Wafer defect detection method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a wafer defect detection method and device, electronic equipment and a storage medium, and belongs to the technical field of image detection. The method comprises the following steps: acquiring a target wafer image and wafer prior structure information; performing time domain feature extraction and distribution difference calculation on the target wafer image to obtain wafer time domain significance features and wafer image information distribution difference features; based on wafer prior structure information, performing phase-frequency separation on the target wafer image to obtain a wafer image amplitude spectrum and a wafer image phase spectrum; performing frequency domain feature extraction on the target wafer image based on the wafer image information distribution difference features, the wafer image amplitude spectrum and the wafer image phase spectrum to obtain wafer frequency domain saliency features; combining the wafer time domain saliency feature and the wafer frequency domain saliency feature to obtain a wafer saliency feature; and carrying out defect detection on the target wafer image based on the wafer saliency characteristics. According to the embodiment of the invention, the accuracy of wafer defect detection can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of image detection, and particularly to a wafer defect detection method and apparatus, an electronic device, and a storage medium. Background Art

[0002] Wafer defect detection refers to observing a wafer image to detect whether there are surface defects on the wafer corresponding to the wafer image, such as cracks, scratches, foreign object occlusion, etc.

[0003] Currently, the wafer defect detection method usually uses a pre-constructed label database to learn and train the extracted defect features, so as to construct a defect recognition model, and then uses the defect recognition model to detect defects in a new wafer image. However, this method is quite dependent on the accuracy of the pre-constructed label database. However, the label database usually marks wafers based on the results of manual wafer defect detection, which is prone to marking errors, resulting in inaccurate data in the label database and low accuracy of wafer defect detection by the defect recognition model. Therefore, how to improve the accuracy of wafer defect detection has become an urgent technical problem to be solved. Summary of the Invention

[0004] The main purpose of the embodiments of the present application is to propose a wafer defect detection method and apparatus, an electronic device, and a storage medium, aiming to improve the accuracy of wafer defect detection.

[0005] To achieve the above object, a first aspect of the embodiments of the present application proposes a wafer defect detection method, the method including:

[0006] Obtain a target wafer image and wafer prior structure information, where the wafer prior structure information represents the structural feature information of a defect-free wafer;

[0007] Extract time-domain features from the target wafer image to obtain wafer time-domain saliency features;

[0008] Based on the target wafer image, calculate the distribution difference of the target wafer image to obtain a wafer image information distribution difference feature;

[0009] Based on the wafer prior structure information, perform phase-frequency separation on the target wafer image to obtain a wafer image amplitude spectrum and a wafer image phase spectrum;

[0010] Based on the wafer image information distribution difference feature, the wafer image amplitude spectrum, and the wafer image phase spectrum, extract frequency-domain features from the target wafer image to obtain wafer frequency-domain saliency features;

[0011] Perform feature merging on the time-domain significant features and the frequency-domain significant features of the wafer to obtain the wafer significant features;

[0012] Based on the wafer significant features, perform defect detection on the target wafer image.

[0013] In some embodiments, the target wafer image includes wafer image pixel points. The extraction of the time-domain features from the target wafer image to obtain the time-domain significant features of the wafer includes:

[0014] Obtain the adjacent image pixel points of the wafer image pixel points;

[0015] Perform distance measurement on the wafer image pixel points and the adjacent image pixel points to obtain the pixel Euclidean distance;

[0016] Based on the pixel Euclidean distance, perform difference calculation on the wafer image pixel points and the adjacent image pixel points to obtain the pixel difference data;

[0017] Based on the pixel difference data and the pixel Euclidean distance, determine the time-domain significant features of the wafer.

[0018] In some embodiments, the determining the time-domain significant features of the wafer based on the pixel difference data and the pixel Euclidean distance includes:

[0019] Based on the pixel difference data, perform weighted processing on the pixel Euclidean distance to obtain the weighted spatial difference;

[0020] Based on the adjacent image pixel points, perform difference merging on the weighted spatial difference to obtain the time-domain significant features of the wafer.

[0021] In some embodiments, the calculating the distribution difference of the target wafer image based on the target wafer image to obtain the wafer image information distribution difference features includes:

[0022] Perform image segmentation on the target wafer image to obtain wafer segmentation regions;

[0023] Perform covariance matrix calculation on the wafer segmentation regions to obtain the regional covariance matrix;

[0024] Based on the regional covariance matrix, determine the wafer image information distribution difference features.

[0025] In some embodiments, the performing covariance matrix calculation on the wafer segmentation regions to obtain the regional covariance matrix includes:

[0026] Query the channels of the wafer segmentation area to obtain the image channels, and record the number of the image channels to obtain the number of image channels;

[0027] Extract the feature points of the wafer segmentation area to obtain the area feature points;

[0028] Based on the image channels, calculate the average value of the area feature points to obtain the average value of the feature points;

[0029] Based on the number of image channels, the area feature points and the average value of the feature points, determine the area covariance matrix.

[0030] In some embodiments, the phase-frequency separation of the target wafer image based on the prior wafer structure information to obtain the wafer image amplitude spectrum and the wafer image phase spectrum includes:

[0031] Perform Fourier transform on the target wafer image to obtain the target image spectrum information;

[0032] Based on the prior wafer structure information, perform spectrum range limitation to obtain the defect-free wafer image spectrum range;

[0033] Based on the defect-free wafer image spectrum range, perform filtering processing on the target image spectrum information to obtain the image filtered spectrum information;

[0034] Extract the amplitude of the image filtered spectrum information to obtain the wafer image amplitude spectrum;

[0035] Extract the phase of the image filtered spectrum information to obtain the wafer image phase spectrum.

[0036] In some embodiments, the frequency-domain feature extraction of the target wafer image based on the wafer image information distribution difference feature, the wafer image amplitude spectrum and the wafer image phase spectrum to obtain the wafer frequency-domain significance feature includes:

[0037] Normalize the wafer image amplitude spectrum based on the wafer image information distribution difference feature to obtain the normalized amplitude spectrum value;

[0038] Perform band-pass filtering on the wafer image phase spectrum to obtain the filtered phase spectrum value;

[0039] Merge the normalized amplitude spectrum value and the filtered phase spectrum value to obtain the merged spectrum data;

[0040] Adjust the real and imaginary parts of the merged spectrum data to obtain the real and imaginary adjusted spectrum data;

[0041] Perform inverse Fourier transform on the real and imaginary adjusted spectrum data to obtain the spatial-domain wafer image;

[0042] Perform spatial domain filtering on the spatial domain wafer image to obtain a filtered spatial domain image;

[0043] Perform modulo square processing on the filtered spatial domain image to obtain the wafer frequency domain saliency feature.

[0044] To achieve the above object, a second aspect of the embodiments of the present application provides a wafer defect detection device, the device includes:

[0045] A data acquisition module, configured to acquire a target wafer image and wafer prior structure information, where the wafer prior structure information represents the structural feature information of a defect-free wafer;

[0046] A time domain feature extraction module, configured to perform time domain feature extraction on the target wafer image to obtain a wafer time domain saliency feature;

[0047] A distribution difference calculation module, configured to calculate the distribution difference of the target wafer image based on the target wafer image to obtain a wafer image information distribution difference feature;

[0048] An image phase-frequency separation module, configured to perform phase-frequency separation on the target wafer image based on the wafer prior structure information to obtain a wafer image amplitude spectrum and a wafer image phase spectrum;

[0049] A frequency domain feature extraction module, configured to perform frequency domain feature extraction on the target wafer image based on the wafer image information distribution difference feature, the wafer image amplitude spectrum, and the wafer image phase spectrum to obtain a wafer frequency domain saliency feature;

[0050] A feature merging module, configured to merge the wafer time domain saliency feature and the wafer frequency domain saliency feature to obtain a wafer saliency feature;

[0051] A wafer defect detection module, configured to perform defect detection on the target wafer image based on the wafer saliency feature.

[0052] To achieve the above object, a third aspect of the embodiments of the present application provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the method of the first aspect is implemented.

[0053] To achieve the above object, a fourth aspect of the embodiments of the present application provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method of the first aspect is implemented.

[0054] The wafer defect detection method and device, electronic device, and storage medium proposed in this application provide a benchmark for subsequent wafer defect detection by acquiring a target wafer image and wafer prior structure information. Then, time-domain feature extraction and distribution difference calculation are performed on the target wafer image to obtain time-domain saliency features and information distribution difference features, preliminarily analyzing the image features of the target wafer image from the time-domain dimension and the pixel distribution perspective. Next, based on the wafer prior structure information, phase-frequency separation is performed on the target wafer image to obtain the wafer image amplitude spectrum and the wafer image phase spectrum. Further, based on the wafer image information distribution difference features, the wafer image amplitude spectrum, and the wafer image phase spectrum, frequency-domain feature extraction is performed on the target wafer image to obtain wafer frequency-domain saliency features, demonstrating the image features of the target wafer image from the frequency dimension. After that, the wafer time-domain saliency features and the wafer frequency-domain saliency features are merged to obtain wafer saliency features, integrating various image features in the time domain and the frequency domain, making the image features of the target wafer image more comprehensive and representative, thereby improving the accuracy and reliability of wafer defect detection. Finally, based on the wafer saliency features, defect detection is performed on the target wafer image, which can avoid misjudgment and missed judgment of wafer image defects, thereby improving the accuracy rate of wafer defect detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 is a flowchart of the wafer defect detection method provided by an embodiment of this application;

[0056] Figure 2 is Figure 1 a flowchart of step S102 in

[0057] Figure 3 is Figure 2 a flowchart of step S204 in

[0058] Figure 4 is Figure 1 a flowchart of step S103 in

[0059] Figure 5 is Figure 4 a flowchart of step S402 in

[0060] Figure 6 is Figure 1 a flowchart of step S104 in

[0061] Figure 7 is Figure 1 a flowchart of step S105 in

[0062] Figure 8 is a schematic structural diagram of the wafer defect detection device provided by an embodiment of this application;

[0063] Figure 9 It is a schematic diagram of the hardware structure of the electronic device provided by the embodiments of the present application. Specific embodiments

[0064] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0065] It should be noted that although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order from the module division in the device or the flowchart. Terms such as "first" and "second" in the specification, claims and the above-mentioned drawings are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence.

[0066] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application and are not intended to limit the present application.

[0067] First, several terms involved in the present application are analyzed:

[0068] Salient feature: A salient feature refers to a part in an image that has obvious differences or prominent manifestations relative to the surrounding information. These features are easily recognizable and noticeable visually or in analysis. For example, edges, corners, texture changes, or regions with strong color contrast in an image usually carry important information and play a key role in applications in fields such as image understanding.

[0069] Image detection field: The image detection field is an important branch of computer vision, which focuses on identifying, locating, and analyzing specific features, objects, or patterns from image data. This field uses various algorithms and technologies, such as edge detection, feature extraction, pattern recognition, and machine learning, to process and analyze image information. Image detection has a wide range of applications in multiple industries, including security monitoring, autonomous driving, medical image analysis, industrial automation, semiconductor defect detection, etc., aiming to improve the efficiency and accuracy of detection through automation, reduce manual intervention, and be able to process large-scale image data.

[0070] Wafer defect detection refers to observing a wafer image to detect whether there are surface defects on the wafer corresponding to the wafer image, such as cracks, scratches, foreign object occlusion, etc.

[0071] Currently, the wafer defect detection method usually utilizes a pre-constructed label database to learn and train the extracted defect features, thereby establishing a defect recognition model, and then using the defect recognition model to detect defects in new wafer images. However, this method relies relatively heavily on the accuracy of the pre-constructed label database. However, the label database usually marks wafers based on the results of manual wafer defect detection, which is prone to marking errors, resulting in inaccurate data in the label database and a low accuracy of wafer defect detection by the defect recognition model. Therefore, how to improve the accuracy of wafer defect detection has become an urgent technical problem to be solved.

[0072] Based on this, the embodiments of the present application provide a wafer defect detection method, device, electronic device, and storage medium, aiming to improve the accuracy of wafer defect detection.

[0073] The wafer defect detection method, device, electronic device, and storage medium provided by the embodiments of the present application are specifically described through the following embodiments. First, the wafer defect detection method in the embodiments of the present application is described.

[0074] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is to use a digital computer or a machine controlled by a digital computer to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results in terms of theory, method, technology, and application system.

[0075] Artificial intelligence basic technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. Artificial intelligence software technologies mainly include several major directions such as computer vision technology, robotics, biometric technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0076] The wafer defect detection method provided by the embodiments of the present application relates to the field of image detection technology. The wafer defect detection method provided by the embodiments of the present application can be applied to a terminal, or to a server side, or can also be software running on a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or can be configured as a server cluster or distributed system composed of multiple physical servers, or can also be configured as a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the wafer defect detection method, etc., but is not limited to the above forms.

[0077] This application can be used in numerous general-purpose or special-purpose computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and so on. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This application can also be practiced in a distributed computing environment where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0078] Figure 1 is an alternative flowchart of the wafer defect detection method provided by an embodiment of this application. Figure 1 The method in may include but is not limited to steps S101 to S107.

[0079] Step S101, obtain a target wafer image and wafer prior structure information, where the wafer prior structure information represents the structural feature information of a defect-free wafer.

[0080] Step S102, perform time-domain feature extraction on the target wafer image to obtain the wafer time-domain saliency feature.

[0081] Step S103, based on the target wafer image, perform distribution difference calculation on the target wafer image to obtain the wafer image information distribution difference feature.

[0082] Step S104, based on the wafer prior structure information, perform phase-frequency separation on the target wafer image to obtain the wafer image amplitude spectrum and the wafer image phase spectrum.

[0083] Step S105, based on the wafer image information distribution difference feature, the wafer image amplitude spectrum, and the wafer image phase spectrum, perform frequency-domain feature extraction on the target wafer image to obtain the wafer frequency-domain saliency feature.

[0084] Step S106, perform feature merging on the wafer time-domain saliency feature and the wafer frequency-domain saliency feature to obtain the wafer saliency feature.

[0085] Step S107, based on the wafer saliency feature, perform defect detection on the target wafer image.

[0086] Steps S101 to S107 illustrated in the embodiments of the present application, by acquiring a target wafer image and wafer prior structure information representing the structural characteristics of a defect-free wafer, perform time-domain feature extraction on the target wafer image to obtain the time-domain saliency features of the wafer. At the same time, calculate the distribution difference of the target wafer image to obtain the distribution difference features of the wafer image information. Then, perform phase-frequency separation on the target wafer image based on the above wafer prior structure information to obtain the amplitude spectrum and phase spectrum of the wafer image. Next, use the distribution difference features of the wafer image information, the amplitude spectrum of the wafer image, and the phase spectrum of the wafer image to perform frequency-domain feature extraction on the target wafer image to obtain the frequency-domain saliency features of the wafer. Then, merge the time-domain saliency features and frequency-domain saliency features of the wafer to obtain the saliency features of the wafer. Finally, perform defect detection on the target wafer image based on the saliency features of the wafer, and it can be determined whether there are surface defects in the target wafer image. Therefore, the present application provides a benchmark for subsequent wafer defect detection by acquiring the target wafer image and wafer prior structure information. Then, perform time-domain feature extraction and distribution difference calculation on the target wafer image to obtain time-domain saliency features and distribution difference features of the information. The image features of the target wafer image are initially analyzed from the time-domain dimension and the pixel distribution perspective. Then, based on the wafer prior structure information, perform phase-frequency separation on the target wafer image to obtain the amplitude spectrum and phase spectrum of the wafer image. Further, based on the distribution difference features of the wafer image information, the amplitude spectrum of the wafer image, and the phase spectrum of the wafer image, perform frequency-domain feature extraction on the target wafer image to obtain the frequency-domain saliency features of the wafer, which shows the image features of the target wafer image from the frequency dimension. After that, merge the time-domain saliency features and frequency-domain saliency features of the wafer to obtain the saliency features of the wafer, integrating various image features in the time domain and frequency domain, making the image features of the target wafer image more comprehensive and representative, thereby improving the accuracy and reliability of wafer defect detection. Finally, perform defect detection on the target wafer image based on the saliency features of the wafer, which can avoid misjudgment and missed judgment of wafer image defects, thereby improving the accuracy rate of wafer defect detection.

[0087] In step S101 of some embodiments, the target wafer image refers to the wafer image used for wafer defect detection. Through wafer defect detection, it can be detected whether there are surface defects on the wafer in the target wafer image. Among them, the surface defect features can be scratches, cracks, etc. on the wafer surface. It should also be noted that the target wafer image is composed of wafer image pixels, and the wafer image pixels contain information such as the color and brightness of the target wafer image. The wafer prior structure information refers to the surface structure characteristic information of a defect-free wafer.

[0088] This application can use a microscope to magnify the wafer to be detected, and then use a digital camera to capture the magnified wafer, so as to obtain a target wafer image. Further, information such as the uniformity and texture direction of the substrate of the defect-free wafer can be obtained through a pre-set defect-free wafer drawing or a microscopic image of the defect-free wafer, and then the information such as the uniformity and texture direction of the substrate of the defect-free wafer can be integrated to obtain the prior structure information of the wafer.

[0089] In step S102 of some embodiments, the wafer time-domain saliency feature refers to the manifestation form of the significant image feature in the target wafer image in the time domain or the spatial domain. Among them, the significant image feature refers to the feature in the target wafer image that can prominently display the surface defect area. For example, the parts in the target wafer image where there are obvious differences in brightness, texture, color, etc. from the normal surface area can be regarded as the significant image features of the target wafer image.

[0090] The embodiments of this application can measure the distance between any wafer image pixel point in the target wafer image and its adjacent adjacent image pixel points to obtain the pixel Euclidean distance. Further, based on the pixel Euclidean distance, the difference data between the wafer image pixel point and its adjacent adjacent image pixel points can be determined to obtain the pixel difference data. Finally, based on the above pixel difference data and the pixel Euclidean distance, the wafer time-domain saliency feature of the target wafer image is determined.

[0091] Specifically, please refer to Figure 2 , in some embodiments, step S102 may include but is not limited to steps S201 to S204:

[0092] Step S201, obtain the adjacent image pixel points of the wafer image pixel point;

[0093] Step S202, measure the distance between the wafer image pixel point and the adjacent image pixel points to obtain the pixel Euclidean distance;

[0094] Step S203, based on the pixel Euclidean distance, calculate the difference between the wafer image pixel point and the adjacent image pixel points to obtain the pixel difference data;

[0095] Step S204, based on the pixel difference data and the pixel Euclidean distance, determine the wafer time-domain saliency feature.

[0096] In step S201 of some embodiments, the adjacent image pixel point refers to the image pixel point within the neighborhood of the adjacent image pixel point, where the neighborhood refers to the pixel area adjacent to the wafer image pixel point in the target wafer image.

[0097] In the embodiments of the present application, the position of a pixel point in the wafer image can be determined by pixel positioning of the target wafer image, and then the surrounding adjacent pixel points can be found based on this position, so as to obtain adjacent image pixel points adjacent to the pixel point of the wafer image. For example, for the pixel point a of the wafer image in the target wafer image, the adjacent image pixel points can be the pixel points in the four directions of up, down, left, and right of the pixel point a of the wafer image, or can be the pixel points in the eight directions including the diagonal directions.

[0098] In step S202 of some embodiments, the pixel Euclidean distance refers to the straight-line distance between the pixel point of the wafer image and the adjacent image pixel point. It should be noted that the pixel Euclidean distance is used to measure the degree of proximity between the pixel point of the wafer image and the adjacent image pixel point in space.

[0099] In the embodiments of the present application, the pixel coordinates of the pixel point of the wafer image and the adjacent image pixel point can be determined according to the pre-established coordinate system. Further, based on the pixel coordinates of the pixel point of the wafer image and the pixel coordinates of the adjacent image pixel point, the straight-line distance between the pixel point of the wafer image and the adjacent image pixel point can be calculated to obtain the pixel Euclidean distance.

[0100] In step S203 of some embodiments, the pixel difference data refers to the data for measuring the differences in aspects such as color and brightness between the pixel point of the wafer image and the adjacent image pixel point.

[0101] In the embodiments of the present application, after obtaining the pixel Euclidean distance between the pixel point of the wafer image and the adjacent image pixel point, the similarity between the pixel point of the wafer image and the adjacent image pixel point can be measured based on the Gaussian function, and this similarity is used as the standard for measuring the difference between the pixel point of the wafer image and the adjacent image pixel point, that is, the pixel difference data.

[0102] Specifically, the following formula can be used to calculate the pixel difference data between the pixel point of the wafer image and the adjacent image pixel point:

[0103]

[0104] where c i represents the pixel point of the wafer image with label i, c j represents the adjacent image pixel point with label j, dist() represents the Euclidean distance function, D represents the pixel difference data, and exp represents the Gaussian function.

[0105] In step S204 of some embodiments, after obtaining the pixel difference data between the pixel point of the wafer image and the adjacent image pixel point, the wafer time-domain saliency feature of the pixel point of the wafer image can be defined by accumulating the product of the difference between the pixel point of the wafer image and all adjacent image pixel points in its neighborhood and the square of the pixel Euclidean distance.

[0106] Specifically, refer to Figure 3 , in some embodiments, step S204 may include but is not limited to steps S301 to S302:

[0107] Step S301, based on the pixel difference data, perform a weighted process on the pixel Euclidean distance to obtain a weighted spatial difference;

[0108] Step S302, based on the adjacent image pixel points, perform a difference merging on the weighted spatial difference to obtain the wafer time-domain significance feature.

[0109] In steps S301 and S302 of some embodiments, the weighted spatial difference can be obtained by multiplying the pixel difference data by the pixel Euclidean distance, that is, the weighted spatial difference. Further, the weighted spatial differences of each adjacent image pixel point are subjected to an addition merging process, so as to obtain the significance feature of the defect of the target wafer image in the time domain or the spatial domain, that is, the wafer time-domain significance feature.

[0110] Specifically, the wafer time-domain significance feature of the target wafer image can be calculated using the following formula:

[0111]

[0112] where SC represents the wafer time-domain significance feature, c i represents the wafer image pixel point with label i, c j represents the adjacent image pixel point with label j, and D represents the pixel difference data.

[0113] In steps S301 to S302 illustrated in this embodiment, based on the pixel difference data, perform a weighted process on the pixel Euclidean distance to obtain a weighted spatial difference, and then based on the adjacent image pixel points, perform a difference merging on the weighted spatial difference to obtain the wafer time-domain significance feature, which can highlight the difference between the pixels of the target wafer image and its adjacent pixels, thereby facilitating the identification and emphasis of the defect features in the image and improving the accuracy and reliability of wafer defect detection.

[0114] In steps S201 to S204 illustrated in this embodiment, by calculating the difference metric and the square of the Euclidean distance between the wafer image pixel points and the adjacent image pixel points, the local structural changes in the target wafer image can be effectively captured, thereby accurately highlighting and quantifying the defect features on the wafer surface, and improving the accuracy and reliability of wafer defect detection.

[0115] In step S103 of some embodiments, the wafer image information distribution difference feature refers to the pixel distribution difference feature between the regions of the target wafer image.

[0116] In the embodiments of the present application, the target wafer image can be segmented into multiple wafer segmentation regions, and then the regional covariance matrix of each wafer segmentation region can be calculated, and the pixel distribution difference between the multiple wafer segmentation regions can be calculated by using the regional covariance matrix.

[0117] Specifically, please refer to Figure 4 , in some embodiments, step S103 may include but is not limited to steps S401 to S403:

[0118] Step S401, perform image segmentation on the target wafer image to obtain wafer segmentation regions;

[0119] Step S402, calculate the covariance matrix of the wafer segmentation regions to obtain the regional covariance matrix;

[0120] Step S403, determine the difference feature of the wafer image information distribution based on the regional covariance matrix.

[0121] In step S401 of some embodiments, the wafer segmentation region refers to the image segmented from the target wafer image.

[0122] In the embodiments of the present application, a superpixel algorithm can be used to perform regional division on the target wafer image, so as to realize the image segmentation of the target wafer image. Specifically, according to the pixel values of the pixel points of the wafer image, regions with similar features such as color and texture in the target wafer image can be divided into the same region, so as to form multiple wafer segmentation regions.

[0123] In step S402 of some embodiments, the regional covariance matrix refers to the matrix describing the local structural features of the wafer segmentation region.

[0124] In the embodiments of the present application, the regional covariance matrix of the wafer segmentation region can be constructed according to the number of image channels of the wafer segmentation region and the average value of the feature points of each channel in the wafer segmentation region.

[0125] Specifically, please refer to Figure 5 , in some embodiments, step S402 may include but is not limited to steps S501 to S504:

[0126] Step S501, query the channels of the wafer segmentation region to obtain the image channels, and record the number of image channels to obtain the number of image channels;

[0127] Step S502, extract the feature points of the wafer segmentation region to obtain the regional feature points;

[0128] Step S503: Calculate the mean value of the regional feature points based on the image channels to obtain the average value of the feature points.

[0129] Step S504: Determine the regional covariance matrix based on the number of image channels, the regional feature points, and the average value of the feature points.

[0130] In step S501 of some embodiments, the image channels refer to different dimensional representations of the image data. For example, a color image has representations in three dimensions: red, green, and blue. Therefore, the image channels of a color image are red, green, and blue. The number of image channels refers to the quantity of image channels. For example, the number of image channels of a color image is 3.

[0131] In the embodiments of the present application, by identifying the image category of the wafer segmentation region, the image channels and the number of image channels of the wafer segmentation region can be determined. For example, when the wafer segmentation region is a color image, the image channels of the wafer segmentation region are red, green, and blue, and the number of image channels is 3.

[0132] In step S502 of some embodiments, the regional feature points refer to key points or points of interest in the wafer segmentation region. For example, the edge points, corner points, etc. in the wafer segmentation region can be the regional feature points of the wafer segmentation region.

[0133] In the embodiments of the present application, the feature points in the wafer segmentation region can be extracted according to the features of the feature points set in advance to obtain the regional feature points. For example, if the preset feature point features are regional edge points and corner points, then the regional feature points of the wafer segmentation region can be determined by searching for the regional edge points and corner points in the wafer segmentation region.

[0134] In step S503 of some embodiments, taking the image channels as the boundary, search for all the regional feature points in each image channel, and then calculate the mean value according to the number of regional feature points in the image channel, and the average value of the feature points in each image channel can be obtained.

[0135] In step S504 of some embodiments, based on the number of image channels, the regional feature points, and the average value of the feature points, the regional covariance matrix of the wafer segmentation region can be calculated using the following formula:

[0136]

[0137] where, v ij represents the regional covariance matrix of the i-th row and j-th column, M represents the number of image channels, m represents the m-th image channel, μ represents the average value of the feature points, i represents the number of rows of the regional covariance matrix, and j represents the number of columns of the regional covariance matrix.

[0138] In steps S501 to S504 illustrated in this embodiment,

[0139] In step S403 of some embodiments, the eigenvalues of the regional covariance matrix can be obtained by performing eigen - decomposition on the regional covariance matrix. Further, by taking the natural logarithm of each eigenvalue, logarithmic eigenvalues can be obtained. Finally, by calculating the variance of the logarithmic eigenvalues, the wafer image information distribution difference feature of the target wafer image can be obtained.

[0140] Specifically, the wafer image information distribution difference feature of the target wafer image can be calculated using the following formula:

[0141]

[0142] where α represents the information distribution difference characteristic, M represents the number of image channels, v ij represents the regional covariance matrix of the i - th row and j - th column, and λ m represents the m - th eigenvalue in the regional covariance matrix.

[0143] In steps S401 to S403 illustrated in this embodiment, by performing image segmentation on the target wafer image, the image can be divided into multiple wafer segmentation regions with similar characteristics, reducing the interference of noise and improving the accuracy of subsequent processing. Then, by calculating the regional covariance matrix for each segmented wafer segmentation region, the local structural characteristics of each wafer segmentation region can be further described, providing a basis for subsequent feature analysis. Finally, based on the regional covariance matrix, the wafer image information distribution difference feature is determined, which can quantify the differences between different wafer segmentation regions, thereby effectively highlighting the defective regions and improving the accuracy and reliability of wafer defect detection.

[0144] In step S104 of some embodiments, the wafer image amplitude spectrum can be the amplitude information of the target wafer image in the frequency domain, and the wafer image amplitude spectrum reflects the brightness and contrast characteristics of the target wafer image. The wafer image phase spectrum can be the phase information of the target wafer image in the frequency domain, and the wafer image phase spectrum reflects the texture and structural characteristics of the target wafer image.

[0145] To improve the accuracy of wafer defect detection, the embodiments of the present application can also transform the target wafer image into the frequency domain and segment the significant image features in the target wafer image from the whole of the target wafer image according to the amplitude and phase in the frequency domain. Specifically, the present application can perform a Fourier transform on the target wafer image to convert the target wafer image into target image spectrum information, and then filter the target image spectrum information according to the above - obtained wafer prior structure information to obtain the spectrum information of the region other than the normal region in the target wafer image, that is, the image - filtered spectrum information. Finally, by respectively extracting the amplitude information and phase information in the image - filtered spectrum information, the wafer image amplitude spectrum and the wafer image phase spectrum can be obtained.

[0146] Further, please refer to Figure 6 , in some embodiments, step S104 may include but is not limited to steps S601 to S605:

[0147] Step S601, perform a Fourier transform on the target wafer image to obtain target image spectrum information;

[0148] Step S602, based on the prior wafer structure information, perform spectrum range limitation to obtain the spectrum range of the defect-free wafer image;

[0149] Step S603, based on the spectrum range of the defect-free wafer image, perform filtering processing on the target image spectrum information to obtain image filtering spectrum information;

[0150] Step S604, perform amplitude extraction on the image filtering spectrum information to obtain the wafer image amplitude spectrum;

[0151] Step S605, perform phase extraction on the image filtering spectrum information to obtain the wafer image phase spectrum.

[0152] In steps S601 to S603 of some embodiments, the target image spectrum information refers to the representation form of the target wafer image in the frequency domain, and the target image spectrum information contains feature information such as brightness, texture, and structure of the target wafer image in the frequency domain. The spectrum range of the defect-free wafer image refers to the spectrum range in the frequency domain of the surface defect-free area in the target wafer image. The image filtering spectrum information refers to the spectrum information in the target image spectrum information excluding the spectrum range of the defect-free wafer image.

[0153] In the embodiments of the present application, the fast Fourier transform formula can be used to convert the time-domain information of the target wafer image into frequency-domain information, that is, the target image spectrum information. Further, according to the prior wafer structure information, the spectrum range of the wafer image without surface defects in the frequency domain is determined, that is, the spectrum range of the defect-free wafer image. Finally, according to the spectrum range of the defect-free wafer image, the filtering range is determined, and based on this filtering range, the target image spectrum information is filtered so that the target image spectrum information no longer contains the spectrum information of the wafer image without surface defects in the frequency domain, and the image filtering spectrum information is obtained.

[0154] In steps S604 and S605 of some embodiments, the amplitude information of the image filtering spectrum information, that is, the wafer image amplitude spectrum, can be obtained by performing a modulus operation on the image filtering spectrum information, or the phase information of the image filtering spectrum information, that is, the wafer image phase spectrum, can be obtained by extracting the phase angle information of the image filtering spectrum information.

[0155] Specifically, the present application can extract the wafer image amplitude spectrum in the target wafer image by using the following amplitude information extraction formula:

[0156]

[0157] where I represents the target wafer image, denotes the Fourier transform, H represents the filter, A(f q ) represents the wafer image amplitude spectrum, and L represents the logarithmic transform.

[0158] Furthermore, the present application can extract the wafer image phase spectrum in the target wafer image by using the following phase information extraction formula:

[0159]

[0160] where I represents the target wafer image, denotes the Fourier transform, H represents the filter, and F I represents the wafer image phase spectrum.

[0161] It should be noted that the filters in the amplitude information extraction formula and the phase information extraction formula are not the same. The filter in the amplitude information extraction formula is usually a band-pass filter, while the filter in the phase information extraction formula is usually a smoothing filter.

[0162] In steps S601 to S605 shown in this embodiment, by performing Fourier transform on the target wafer image to obtain the target image spectral information, and then performing filtering based on the spectral range of the defect-free wafer image provided by the wafer prior structure information, the defect information in the target wafer image, that is, the wafer image amplitude spectrum and the wafer image phase spectrum, can be highlighted, providing an important basis for subsequent wafer defect detection and analysis, and helping to improve the efficiency and accuracy of wafer defect detection.

[0163] In step S105 of some embodiments, the wafer frequency domain saliency feature refers to the manifestation form of the significant image features in the target wafer image in the frequency domain.

[0164] In the embodiments of the present application, the wafer image amplitude spectrum can be normalized, then the wafer image phase spectrum can be band-pass filtered, and the processed wafer image amplitude spectrum and the wafer image phase spectrum can be combined to form combined spectrum data. The combined spectrum data is adjusted for real and imaginary parts to obtain real and imaginary adjusted spectrum data, and then the real and imaginary adjusted spectrum data is converted back to the spatial domain to obtain the spatial domain wafer image. Further, the spatial domain image is filtered to obtain the filtered spatial domain image. Finally, the filtered spatial domain image is processed by taking the modulus square to obtain the wafer frequency domain saliency feature.

[0165] Specifically, please refer toFigure 7 , in some embodiments, step S105 may include but is not limited to steps S701 to S707:

[0166] Step S701, normalize the amplitude spectrum of the wafer image based on the difference characteristics of the wafer image information distribution to obtain the normalized amplitude spectrum value;

[0167] Step S702, perform band-pass filtering on the phase spectrum of the wafer image to obtain the filtered phase spectrum value;

[0168] Step S703, perform a merging process on the normalized amplitude spectrum value and the filtered phase spectrum value to obtain the merged spectrum data;

[0169] Step S704, perform real and imaginary part adjustment on the merged spectrum data to obtain the real and imaginary adjusted spectrum data;

[0170] Step S705, perform inverse Fourier transform on the real and imaginary adjusted spectrum data to obtain the wafer image in the spatial domain;

[0171] Step S706, perform spatial domain filtering on the wafer image in the spatial domain to obtain the filtered spatial domain image;

[0172] Step S707, perform modulus square processing on the filtered spatial domain image to obtain the wafer frequency domain saliency feature.

[0173] In steps S701 to S703 of some embodiments, the normalized amplitude spectrum value refers to the spectrum value after normalization processing of the amplitude spectrum value in the amplitude spectrum of the wafer image. The filtered phase spectrum value refers to the spectrum value after filtering processing of the phase spectrum value in the phase spectrum of the wafer image. The merged spectrum data refers to the data after adding the normalized amplitude spectrum value and the filtered phase spectrum value.

[0174] In the embodiments of the present application, the amplitude spectrum of the wafer image is normalized according to the difference characteristics of the wafer image information distribution to obtain the normalized amplitude spectrum value, so that the spectrum values of the amplitude spectrum of the wafer image can be compared and merged on the same scale. Further, the phase spectrum of the wafer image is subjected to band-pass filtering to obtain the filtered phase spectrum value, which can highlight the specific frequency components in the phase spectrum of the wafer image, thereby removing unnecessary noise and interference. Finally, the normalized amplitude spectrum value and the filtered phase spectrum value are merged to obtain the merged spectrum data, which can make full use of the information of both, so as to facilitate a more comprehensive and accurate description of the characteristics of the surface defect area in the target wafer image.

[0175] In steps S704 to S707 of some embodiments, by adjusting the real and imaginary parts of the combined spectrum data, the real-imaginary adjusted spectrum data can be obtained, ensuring that the image after the inverse Fourier transform of the combined spectrum data is a real image. Further, by performing the inverse Fourier transform on the real-imaginary adjusted spectrum data, a spatial domain wafer image is obtained, and by performing spatial domain filtering on the spatial domain wafer image, a filtered spatial domain image can be obtained, which can further highlight the defect features of the surface defect area of the target wafer image. Finally, by performing the modulo-square operation on the filtered spatial domain image, the representation of the significant image feature of the target wafer image in the frequency domain can be obtained, that is, the wafer frequency domain significant feature.

[0176] Specifically, the present application can extract the wafer frequency domain significant feature of the target wafer image by using the following formula:

[0177]

[0178] where S Asa represents the wafer frequency domain significant feature, k represents the image filtering spectrum information with the number 1, K represents the frequency component of the image filtering spectrum information, represents the inverse Fourier transform function, Z represents the real-imaginary part adjustment function, α represents the feature of the difference in the distribution of wafer image information, A(f q ) represents the amplitude spectrum of the wafer image, F I represents the phase spectrum of the wafer image, represents the band-pass filter.

[0179] In steps S701 to S707 illustrated in this embodiment, according to the difference characteristics of the distribution of wafer image information, the amplitude spectrum of the wafer image is normalized to obtain the normalized amplitude spectrum value, which can eliminate the brightness difference between different images, enabling all wafer image amplitude spectra to be compared on the same scale. The phase spectrum of the wafer image is band-pass filtered to obtain the filtered phase spectrum value, which can highlight specific frequency components in the wafer image phase spectrum, thereby removing unnecessary noise and interference. Further, the normalized amplitude spectrum value and the filtered phase spectrum value are combined to obtain the combined spectrum data, which can make full use of the information of the wafer image amplitude spectrum and the wafer image phase spectrum, enhancing the saliency of the surface defect area in the target wafer image. Secondly, the real and imaginary parts of the combined spectrum data are adjusted to obtain the real-imaginary adjusted spectrum data, which can ensure that the image after the inverse Fourier transform of the combined spectrum data is a real image, facilitating subsequent processing. When the real-imaginary adjusted spectrum data undergoes the inverse Fourier transform, the frequency domain information of the image filtering spectrum information can be converted back to the spatial domain, thereby obtaining the spatial domain wafer image. Then, by filtering the spatial domain wafer image, the characteristic performance of the surface defect area in the target wafer image can be further highlighted, and the background noise in the target wafer image can be removed to obtain the filtered spatial domain image. Finally, by performing the modulus square operation on the filtered spatial domain image, the wafer frequency domain saliency feature is obtained, making the surface defect area in the target wafer image more prominent in the image, facilitating subsequent wafer defect detection and analysis.

[0180] In step S106 of some embodiments, the wafer saliency feature refers to the joint optimization feature formed by combining the wafer time domain saliency feature of the target wafer image in the time domain and the wafer frequency domain saliency feature in the frequency domain.

[0181] In the embodiments of the present application, the following feature combination formula can be used to combine the wafer time domain saliency feature and the wafer frequency domain saliency feature to obtain the wafer saliency feature:

[0182]

[0183] Among them, S represents the wafer saliency feature, h(x, y) represents the constraint matrix, I c represents the wafer image pixel point in the target wafer image, S Asa (I c ) represents the wafer frequency domain saliency feature of the wafer image pixel point I c , SC(I c ) represents the wafer time domain saliency feature of the wafer image pixel point I c , S p (I c ) represents the phase image of the wafer image pixel point in the spectral transform target area obtained after phase information processing. denotes the inverse Fourier transform function, Z denotes the real and imaginary part adjustment function, and P denotes the phase function. denotes the band - pass filter, and N denotes the number of adjacent image pixels of the wafer image pixel I c of the adjacent image pixels, x denotes the abscissa of the wafer image pixel I c of the wafer image pixel I, and y denotes the ordinate of the wafer image pixel I c of the wafer image pixel I. denotes the wafer image pixel I c the abscissa of the adjacent image pixel numbered 0 within the neighborhood range of the wafer image pixel I denotes the wafer image pixel I c the ordinate of the adjacent image pixel numbered 0 within the neighborhood range of the wafer image pixel I, and both h1(i, j) and h2(i, j) denote the weight matrix of the adjacent image pixels of the wafer image pixel I c of the wafer image pixel I, o(x, y) denotes the parameter matrix used to determine whether the wafer image pixel belongs to the region where the wafer significant feature is located, and threshold denotes the threshold.

[0184] It should be noted that the above - mentioned phase image highlights the phase features of the target region, and the phase function is used to extract the phase image from the filtered phase spectrum values.

[0185] In step S107 of some embodiments, after obtaining the wafer significant features of the target wafer image, it is possible to determine whether the wafer corresponding to the target wafer image has surface defects through the wafer significant features. Specifically, when the wafer significant feature of the target wafer image is greater than or equal to the pre - set feature threshold, that is, the target wafer image has a wafer significant feature, it can be determined that the wafer corresponding to the target wafer image has surface defects; when the wafer significant feature of the target wafer image is less than the pre - set feature threshold, that is, the target wafer image does not have a wafer significant feature, it can be determined that the wafer corresponding to the target wafer image does not have surface defects.

[0186] This application provides a benchmark for subsequent wafer defect detection by obtaining a target wafer image and wafer prior structure information. Then, time-domain feature extraction and distribution difference calculation are performed on the target wafer image to obtain time-domain saliency features and information distribution difference features, preliminarily analyzing the image features of the target wafer image from the time-domain dimension and pixel distribution perspective. Next, based on the wafer prior structure information, phase-frequency separation is performed on the target wafer image to obtain the wafer image amplitude spectrum and the wafer image phase spectrum. Further, based on the wafer image information distribution difference features, the wafer image amplitude spectrum, and the wafer image phase spectrum, frequency-domain feature extraction is performed on the target wafer image to obtain wafer frequency-domain saliency features, demonstrating the image features of the target wafer image from the frequency dimension. After that, the wafer time-domain saliency features and the wafer frequency-domain saliency features are merged to obtain wafer saliency features, integrating various image features in the time domain and frequency domain, making the image features of the target wafer image more comprehensive and representative, thereby improving the accuracy and reliability of wafer defect detection. Finally, based on the wafer saliency features, defect detection is performed on the target wafer image, which can avoid misjudgment and missed judgment of wafer image defects, thus improving the accuracy rate of wafer defect detection.

[0187] Please refer to Figure 8 , this embodiment of the application also provides a wafer defect detection device that can implement the above wafer defect detection method. The device includes:

[0188] A data acquisition module 801, configured to acquire a target wafer image and wafer prior structure information, where the wafer prior structure information represents the structural feature information of a defect-free wafer;

[0189] A time-domain feature extraction module 802, configured to perform time-domain feature extraction on the target wafer image to obtain wafer time-domain saliency features;

[0190] A distribution difference calculation module 803, configured to perform distribution difference calculation on the target wafer image based on the target wafer image to obtain wafer image information distribution difference features;

[0191] An image phase-frequency separation module 804, configured to perform phase-frequency separation on the target wafer image based on the wafer prior structure information to obtain the wafer image amplitude spectrum and the wafer image phase spectrum;

[0192] A frequency-domain feature extraction module 805, configured to perform frequency-domain feature extraction on the target wafer image based on the wafer image information distribution difference features, the wafer image amplitude spectrum, and the wafer image phase spectrum to obtain wafer frequency-domain saliency features;

[0193] A feature merging module 806, configured to merge the wafer time-domain saliency features and the wafer frequency-domain saliency features to obtain wafer saliency features;

[0194] The wafer defect detection module 807 is used to detect defects in the target wafer image based on the significant features of the wafer.

[0195] The specific implementation manner of this wafer defect detection device is basically the same as the specific embodiments of the above wafer defect detection method, and will not be elaborated here.

[0196] An embodiment of this application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the above-mentioned wafer defect detection method is implemented. This electronic device can be any intelligent terminal including a tablet computer, an in-vehicle computer, etc.

[0197] Please refer to Figure 9 , Figure 9 which shows the hardware structure of an electronic device in another embodiment. The electronic device includes:

[0198] A processor 901, which can be implemented in ways such as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided by the embodiments of this application;

[0199] A memory 902, which can be implemented in forms such as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When implementing the technical solutions provided by the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 902, and the processor 901 is called to execute the wafer defect detection method of the embodiments of this application;

[0200] An input / output interface 903, which is used to implement information input and output;

[0201] A communication interface 904, which is used to implement communication interaction between this device and other devices, and can implement communication through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.);

[0202] A bus 905, which transmits information between various components of the device (such as the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);

[0203] Among them, the processor 901, the memory 902, the input / output interface 903, and the communication interface 904 are communicatively connected to each other inside the device through the bus 905.

[0204] An embodiment of the present application also provides a computer-readable storage medium storing a computer program, which when executed by a processor implements the above-mentioned wafer defect detection method.

[0205] As a non-transitory computer-readable storage medium, the memory can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor through a network. Examples of the above networks include, but are not limited to, the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0206] The wafer defect detection method, wafer defect detection device, electronic device, and storage medium provided by the embodiments of the present application obtain a target wafer image and wafer prior structure information representing the structural characteristics of a defect-free wafer, extract time-domain features from the target wafer image to obtain wafer time-domain saliency features. At the same time, calculate the distribution difference of the target wafer image to obtain the wafer image information distribution difference feature. Then, perform phase-frequency separation on the target wafer image based on the above wafer prior structure information to obtain the wafer image amplitude spectrum and the wafer image phase spectrum. Secondly, use the wafer image information distribution difference feature, the wafer image amplitude spectrum, and the wafer image phase spectrum to extract frequency-domain features from the target wafer image to obtain wafer frequency-domain saliency features. Then, merge the wafer time-domain saliency features and the frequency-domain saliency features to obtain wafer saliency features. Finally, perform defect detection on the target wafer image based on the wafer saliency features, and can determine whether there are surface defects in the target wafer image.

[0207] The embodiments described in the embodiments of the present application are for more clearly illustrating the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. Those skilled in the art will know that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present application are equally applicable to similar technical problems.

[0208] Those skilled in the art can understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or combine certain steps, or different steps.

[0209] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0210] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in systems and devices, can be implemented as software, firmware, hardware, and their appropriate combinations.

[0211] It should be understood that in this application, the terms "first", "second", "third", "fourth", etc. (if any) in the specification and the above drawings are used to distinguish similar objects and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of this application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that comprises a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0212] It should be understood that in this application, "at least one (item)" means one or more, and "a plurality" means two or more. "And / or" is used to describe the association relationship of associated objects and indicates that three relationships can exist. For example, "A and / or B" can mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally means that the associated objects before and after are in an "or" relationship. "At least one (one) of the following" or its similar expressions refer to any combination of these items, including any combination of single items (ones) or plural items (ones). For example, at least one (one) of a, b, or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0213] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the above division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. The displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be in electrical, mechanical or other forms.

[0214] The units described above as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0215] In addition, in each embodiment of the present application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0216] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes multiple instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present application. The foregoing storage medium includes: various media that can store programs such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0217] The preferred embodiments of the embodiments of the present application have been described above with reference to the drawings, but this does not limit the scope of the rights of the embodiments of the present application. Any modification, equivalent replacement, and improvement made by those skilled in the art without departing from the scope and essence of the embodiments of the present application shall be within the scope of the rights of the embodiments of the present application.

Claims

1. A wafer defect detection method, characterized in that, The method includes: Obtaining a target wafer image and wafer prior structure information, where the wafer prior structure information represents the structural feature information of a defect-free wafer; Performing time-domain feature extraction on the target wafer image to obtain the wafer time-domain saliency feature; Based on the target wafer image, calculating the distribution difference of the target wafer image to obtain the wafer image information distribution difference feature; Based on the wafer prior structure information, performing phase-frequency separation on the target wafer image to obtain the wafer image amplitude spectrum and the wafer image phase spectrum; Based on the wafer image information distribution difference feature, the wafer image amplitude spectrum, and the wafer image phase spectrum, performing frequency-domain feature extraction on the target wafer image to obtain the wafer frequency-domain saliency feature; Merging the wafer time-domain saliency feature and the wafer frequency-domain saliency feature to obtain the wafer saliency feature; Based on the wafer saliency feature, performing defect detection on the target wafer image.

2. The method according to claim 1, characterized in that The target wafer image includes wafer image pixels. The performing time-domain feature extraction on the target wafer image to obtain the wafer time-domain saliency feature includes: Obtaining adjacent image pixels of the wafer image pixels; Measuring the distance between the wafer image pixels and the adjacent image pixels to obtain the pixel Euclidean distance; Based on the pixel Euclidean distance, calculating the difference between the wafer image pixels and the adjacent image pixels to obtain the pixel difference data; Based on the pixel difference data and the pixel Euclidean distance, determining the wafer time-domain saliency feature.

3. The method according to claim 2, wherein The determining the wafer time-domain saliency feature based on the pixel difference data and the pixel Euclidean distance includes: Based on the pixel difference data, performing weighted processing on the pixel Euclidean distance to obtain the weighted spatial difference; Based on the adjacent image pixels, merging the differences of the weighted spatial difference to obtain the wafer time-domain saliency feature.

4. The method according to claim 1, wherein The calculating the distribution difference of the target wafer image based on the target wafer image to obtain the wafer image information distribution difference feature includes: Performing image segmentation on the target wafer image to obtain wafer segmentation regions; Calculating the covariance matrix of the wafer segmentation regions to obtain the regional covariance matrix; Based on the regional covariance matrix, determining the wafer image information distribution difference feature.

5. The method according to claim 4, wherein The calculating the covariance matrix of the wafer segmentation regions to obtain the regional covariance matrix includes: Querying the image channels of the wafer segmentation regions to obtain the image channels, and recording the number of the image channels to obtain the number of image channels; Extracting feature points of the wafer segmentation regions to obtain regional feature points; Based on the image channels, calculating the average value of the regional feature points to obtain the average value of the feature points; Based on the number of image channels, the regional feature points, and the average value of the feature points, determining the regional covariance matrix.

6. The method according to any one of claims 1-5, characterized in that, The performing phase-frequency separation on the target wafer image based on the wafer prior structure information to obtain the wafer image amplitude spectrum and the wafer image phase spectrum includes: Perform a Fourier transform on the target wafer image to obtain target image spectrum information; Based on the prior wafer structure information, perform spectrum range limitation to obtain the spectrum range of the defect-free wafer image; Based on the spectrum range of the defect-free wafer image, perform filtering processing on the target image spectrum information to obtain image filtering spectrum information; Extract the amplitude of the image filtering spectrum information to obtain the amplitude spectrum of the wafer image; Extract the phase of the image filtering spectrum information to obtain the phase spectrum of the wafer image.

7. The method according to any one of claims 1-5, characterized in that, Based on the difference feature of the wafer image information distribution, the amplitude spectrum of the wafer image, and the phase spectrum of the wafer image, perform frequency domain feature extraction on the target wafer image to obtain the significant frequency domain features of the wafer, including: Normalize the amplitude spectrum of the wafer image based on the difference feature of the wafer image information distribution to obtain a normalized amplitude spectrum value; Perform band-pass filtering on the phase spectrum of the wafer image to obtain a filtered phase spectrum value; Perform merging processing on the normalized amplitude spectrum value and the filtered phase spectrum value to obtain merged spectrum data; Perform real and imaginary part adjustment on the merged spectrum data to obtain real and imaginary adjusted spectrum data; Perform inverse Fourier transform on the real and imaginary adjusted spectrum data to obtain a spatial domain wafer image; Perform spatial domain filtering on the spatial domain wafer image to obtain a filtered spatial domain image; Perform modulus square processing on the filtered spatial domain image to obtain the significant frequency domain features of the wafer.

8. A wafer defect detection device, characterized in that, The device includes: A data acquisition module for acquiring a target wafer image and prior wafer structure information, where the prior wafer structure information represents the structural feature information of a defect-free wafer; A time domain feature extraction module for performing time domain feature extraction on the target wafer image to obtain significant time domain features of the wafer; A distribution difference calculation module for performing distribution difference calculation on the target wafer image based on the target wafer image to obtain the difference feature of the wafer image information distribution; An image phase-frequency separation module for performing phase-frequency separation on the target wafer image based on the prior wafer structure information to obtain the amplitude spectrum of the wafer image and the phase spectrum of the wafer image; A frequency domain feature extraction module for performing frequency domain feature extraction on the target wafer image based on the difference feature of the wafer image information distribution, the amplitude spectrum of the wafer image, and the phase spectrum of the wafer image to obtain the significant frequency domain features of the wafer; A feature merging module for merging the significant time domain features of the wafer and the significant frequency domain features of the wafer to obtain the significant features of the wafer; A wafer defect detection module for performing defect detection on the target wafer image based on the significant features of the wafer.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, it implements the wafer defect detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the wafer defect detection method according to any one of claims 1 to 7.

Images