polarized light images of the wafer to be detected as the image to be detected based on different scanning modes through a preset imaging system, wherein the imaging system comprises an imaging device, a polarizer and a light source, and the light source parameters of the imaging system are preset imaging parameters.
3. The method for inspecting a wafer defect according to claim 2, further comprising, after the step of acquiring the image to be inspected of the wafer to be inspected:
image preprocessing is performed on the plurality of images to be detected, wherein the image preprocessing comprises one or more of image denoising processing, image enhancement processing and image normalization processing.
4. The method according to claim 1, wherein the step of performing defect localization on the image to be inspected based on a defect detection localization model to obtain location information of at least one first defect further comprises:
inputting the image to be detected into the defect detection positioning model, and acquiring at least one coordinate position output by the defect detection positioning model as position information of at least one first defect in the image to be detected.
5. The method of claim 1, wherein the step of performing feature analysis on the partial defect image corresponding to the first defects for each of the detected first defects to perform a first filtering process on the first defects to determine at least one second defect and at least one third defect among the first defects, further comprises:
Aiming at each first defect, acquiring a local defect image corresponding to the first defect from the image to be detected according to the position information corresponding to the first defect;
Extracting features of the local defect image to obtain local defect features corresponding to the local defect image, wherein the local defect features comprise one or more of brightness distribution features, regional features and gray level features;
And filtering the local defect characteristics based on a preset characteristic threshold to obtain at least one second defect meeting the preset characteristic threshold and at least one third defect not meeting the preset characteristic threshold, wherein the third defect is a weak wafer defect.
6. The wafer defect detection method of claim 1, wherein the subdivision filter model comprises a plurality of cascading sub-models, each cascading sub-model comprising 1 feature extraction module and 1 filter module;
the step of performing feature analysis on the third defects based on the subdivision filter model to perform second filtering processing on each third defect to obtain fourth defects, further includes:
and inputting at least one third defect into the plurality of cascading submodels to obtain a filtering result output by the last cascading submodel, and determining at least one fourth defect according to the filtering result.
7. The method according to claim 1, wherein the step of classifying the second and fourth defects based on a defect classification model to obtain the target defect of the wafer to be inspected further comprises:
Inputting at least one second defect and at least one fourth defect into the defect classification model to perform feature extraction and defect classification on the at least one second defect and the at least one fourth defect through the defect classification model, and obtaining the target defect of the wafer to be detected.
8. The method according to claim 1, wherein the target defect includes location information of the target defect and defect class identification;
The method further comprises the steps of:
And identifying the position information and the defect type identification of the target defect in the coordinate image corresponding to the wafer to be detected, and outputting a defect map of the wafer to be detected.
9. A wafer defect inspection apparatus, the apparatus comprising:
the image acquisition module is used for acquiring an image to be detected of the wafer to be detected, wherein the image to be detected is obtained through polarized light imaging;
The defect detection positioning module is used for carrying out defect positioning on the image to be detected based on a defect detection positioning model to obtain the position information of at least one first defect, wherein the defect detection positioning model is a deep convolution network model;
The first filtering module is used for carrying out feature analysis on the local defect image corresponding to each detected first defect so as to carry out first filtering processing on the first defects, so as to determine at least one second defect and at least one third defect in the first defects, wherein the third defects are weak wafer defects;
The second filtering module is used for carrying out feature analysis on the third defects based on a subdivision filtering model so as to carry out second filtering processing on each third defect to obtain fourth defects, wherein the subdivision filtering model is a multi-stage convolutional neural network;
and the defect classification module is used for classifying the defects of the second defect and the fourth defect based on a defect classification model so as to acquire the target defect of the wafer to be detected.
10. The wafer defect detection apparatus of claim 9, wherein the apparatus further comprises:
And the image preprocessing module is used for carrying out image preprocessing on the plurality of images to be detected, wherein the image preprocessing comprises one or more of image denoising processing, image enhancement processing and image normalization processing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410501637.2A CN118115496B (en) | 2024-04-25 | 2024-04-25 | Wafer defect detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410501637.2A CN118115496B (en) | 2024-04-25 | 2024-04-25 | Wafer defect detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118115496A true CN118115496A (en) | 2024-05-31 |
| CN118115496B CN118115496B (en) | 2024-08-13 |
Family
ID=91208857
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410501637.2A Active CN118115496B (en) | 2024-04-25 | 2024-04-25 | Wafer defect detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118115496B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118329777A (en) * | 2024-06-11 | 2024-07-12 | 深圳市华拓半导体技术有限公司 | Method and device for detecting surface defects of bare wafer, terminal equipment and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010165700A (en) * | 2009-01-13 | 2010-07-29 | Panasonic Corp | Method of inspecting defect of wafer and method of manufacturing semiconductor device |
| KR20200086485A (en) * | 2019-01-09 | 2020-07-17 | 에스케이텔레콤 주식회사 | Method and Apparatus for Recognizing Defect Patterns in Wafer Map |
| CN113538392A (en) * | 2021-07-26 | 2021-10-22 | 长江存储科技有限责任公司 | Wafer detection method, wafer detection equipment and storage medium |
| CN113570003A (en) * | 2021-09-23 | 2021-10-29 | 深圳新视智科技术有限公司 | Feature fusion defect detection method and device based on attention mechanism |
| CN116542908A (en) * | 2023-03-30 | 2023-08-04 | 深圳市格灵精睿视觉有限公司 | Wafer defect detection method and device, electronic equipment and storage medium |
| WO2023159961A1 (en) * | 2022-02-25 | 2023-08-31 | 季华实验室 | Defect detection method, apparatus and device, and computer-readable storage medium |
| CN117611879A (en) * | 2023-11-06 | 2024-02-27 | 上海朋熙半导体有限公司 | Defect detection method, device, equipment and computer readable medium |
-
2024
- 2024-04-25 CN CN202410501637.2A patent/CN118115496B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010165700A (en) * | 2009-01-13 | 2010-07-29 | Panasonic Corp | Method of inspecting defect of wafer and method of manufacturing semiconductor device |
| KR20200086485A (en) * | 2019-01-09 | 2020-07-17 | 에스케이텔레콤 주식회사 | Method and Apparatus for Recognizing Defect Patterns in Wafer Map |
| CN113538392A (en) * | 2021-07-26 | 2021-10-22 | 长江存储科技有限责任公司 | Wafer detection method, wafer detection equipment and storage medium |
| CN113570003A (en) * | 2021-09-23 | 2021-10-29 | 深圳新视智科技术有限公司 | Feature fusion defect detection method and device based on attention mechanism |
| WO2023159961A1 (en) * | 2022-02-25 | 2023-08-31 | 季华实验室 | Defect detection method, apparatus and device, and computer-readable storage medium |
| CN116542908A (en) * | 2023-03-30 | 2023-08-04 | 深圳市格灵精睿视觉有限公司 | Wafer defect detection method and device, electronic equipment and storage medium |
| CN117611879A (en) * | 2023-11-06 | 2024-02-27 | 上海朋熙半导体有限公司 | Defect detection method, device, equipment and computer readable medium |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118329777A (en) * | 2024-06-11 | 2024-07-12 | 深圳市华拓半导体技术有限公司 | Method and device for detecting surface defects of bare wafer, terminal equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118115496B (en) | 2024-08-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20230342925A1 (en) | System and method for calculating focus variation for a digital microscope | |
| CN107507173B (en) | No-reference definition evaluation method and system for full-slice image | |
| CN118115496B (en) | Wafer defect detection method and device | |
| CN110402487B (en) | Method and apparatus for classifying defects detected from a wafer | |
| WO2021139258A1 (en) | Image recognition based cell recognition and counting method and apparatus, and computer device | |
| CN111292305A (en) | Improved YOLO-V3 metal processing surface defect detection method | |
| CN104487843B (en) | Image processing apparatus, program, image processing method, computer-readable medium and image processing system | |
| CN111325713A (en) | Wood defect detection method, system and storage medium based on neural network | |
| CN110110799A (en) | Cell sorting method, device, computer equipment and storage medium | |
| CN109671058B (en) | Defect detection method and system for large-resolution image | |
| CN113962975B (en) | System for carrying out quality evaluation on pathological slide digital image based on gradient information | |
| CN113706490B (en) | Wafer defect detection method | |
| JP2022027473A (en) | Generation of training data usable for inspection of semiconductor sample | |
| CN112215790A (en) | KI67 index analysis method based on deep learning | |
| KR20180081820A (en) | Registration and design in the die internal inspection Reduction of the noise caused by the peripheral part | |
| CN110853006A (en) | Method for evaluating quality of digital pathological image acquired by scanner | |
| CN112150460A (en) | Detection method, detection system, device, and medium | |
| Kanwal et al. | Quantifying the effect of color processing on blood and damaged tissue detection in whole slide images | |
| CN117495856B (en) | Wafer surface detection method, device, equipment and medium based on deep learning | |
| CN116563298A (en) | Cross line center sub-pixel detection method based on Gaussian fitting | |
| US20030185432A1 (en) | Method and system for image registration based on hierarchical object modeling | |
| CN116363097A (en) | Defect detection method and system for photovoltaic panel | |
| US20230206417A1 (en) | Detection of defects using a computationally efficient segmentation approach | |
| US20030185431A1 (en) | Method and system for golden template image extraction | |
| CN111476129A (en) | Soil impurity detection method based on deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |