Micro defect enhancement detection method and system based on fusion local relative gradient
By fusing local relative gradients, defect feature enhancement image information is generated, which solves the problem of low real-time defect detection in the prior art, and realizes high-precision and real-time micro defect detection.
Patent Information
- Application Number
- CN202411779822.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-05
- Publication Date
- 2025-04-15
AI Technical Summary
In the prior art defect detection methods require a large amount of labeled data during training, resulting in low real-time performance and difficult to meet the needs of high-precision and real-time detection in semiconductor processes and other fields.
A small defect enhancement detection method based on fused local relative gradients is adopted. By obtaining the image information to be detected, the target area and multiple target domain information are determined, the local gradient feature information is generated, and the fusion process is performed to generate defect feature enhancement image information.
Significantly enhance the characteristics of micro defects, improve detection performance, reduce dependence on sample training, and improve real-time performance. It is suitable for surface defect detection of products such as semiconductor wafers and precision optical components.
Smart Images
Figure CN119831933A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of image processing, and in particular to a method and system for enhancing detection of tiny defects based on fusion of local relative gradients. Background Art
[0002] As semiconductor processes continue to move toward smaller dimensions, the requirements for defect detection are also increasing. This trend requires detection methods to have not only higher accuracy and sensitivity, but also excellent real-time performance.
[0003] At present, deep learning-based methods are usually used for defect detection. However, this method usually needs to rely on a large amount of labeled data during the training process. The entire training process is very time-consuming and has the problem of low real-time performance, which needs further improvement. Summary of the invention
[0004] Based on this, an embodiment of the present application provides a method and system for enhancing detection of tiny defects based on fusion of local relative gradients to solve the problem of low real-time performance in the prior art.
[0005] In a first aspect, an embodiment of the present application provides a method for enhanced detection of micro defects based on fusion of local relative gradients, the method comprising:
[0006] Obtaining image information to be detected;
[0007] Determining target area information and multiple target area information according to the image information to be detected;
[0008] Generate multiple local gradient feature information according to the multiple target area information and the preset local relative gradient feature operator information;
[0009] Based on the target area information, a plurality of local gradient feature information are fused to generate defect feature enhanced image information.
[0010] Compared with the prior art, the beneficial effects are as follows: the minute defect enhancement detection method based on fusion of local relative gradients provided in the embodiment of the present application, the terminal device can first obtain the image information to be detected, and then determine the target area information and multiple target field information according to the image information to be detected, and then generate multiple local gradient feature information according to the multiple target field information and the preset local relative gradient feature operator information, and finally, based on the target area information, the multiple local gradient feature information are fused and processed to generate defect feature enhanced image information, thereby significantly enhancing the minute defect features and greatly improving the detection performance, which is more suitable for surface defect detection of products such as semiconductor wafers and precision optical components, omits a large amount of sample training process, and solves the current problem of low real-time performance to a certain extent.
[0011] In a second aspect, an embodiment of the present application provides a system for enhancing detection of micro defects based on fusion of local relative gradients, the system comprising:
[0012] The module for acquiring information of the image to be detected is used to acquire information of the image to be detected;
[0013] A target area information determination module is used to determine target area information and multiple target area information according to the image information to be detected;
[0014] A local gradient feature information generating module: used to generate a plurality of local gradient feature information according to a plurality of target domain information and preset local relative gradient feature operator information;
[0015] Defect feature enhanced image information generation module: used to fuse the multiple local gradient feature information based on the target area information to generate defect feature enhanced image information.
[0016] In a third aspect, an embodiment of the present application provides a terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method of the first aspect described above when executing the computer program.
[0017] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method of the first aspect described above are implemented.
[0018] It can be understood that the beneficial effects of the second to fourth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.
[0020] Figure 1 It is a flow chart of a method for enhanced detection of micro defects provided in one embodiment of the present application;
[0021] Figure 2 is a flow chart of step S200 in the method for enhanced detection of micro defects provided in one embodiment of the present application;
[0022] Figure 3 is a flow chart of step S230 in the method for enhanced detection of micro defects provided in one embodiment of the present application;
[0023] Figure 4It is a schematic diagram of the process before step S235 in the method for enhanced detection of micro defects provided in one embodiment of the present application;
[0024] Figure 5 4 is a flow chart of step S400 in the method for enhanced detection of micro defects provided in one embodiment of the present application;
[0025] Figure 6 It is a module block diagram of a system for enhancing detection of micro defects provided by an embodiment of the present application;
[0026] Figure 7 It is a schematic diagram of a terminal device provided in one embodiment of the present application. DETAILED DESCRIPTION
[0027] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.
[0028] In the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.
[0029] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.
[0030] In order to illustrate the technical solution described in this application, a specific embodiment is provided below for illustration.
[0031] See also Figure 1 , Figure 1: is a flow chart of a method for enhancing detection of micro defects based on fusion of local relative gradients provided in an embodiment of the present application. In this embodiment, the execution subject of the method for enhancing detection of micro defects is a terminal device. It is understandable that the types of terminal devices include but are not limited to mobile phones, tablet computers, laptop computers, ultra-mobile personal computers (Ultra-Mobile Personal Computer, UMPC), netbooks, personal digital assistants (Personal Digital Assistant, PDA), etc., and the embodiment of the present application does not impose any restrictions on the specific types of terminal devices.
[0032] See also Figure 1 The method for enhanced detection of micro defects provided in the embodiment of the present application includes but is not limited to the following steps:
[0033] In S100, image information to be detected is obtained.
[0034] Specifically, the terminal device can obtain image information to be detected.
[0035] In S200, target area information and multiple target field information are determined according to the image information to be detected.
[0036] Specifically, after the terminal device obtains the image information to be detected, the terminal device can determine the target area information and multiple target field information based on the image information to be detected, wherein the target area information is used to describe the defect area in the image information to be detected, and the target field information is used to describe the area located around the defect area in the image information to be detected.
[0037] In some possible implementations, in order to determine effective target area information and multiple target area information, please refer to Figure 2 Step S200 includes but is not limited to the following steps:
[0038] In S210, target area information is generated according to the image information to be detected and a preset edge detection algorithm.
[0039] Specifically, the terminal device can perform edge extraction processing on the image information to be detected based on a preset edge detection algorithm to generate target area information, wherein the edge detection algorithm can be an edge detection algorithm based on a SobeL operator, an edge detection algorithm based on a Prewitt operator, or an edge detection algorithm based on a Laplacian operator.
[0040] In S220, the surrounding area information is determined according to the target area information and the preset first sampling range distance value information.
[0041] Specifically, after the terminal device generates the target area information, the terminal device can quickly determine the surrounding area information based on the target area information and the preset first sampling range distance value information, wherein the surrounding area information is used to describe the area centered on the target area information and enclosed by the first sampling range distance value information as a radius.
[0042] In S230, the surrounding area information is divided into adaptive thresholds to generate a plurality of target area information.
[0043] Specifically, after the terminal device determines the surrounding area information, the terminal device can perform adaptive threshold segmentation on the surrounding area information and utilize the adaptive threshold segmentation strategy to effectively generate multiple target area information.
[0044] In some possible implementations, in order to generate multiple target domain information to facilitate subsequent improvement of detection performance, please refer to Figure 3 Step S230 includes but is not limited to the following steps:
[0045] In S231, based on a preset basic division grid, the target region information is divided and processed to generate a plurality of target sub-region information, and the surrounding region information is divided and processed to generate a plurality of surrounding sub-region information.
[0046] Specifically, the terminal device can divide the target area information in equal proportion based on a preset basic division grid to generate multiple target sub-area information. At the same time, the terminal device can divide the surrounding area information based on the basic division grid to generate multiple surrounding sub-area information, wherein the size of the target sub-area information and the surrounding sub-area information are both equal to the size of the basic division grid.
[0047] In S232, for each piece of surrounding sub-region information: determine the related sub-region information according to the surrounding sub-region information.
[0048] Specifically, after the terminal device generates multiple target sub-region information and multiple surrounding sub-region information, the terminal device can perform this processing for each surrounding sub-region information: based on the boundary of the surrounding sub-region information, quickly determine the involved sub-region information, wherein the involved sub-region information is used to describe the target sub-region information adjacent to the surrounding sub-region information.
[0049] In S233, the surrounding area information is grayed to generate first maximum gray value information and first minimum gray value information, and the sub-area information is grayed to generate second maximum gray value information and second minimum gray value information.
[0050] Specifically, after the terminal device determines the sub-region information involved, the terminal device can grayscale the surrounding area information to generate a first maximum grayscale value information and a first minimum grayscale value information, and at the same time grayscale the sub-region information to generate a second maximum grayscale value information and a second minimum grayscale value information, wherein the first maximum grayscale value information is used to describe the maximum grayscale value corresponding to the surrounding area information after the grayscale processing, and the first minimum grayscale value information is used to describe the minimum grayscale value corresponding to the surrounding area information after the grayscale processing; the second maximum grayscale value information is used to describe the maximum grayscale value corresponding to the sub-region information after the grayscale processing, and the second minimum grayscale value information is used to describe the minimum grayscale value corresponding to the sub-region information after the grayscale processing.
[0051] In S234, first region mean information is generated according to the average value of the first maximum gray value information and the first minimum gray value information, and second region mean information is generated according to the average value of the second maximum gray value information and the second minimum gray value information.
[0052] Specifically, after the terminal device generates the first maximum grayscale value information, the first minimum grayscale value information, the second maximum grayscale value information, and the second minimum grayscale value information, the terminal device can generate the first region mean information according to the average value of the sum of the first maximum grayscale value information and the first minimum grayscale value information, and at the same time generate the second region mean information according to the average value of the sum of the second maximum grayscale value information and the second minimum grayscale value information.
[0053] In S235 , it is determined whether the difference between the first region mean information and the second region mean information is less than a preset dynamic threshold information.
[0054] Specifically, after the terminal device generates the first area mean information and the second area mean information, the terminal device may determine whether the difference between the first area mean information and the second area mean information is less than a preset dynamic threshold information.
[0055] In S236, if the difference between the first region mean information and the second region mean information is less than the dynamic threshold information, feature fusion processing is performed on the surrounding region information based on the sub-region information to generate target area information.
[0056] Specifically, if the difference between the mean information of the first area and the mean information of the second area is less than the dynamic threshold information, it indicates that there is a high correlation between the surrounding area information and the sub-area information involved. The terminal device can perform feature fusion processing on the surrounding area information based on the sub-area information involved, that is, fuse the features of the sub-area information involved into the surrounding area information to generate target area information, thereby realizing the fusion of multi-scale results, improving the suppression effect of background texture, and improving detection stability.
[0057] In some possible implementations, in order to further facilitate subsequent improvement of detection performance, after step S236, the method further includes but is not limited to the following steps:
[0058] In S237, based on the preset defect contour template information, the target area information is morphologically reconstructed to generate optimized area information.
[0059] Specifically, after the terminal device generates the target field information, the terminal device can perform morphological reconstruction processing on the target field information based on the preset defect contour template information, remove the part of the target field information that exceeds the defect contour template information, and generate optimized field information, wherein the defect contour template information is used to describe the outer contour corresponding to the tiny defects in the semiconductor wafer.
[0060] In some possible implementations, to introduce an adaptive threshold segmentation strategy, see Figure 4 Before step S235, the method further includes but is not limited to the following steps:
[0061] In S2351, a plurality of sampling sub-region information within a specified range is determined based on the surrounding sub-region information.
[0062] Specifically, the terminal device can quickly determine multiple sampling sub-area information within a specified range based on the surrounding sub-area information, wherein the specified range is used to describe an area centered on the surrounding sub-area information and enclosed by the second sampling range distance value information as a radius; the second sampling range distance value information is a specified multiple of the first sampling range distance value information, and the specified multiple is any natural number greater than 0 and less than 1. The specific value of the specified multiple is not limited. By way of example, the specified multiple can be 0.3, 0.5 or 0.7.
[0063] In S2352, dynamic threshold information is determined based on the grayscale average of multiple sampling sub-region information.
[0064] Specifically, after the terminal device determines multiple sampling sub-region information, the terminal device can calculate the sum of the grayscale values of the multiple sampling sub-region information, and calculate the grayscale average value based on the sum value, and determine the grayscale average value as the dynamic threshold information.
[0065] In S300, a plurality of local gradient feature information is generated according to a plurality of target area information and preset local relative gradient feature operator information.
[0066] Specifically, the terminal device can generate multiple local gradient feature information based on multiple target area information and preset local relative gradient feature operator information, so as to further integrate gradient features, capture subtle changes and local features in different regions, and enhance the overall feature extraction capability.
[0067] In some possible implementations, in order to meet the requirements for detecting micro defects, step S300 includes but is not limited to the following steps:
[0068] In S310, local gradient feature information is generated in sequence based on the local relative gradient feature operator information and the plurality of target domain information and target region information.
[0069] Specifically, the terminal device can generate local gradient feature information in sequence based on the local relative gradient feature operator information and multiple target domain information and target area information, wherein the local gradient feature information is used to describe the relative gradient between the target domain information and the target area information, and the local gradient feature information includes gradient direction information.
[0070] In S400, based on the target area information, a plurality of local gradient feature information is fused to generate defect feature enhanced image information.
[0071] Specifically, the terminal device can fuse multiple local gradient feature information based on the target area information to generate defect feature enhanced image information, so that the fused image is easier to identify defects, effectively reducing the probability of false detection and missed detection, and significantly improving the detection capability of tiny defects.
[0072] In some possible implementations, to further improve detection performance, see Figure 5 Step S400 includes but is not limited to the following steps:
[0073] In S410, gradient direction histogram information is constructed according to a plurality of local gradient feature information and target region information.
[0074] Specifically, the terminal device can construct gradient direction histogram information based on multiple local gradient feature information and target area information, wherein the gradient direction histogram information includes multiple unit blocks containing gradient directions.
[0075] In S420, dominant direction feature information is determined according to the gradient direction histogram information.
[0076] Specifically, after the terminal device constructs the gradient direction histogram information, the terminal device can determine the dominant direction feature information according to the gradient direction histogram information, wherein the dominant direction feature information is used to describe the direction corresponding to the unit blocks with the same gradient direction and the largest number.
[0077] In S430, based on the dominant direction feature information, a plurality of local gradient feature information is fused to generate defect feature enhanced image information.
[0078] Specifically, after the terminal device determines the dominant direction feature information, the terminal device can fuse multiple local gradient feature information based on the dominant direction feature information to generate defect feature enhanced image information, thereby greatly improving the detection performance, improving calculation efficiency, and improving practicality. Smile defect detection can be achieved without a large number of sample training, and the overall parameter adjustment process is relatively simple, more convenient to deploy, and better portability.
[0079] It should be noted that after the terminal device generates defect feature enhanced image information, the terminal device can also use an adaptive filter, such as an adaptive high-pass filter, to enhance the local contrast of the image, and then use a mean filter to implement mean filtering type smoothing of the image to reduce random noise, and then take the median of the neighborhood pixels to further suppress the noise to implement median filtering, and finally use a Gaussian filter to implement Gaussian filtering type smoothing of the image, which not only further removes the noise in the image, but also retains various details in the image.
[0080] The implementation principle of the method for enhanced detection of tiny defects based on fusion of local relative gradients in the embodiment of the present application is as follows: the terminal device may first obtain the image information to be detected, and then determine the target area information and multiple target field information based on the image information to be detected, and then generate multiple local gradient feature information based on the multiple target field information and the preset local relative gradient feature operator information, and finally, based on the target area information, fuse the multiple local gradient feature information to generate defect feature enhanced image information, thereby significantly enhancing the features of tiny defects and improving the detection performance, so that the defect detection rate below 5 microns reaches 96%, the false alarm rate is less than 0.01%, and the positioning accuracy is better than 1 micron, and the algorithm is more robust, which is particularly suitable for surface defect detection of products such as semiconductor wafers and precision optical components.
[0081] It should be noted that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0082] The embodiment of the present application also provides a system for enhancing the detection of micro defects based on the fusion of local relative gradients. For ease of explanation, only the parts related to the present application are shown, such as Figure 6 As shown, the system 60 includes:
[0083] The image information acquisition module 61 is used to acquire the image information to be detected;
[0084] The target area information determination module 62 is used to determine the target area information and multiple target area information according to the image information to be detected;
[0085] Local gradient feature information generating module 63: used to generate multiple local gradient feature information according to multiple target area information and preset local relative gradient feature operator information;
[0086] The defect feature enhanced image information generating module 64 is used to fuse a plurality of local gradient feature information based on the target area information to generate defect feature enhanced image information.
[0087] Optionally, the target domain information determination module 62 includes:
[0088] Target area information generation submodule: used to generate target area information according to the image information to be detected and the preset edge detection algorithm;
[0089] The peripheral area information determination submodule is used to determine the peripheral area information according to the target area information and the preset first sampling range distance value information, wherein the peripheral area information is used to describe the area surrounded by the target area information as the center and the first sampling range distance value information as the radius;
[0090] Target area information generation submodule: used to perform adaptive threshold division on surrounding area information and generate multiple target area information.
[0091] Optionally, the target domain information generation submodule includes:
[0092] A target sub-region information generating unit is used to divide the grid based on a preset basis, divide the target region information, generate a plurality of target sub-region information, and divide the surrounding region information to generate a plurality of surrounding sub-region information;
[0093] The involved sub-region information determining unit is used to determine the involved sub-region information according to the surrounding sub-region information for each piece of surrounding sub-region information, wherein the involved sub-region information is used to describe the target sub-region information adjacent to the surrounding sub-region information;
[0094] A minimum grayscale value information generating unit: used to perform grayscale processing on the surrounding area information to generate first maximum grayscale value information and first minimum grayscale value information, and to perform grayscale processing on the sub-area information to generate second maximum grayscale value information and second minimum grayscale value information;
[0095] A regional mean information generating unit: used to generate first regional mean information according to the average value of the first maximum gray value information and the first minimum gray value information, and to generate second regional mean information according to the average value of the second maximum gray value information and the second minimum gray value information;
[0096] A dynamic threshold information determination unit: used to determine whether the difference between the first area average information and the second area average information is less than a preset dynamic threshold information;
[0097] Target area information generation unit: used to generate target area information by performing feature fusion processing on the surrounding area information based on the sub-area information if the difference between the first area mean information and the second area mean information is less than the dynamic threshold information.
[0098] Optionally, the system 60 further includes:
[0099] Optimized domain information generation module: used to perform morphological reconstruction processing on the target domain information based on the preset defect contour template information to generate optimized domain information.
[0100] Optionally, the system 60 further includes:
[0101] A sampling sub-region information determination module: used to determine multiple sampling sub-region information within a specified range according to the surrounding sub-region information, wherein the specified range is used to describe an area centered on the surrounding sub-region information and surrounded by the second sampling range distance value information as a radius, and the second sampling range distance value information is a specified multiple of the first sampling range distance value information, and the specified multiple is any natural number greater than 0 and less than 1;
[0102] Dynamic threshold information determination module: used to determine dynamic threshold information according to the grayscale average value of multiple sampling sub-region information.
[0103] Optionally, the local gradient feature information generating module 63 includes:
[0104] A local gradient feature information generation submodule: used to sequentially generate local gradient feature information based on local relative gradient feature operator information and multiple target domain information and target region information, wherein the local gradient feature information is used to describe the relative gradient between the target domain information and the target region information, and the local gradient feature information includes gradient direction information;
[0105] Accordingly, the defect feature enhanced image information generating module 64 includes:
[0106] A gradient direction histogram information construction submodule: used to construct gradient direction histogram information according to a plurality of local gradient feature information and target area information, wherein the gradient direction histogram information includes a plurality of unit blocks containing gradient directions;
[0107] The dominant direction feature information determination submodule is used to determine the dominant direction feature information according to the gradient direction histogram information, wherein the dominant direction feature information is used to describe the direction corresponding to the unit blocks with the same gradient direction and the largest number;
[0108] Defect feature enhanced image information generation submodule: used to fuse multiple local gradient feature information based on dominant direction feature information to generate defect feature enhanced image information.
[0109] It should be noted that the information interaction, execution process and other contents between the above-mentioned modules are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.
[0110] The present application also provides a terminal device, such as Figure 7 As shown, the terminal device 70 of this embodiment includes: a processor 71, a memory 72, and a computer program 73 stored in the memory 72 and executable on the processor 71. When the processor 71 executes the computer program 73, the steps in the embodiment of the above-mentioned method for enhancing detection of micro defects are implemented, for example Figure 1 Steps S100 to S400 shown; or, when the processor 71 executes the computer program 73, the functions of each module in the above device are realized, for example Figure 6 The functions of modules 61 to 64 are shown.
[0111] The terminal device 70 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server, and the terminal device 70 includes but is not limited to a processor 71 and a memory 72. Those skilled in the art will appreciate that Figure 7 It is only an example of the terminal device 70 and does not constitute a limitation of the terminal device 70. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the terminal device 70 may also include input and output devices, network access devices, buses, etc.
[0112] Among them, the processor 71 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.; the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.
[0113] The memory 72 may be an internal storage unit of the terminal device 70, such as a hard disk or memory of the terminal device 70, or the memory 72 may be an external storage device of the terminal device 70, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the terminal device 70; further, the memory 72 may also include both an internal storage unit and an external storage device of the terminal device 70, and the memory 72 may also store a computer program 73 and other programs and data required by the terminal device 70, and the memory 72 may also be used to temporarily store data that has been output or is to be output.
[0114] One embodiment of the present application also provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the steps of each of the above method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc.; the computer-readable medium may include: any entity or device that can carry computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.
[0115] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, all equivalent changes made according to the methods, principles, and structures of the present application should be included in the protection scope of the present application.
Claims
1. A method for enhanced detection of small defects based on fusion of local relative gradients, characterized in that: The method comprises: Obtaining image information to be detected; Determining target area information and multiple target area information according to the image information to be detected; Generate multiple local gradient feature information according to the multiple target area information and the preset local relative gradient feature operator information; Based on the target area information, a plurality of local gradient feature information are fused to generate defect feature enhanced image information.
2. The method according to claim 1, characterized in that The step of determining target area information and multiple target field information according to the image information to be detected includes: Generate target area information according to the image information to be detected and a preset edge detection algorithm; Determine the surrounding area information according to the target area information and the preset first sampling range distance value information, wherein the surrounding area information is used to describe the area centered on the target area information and surrounded by the first sampling range distance value information as a radius; Adaptively divide the surrounding area information into thresholds to generate multiple target area information.
3. The method according to claim 2, characterized in that The adaptive threshold division of the surrounding area information to generate multiple target area information includes: Based on a preset basic division grid, the target area information is divided and processed to generate a plurality of target sub-area information, and the surrounding area information is divided and processed to generate a plurality of surrounding sub-area information; For each of the peripheral sub-region information: determining the involved sub-region information according to the peripheral sub-region information, wherein the involved sub-region information is used to describe the target sub-region information adjacent to the peripheral sub-region information; Performing grayscale processing on the peripheral area information to generate first maximum grayscale value information and first minimum grayscale value information, and performing grayscale processing on the sub-area information to generate second maximum grayscale value information and second minimum grayscale value information; Generate first region mean information according to an average value of the first maximum gray value information and the first minimum gray value information, and generate second region mean information according to an average value of the second maximum gray value information and the second minimum gray value information; Determine whether a difference between the first region mean information and the second region mean information is less than a preset dynamic threshold information; If the difference between the first region mean information and the second region mean information is smaller than the dynamic threshold information, the surrounding region information is subjected to feature fusion processing based on the involved sub-region information to generate target region information.
4. The method according to claim 3, characterized in that If the difference between the first region mean information and the second region mean information is less than the dynamic threshold information, then based on the involved sub-region information, feature fusion processing is performed on the surrounding region information to generate target region information, the method further includes: Based on the preset defect contour template information, the target domain information is morphologically reconstructed to generate optimized domain information.
5. The method according to claim 3, characterized in that: Before determining whether the difference between the first region mean information and the second region mean information is less than a preset dynamic threshold information, the method further includes: Determine, according to the peripheral sub-region information, a plurality of sampling sub-region information within a specified range, wherein the specified range is used to describe a region centered on the peripheral sub-region information and surrounded by the second sampling range distance value information as a radius, the second sampling range distance value information is a specified multiple of the first sampling range distance value information, and the specified multiple is any natural number greater than 0 and less than 1; Dynamic threshold information is determined according to the grayscale average of the plurality of sampling sub-region information.
6. The method according to claim 2, characterized in that The generating of a plurality of local gradient feature information according to the plurality of target domain information and the preset local relative gradient feature operator information comprises: Based on the local relative gradient feature operator information, local gradient feature information is sequentially generated according to the plurality of target domain information and target region information, wherein the local gradient feature information is used to describe the relative gradient between the target domain information and the target region information, and the local gradient feature information includes gradient direction information; Accordingly, based on the target area information, a plurality of local gradient feature information are fused to generate defect feature enhanced image information, including: Constructing gradient direction histogram information according to multiple local gradient feature information and the target area information, wherein the gradient direction histogram information includes multiple unit blocks including gradient directions; Determine dominant direction feature information according to the gradient direction histogram information, wherein the dominant direction feature information is used to describe the direction corresponding to the unit blocks with the same gradient direction and the largest number; Based on the dominant direction feature information, a plurality of local gradient feature information are fused to generate defect feature enhanced image information.
7. A micro defect enhancement detection system based on fusion of local relative gradients, characterized in that: The system comprises: The module for acquiring information of the image to be detected is used to acquire information of the image to be detected; A target area information determination module is used to determine target area information and multiple target area information according to the image information to be detected; A local gradient feature information generating module: used to generate a plurality of local gradient feature information according to a plurality of target domain information and preset local relative gradient feature operator information; Defect feature enhanced image information generation module: used to fuse the multiple local gradient feature information based on the target area information to generate defect feature enhanced image information.
8. The system according to claim 7, characterized in that The target domain information determination module includes: The target area information generating submodule is used to generate the target area information according to the image information to be detected and a preset edge detection algorithm; A peripheral area information determination submodule: used to determine the peripheral area information according to the target area information and the preset first sampling range distance value information, wherein the peripheral area information is used to describe the area centered on the target area information and surrounded by the first sampling range distance value information as a radius; Target area information generation submodule: used to perform adaptive threshold division on the surrounding area information to generate multiple target area information.
9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.