CN117649451B - Visual positioning method for positioning edge finder wafer - Google Patents

Visual positioning method for positioning edge finder wafer Download PDF

Info

Publication number
CN117649451B
CN117649451B CN202410123196.7A CN202410123196A CN117649451B CN 117649451 B CN117649451 B CN 117649451B CN 202410123196 A CN202410123196 A CN 202410123196A CN 117649451 B CN117649451 B CN 117649451B
Authority
CN
China
Prior art keywords
edge
image
gray
wafer
gray level
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202410123196.7A
Other languages
Chinese (zh)
Other versions
CN117649451A (en
Inventor
李玉廷
李恒飞
曹雄新
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Huxitech Technology Co ltd
Original Assignee
Shenzhen Huxitech Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Huxitech Technology Co ltd filed Critical Shenzhen Huxitech Technology Co ltd
Priority to CN202410123196.7A priority Critical patent/CN117649451B/en
Publication of CN117649451A publication Critical patent/CN117649451A/en
Application granted granted Critical
Publication of CN117649451B publication Critical patent/CN117649451B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G06T7/77—Determining position or orientation of objects or cameras using statistical methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T1/00—General purpose image data processing
    • G06T1/0014—Image feed-back for automatic industrial control, e.g. robot with camera
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration using local operators
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/64—Analysis of geometric attributes of convexity or concavity
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20024—Filtering details
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30148—Semiconductor; IC; Wafer

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Geometry (AREA)
  • Robotics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Image Analysis (AREA)

Abstract

The invention relates to the technical field of wafer positioning, in particular to a visual positioning method for positioning a wafer of an edge finder, which comprises the steps of acquiring a gray image of a wafer image, and determining a noise intensity characteristic value of the gray image according to the position distribution of each edge pixel point of the gray image; acquiring a gray scale run matrix of a gray scale image, and further determining texture information density parameters of the gray scale image; determining the size of a search window according to the position distribution of the wafer notch pixel points and the edge pixel points in the gray level image, and the noise intensity characteristic value and the texture information density parameter of the gray level image; and carrying out non-local mean filtering on the gray level image according to the size of the search window, and carrying out wafer positioning according to the gray level image after filtering. According to the invention, the size of the search window in the process of carrying out non-local mean value filtering on the gray level image of the wafer image is determined in a self-adaptive manner, so that the filtering effect is effectively ensured, and the positioning accuracy of the wafer is improved.

Description

Visual positioning method for positioning edge finder wafer
Technical Field
The invention relates to the technical field of wafer positioning, in particular to a visual positioning method for positioning a wafer of an edge finder.
Background
An edge finder (Aligner) is an apparatus used in a wafer manufacturing process, and fig. 1 shows a schematic structural diagram of the edge finder, which is mainly used for positioning and aligning a wafer, and mainly used for finding the edge and the center of the wafer, so as to ensure that each wafer process step can be accurately aligned and covered to a target area in a subsequent processing and manufacturing process, thereby improving the quality and production efficiency of a product and reducing the production cost.
In the process of positioning the wafer, the edge finder samples the wafer for multiple times through three cameras, processes the images to acquire other position information such as the circle center and the edge of the wafer, and finally realizes the wafer positioning. Because of the inevitable image noise generated during image acquisition and transmission, denoising of the wafer image is often required before wafer positioning.
Because the edge information in the wafer image is needed in the wafer positioning process, and the common filter can generate a smoothing effect on the edge pixel points in the process of removing the noise of the wafer image, partial edge information loss can be caused, and therefore the edge protection filter is needed to be selected to denoise the wafer image. The non-local mean filtering can retain edge information in an image to a large extent while realizing image filtering, and is generally used for denoising a wafer image.
When non-local mean filtering is performed on a wafer image, important parameters such as search window size, neighborhood block size, similarity weight and the like need to be determined. The size of the search window determines the range considered when pixel denoising is performed, and the size of the neighborhood block determines local context information of the pixel point, so that in order to ensure the denoising effect, the texture characteristics of the image and the denoising needs need to be comprehensively considered to determine the proper size of the search window and the size of the neighborhood block. However, in the prior art, the fixed search window size and the neighborhood block size are usually determined empirically, at this time, because different wafer images are affected by different noise, and the wafer areas are usually provided with notches, the situations that the search window size and the neighborhood block size are not matched with the wafer images easily occur due to different notch types and sizes of different wafers, so that the denoising effect is not ideal, and the accuracy of wafer positioning is finally affected.
Disclosure of Invention
The invention aims to provide a visual positioning method for positioning a wafer of an edge finder, which is used for solving the problem of low accuracy of wafer positioning caused by poor denoising effect of a wafer image in the prior art.
In order to solve the technical problems, the invention provides a visual positioning method for positioning a wafer of an edge finder, which comprises the following steps:
Acquiring a gray level image of a wafer image, performing edge detection on the gray level image to obtain each edge pixel point, and determining a noise intensity characteristic value of the gray level image according to the position distribution of the edge pixel points;
acquiring a gray scale run matrix of the gray scale image, and determining texture information density parameters of the gray scale image according to the distribution condition of each element value in the gray scale run matrix and the noise intensity characteristic value of the gray scale image;
Determining a wafer notch pixel point in the gray level image, and determining a search window size in the process of carrying out non-local mean filtering on the gray level image according to the position distribution of the wafer notch pixel point and the edge pixel point, the noise intensity characteristic value and the texture information density parameter of the gray level image;
and carrying out non-local mean filtering on the gray level image according to the size of a search window in the process of carrying out non-local mean filtering on the gray level image, and carrying out wafer positioning according to the gray level image after filtering.
Further, determining a noise intensity characteristic value of the gray image includes:
According to the position distribution of the edge pixel points, carrying out linear detection on the edge pixel points to obtain each edge line segment, and determining the total number of the edge pixel points contained in each edge line segment;
determining the included angles between the edge line segments and the set direction, counting the included angles, and determining the occurrence frequency of each type of included angle;
Determining edge pixel points which are not positioned on the edge line segment in all the edge pixel points as nonlinear edge points, detecting connected domains of the nonlinear edge points, obtaining edge connected domains where each nonlinear edge point is positioned, and determining the curvature of the edge connected domains;
and determining the noise intensity characteristic value of the gray image according to the total number of edge pixel points contained in each edge line segment, the occurrence frequency of each type of included angle and the curvature of the edge connected domain where each nonlinear edge point is positioned.
Further, determining a noise intensity characteristic value of the gray image, wherein the corresponding calculation formula is as follows:
；
Wherein, A noise intensity characteristic value representing the gray image; /(I)Representing the/>, in the included angles of all edge line segments and the set direction, in the gray level imageThe frequency of occurrence of seed angles; /(I)The number of kinds of included angles which appear in included angles between all edge line segments and a set direction in the gray level image is represented; /(I)Representing the/>, in the included angles of all edge line segments and the set direction, in the gray level imageThe average total number of edge pixel points contained in the edge line segment corresponding to the included angle; Representing the total number of edge pixel points contained in all edge line segments in the gray image; /(I) Representing the base of the logarithmic function; /(I)Representing the/>, in the gray scale imageCurvature of each nonlinear edge point on the edge connected domain where the nonlinear edge point is located; Representing the average value of the curvature of all nonlinear edge points in the gray level image on the edge connected domain where the nonlinear edge points are positioned; /(I) Representing the total number of all non-linear edge points in the gray scale image; /(I)Representing a hyperbolic tangent function.
Further, determining texture information density parameters of the gray level image, wherein the corresponding calculation formula is as follows:
；
；
Wherein, A texture information density parameter representing the gray scale image; /(I)Representing the first of the gray scale run matricesThe/>, in the row corresponding to the gray levelColumn element values; /(I)Representing the total number of gray levels in the gray scale run matrix; /(I)A noise intensity characteristic value representing the gray image; /(I)Representing a hyperbolic tangent function; /(I)Representing the/>, in the gray run matrixThe gray scales belong to probability parameters of the wafer chips; /(I)Representing the number of columns of the gray run matrix; /(I)Representing the/>, in the gray run matrixThe/>, in the row corresponding to the gray levelA maximum difference value between a column element value and its adjacent two column element values; /(I)Representing the/>, in the gray run matrixThe/>, in the row corresponding to the gray levelA derivative corresponding to the column element value; /(I)Representing the/>, in the gray run matrixDerivative means of element values of each column in a row corresponding to each gray level; /(I)Representing the normalization function.
Further, determining a search window size in the process of non-local mean filtering the gray scale image includes:
determining an edge line segment containing the wafer slot pixel points, and determining the edge line segment containing the wafer slot pixel points as a target edge line segment;
And determining the size of a search window in the process of carrying out non-local mean filtering on the gray level image according to the total number of the pixel points of the wafer slot opening in the gray level image, the difference of the included angles of each target edge line segment and the set direction, the length of the edge line segment, the noise intensity characteristic value and the texture information density parameter of the gray level image.
Further, determining the size of a search window in the process of carrying out non-local mean filtering on the gray level image, wherein the corresponding calculation formula is as follows:
；
Wherein, Representing the size of a search window in the process of carrying out non-local mean filtering on the gray level image; /(I)Representing the total number of wafer notch pixel points in the gray scale image; /(I)Representing the first pixel point containing the wafer slot in the gray level imageAn included angle between the line segment of the item mark edge and the set direction; /(I)Representing the average value of included angles between all target edge line segments containing wafer slot pixel points and a set direction in the gray level image; /(I)Representing the total number of all target edge line segments containing wafer notch pixel points in the gray level image; /(I)A noise intensity characteristic value representing the gray image; /(I)A texture information density parameter representing the gray scale image; /(I)Representing the maximum value of the lengths of all the edge line segments in the gray scale image.
Further, before the non-local mean filtering is performed on the gray scale image, the method further includes:
Determining the size of a neighborhood block in the process of carrying out non-local mean filtering on the gray image according to the noise intensity characteristic value of the gray image and the size of the search window, wherein the noise intensity characteristic value and the size of the neighborhood block form a negative correlation, and the size of the search window and the size of the neighborhood block form a positive correlation;
and carrying out non-local mean filtering on the gray level image according to the size of the search window and the size of the neighborhood block in the non-local mean filtering process of the gray level image.
Further, determining the size of a neighborhood block in the process of carrying out non-local mean filtering on the gray level image, wherein the corresponding calculation formula is as follows:
；
Wherein, Representing the size of a neighborhood block in the process of carrying out non-local mean filtering on the gray level image; /(I)A noise intensity characteristic value representing the gray image; /(I)Representing natural constants; /(I)Representing the search window size during non-local mean filtering of the gray scale image.
Further, performing edge detection on the gray level image to obtain each edge pixel point, including:
Performing edge detection on the gray level image by adopting a Canny edge detection method to obtain each edge pixel point;
And eliminating discrete edge pixel points in the edge pixel points, and taking the edge pixel points after eliminating the rest edge pixel points as final edge pixel points.
Further, if the eight neighboring pixel points of the edge pixel point do not include other edge pixel points, the edge pixel point is determined to be a discrete edge pixel point.
The invention has the following beneficial effects: in order to facilitate subsequent positioning of the wafer, a grayscale image of the wafer image is first acquired. Considering that the wafer positioning needs to use the edge information in the image, the noise influence degree of the edge needs to be analyzed, so that the proper search window size is convenient to determine later, and the complete edge information can be reserved in the denoised image. And carrying out edge detection on the gray level image to obtain each edge pixel point. Under the condition of not being influenced by noise, the similar edge pixel points have consistent distribution characteristics, and when the degree of influence by noise is larger, the similar edge pixel points are less obvious in consistent distribution characteristics, so that according to the position distribution of the edge pixel points, the consistent distribution condition of the edge pixel points is analyzed, and the noise intensity characteristic value of the gray image is determined. In order to ensure the denoising effect of the whole wafer image, the texture information of the whole wafer image needs to be considered, and for the image with larger texture information density, a larger search window needs to be selected to capture the texture information in the image. Because the chips on the wafer image are orderly arranged, the texture information of the chips can be embodied through the gray scale run matrix, so that the gray scale run matrix of the gray scale image is obtained, and according to the distribution characteristics of the chips in the wafer image, the distribution condition of each element value in the gray scale run matrix and the noise intensity characteristic value of the gray scale image are combined, the texture information density of the wafer image is extracted, and the texture information density parameter of the gray scale image is obtained. In addition, different types of notches may appear in the wafer edge portion in the wafer image, and the notch area has different search window size requirements for non-local mean filtering due to noise. Therefore, the pixel points of the wafer slot opening in the gray level image are identified, the feature of the wafer slot opening affected by noise is identified according to the position distribution of the pixel points of the wafer slot opening and the edge pixel points, and meanwhile, the size of a search window in the process of carrying out non-local mean value filtering on the gray level image is adaptively determined by combining the noise intensity feature value and the texture information density parameter of the gray level image, so that the effective filtering of the gray level image is realized, and the accurate positioning of the wafer is finally realized. According to the invention, the reasonable search window size when the non-local mean filtering is carried out on the image can be adaptively determined by comprehensively considering the degree of influence of noise on the edge information of the wafer image, the texture characteristics of the whole wafer image and the notch characteristics of the wafer, so that the effect of filtering the wafer image is improved, and the positioning accuracy of the wafer is finally improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic diagram of a prior art edge finder;
FIG. 2 is a flow chart of a visual positioning method for edge finder wafer positioning according to an embodiment of the present invention;
FIG. 3 is a grayscale image of a wafer image with flat grooves according to an embodiment of the present invention;
Fig. 4 is a gray scale image of a wafer image with V-grooves according to an embodiment of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the present invention to achieve the preset purpose, the following detailed description is given below of the specific implementation, structure, features and effects of the technical solution according to the present invention with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. In addition, all parameters or indices in the formulas referred to herein are values after normalization that eliminate the dimensional effects.
An embodiment of a visual positioning method for positioning an edge finder wafer:
in order to solve the technical problem of inaccurate wafer positioning caused by poor denoising effect of a wafer image in the prior art, the embodiment provides a visual positioning method for positioning a wafer by an edge finder, and a flow chart corresponding to the method is shown in fig. 2, and the method comprises the following steps:
Step S1: and acquiring a gray level image of the wafer image, performing edge detection on the gray level image to obtain each edge pixel point, and determining a noise intensity characteristic value of the gray level image according to the position distribution of the edge pixel points.
When the edge finder locates the wafer through a visual method, the three cameras are used for acquiring images of the wafer for multiple times, so that corresponding wafer images are obtained. In order to facilitate the subsequent reduction of the calculation cost for denoising the wafer image, the acquired wafer image is subjected to graying treatment, so that a corresponding gray image is obtained.
In the process of filtering and denoising the gray image by adopting a non-local mean value filtering algorithm, a search window of each pixel point in the gray image is required to be determined, and the filtering result is calculated by combining the similarity degree of the neighborhood blocks of each pixel point in the search window and then weighted average. The search window is used for searching the range of the similar blocks, and the neighborhood blocks are basic units used for calculating the similarity degree. In order to ensure the filtering effect, the intensity of the gray image affected by noise needs to be analyzed, so as to adaptively determine the appropriate size of the search window and the size of the neighborhood block.
Considering that chips which are orderly arranged on a wafer exist, image information represented by the chips can provide more reference information when judging noise intensity, and considering that edge information in an image is required to be used for positioning the wafer, the degree of influence of noise on the edge needs to be analyzed. Based on the image, the edge detection method is used for carrying out edge detection on the gray image, and each edge pixel point in the gray image is obtained. Because the acquired wafer image has noise points, and some noise points appear as single points, the discrete edge pixel points in all edge pixel points are deleted, wherein the discrete edge pixel points refer to edge point pixel points without other edge pixel points in eight adjacent areas, and the rest edge pixel points after deletion are used as final all edge pixel points on the gray level image. And according to the final positions of all the edge pixel points on the gray level image, carrying out straight line detection on the edge pixel points by a Hough straight line detection method to obtain all the edge straight lines, and determining a line segment formed by two edge pixel points positioned at two ends of the edge straight lines as an edge line segment corresponding to the edge straight lines to obtain all the edge line segments.
Considering that the arrangement of chips on a wafer is relatively regular, the detected edge line segments are mainly obtained by chip arrangement gaps, when the noise is affected, noise pixel points can be connected with original chip gap line pixel points and then detected as straight lines, but the straight line direction of the noise pixel points is different from other gap line directions, and the distribution of the noise pixel points is not regular, so that uncertainty exists in the extending direction of each edge line segment. In addition, in the normal gray level image, the nonlinear edge points are mainly circular edge portions of the wafer, and the curvatures of the circular points are the same, but after being affected by noise, the curvatures of the nonlinear edge points of the wafer in the connected domain have differences, and the greater the differences, the greater the degree of interference by noise.
Based on the characteristics, the included angles between each detected edge line segment and the set direction in the gray level image are determined, and the included angles are used for representing the extending direction of the edge line segment. In the present embodiment, the setting direction is a horizontal direction in the grayscale image, and as another embodiment, the setting direction may be a vertical direction in the grayscale image, which is not limited herein. And meanwhile, determining edge pixel points which are not positioned in each edge line segment in the final edge pixel points on the determined gray level image, and taking the edge pixel points as non-linear edge points. And detecting the connected domain of the nonlinear edge points, so as to obtain an edge connected domain where each nonlinear edge point is located, wherein the edge connected domain is a circular arc section of the wafer edge formed by a plurality of nonlinear edge points, and determining the curvature of each nonlinear edge point on the edge connected domain where the nonlinear edge point is located. Since the specific implementation process of determining the edge connected domain in which each nonlinear edge point is located and the curvature of each nonlinear edge point on the edge connected domain in which each nonlinear edge point is located is a technology well known to those skilled in the art, the description thereof will not be repeated here.
On the basis, according to the included angle between each edge line segment and the set direction in the gray image and combining the curvature of each nonlinear edge point on the edge connected domain where the nonlinear edge point is positioned, determining the noise intensity characteristic value of the gray image, wherein the corresponding calculation formula is as follows:
；
Wherein, A noise intensity characteristic value representing a gray image; /(I)Representing the/>, in the included angles of all edge line segments and the set direction in the gray level imageThe frequency of occurrence of seed angles; /(I)The number of kinds of included angles appearing in included angles between all edge line segments and a set direction in the gray level image is represented; /(I)Representing the/>, in the included angles of all edge line segments and the set direction in the gray level imageThe average total number of edge pixel points contained in the edge line segment corresponding to the included angle; /(I)Representing the total number of edge pixel points contained in all edge line segments in the gray image; /(I)Representing the base of the logarithmic function, the present embodiment sets/>；Representing the/>, in a gray scale imageCurvature of each nonlinear edge point on the edge connected domain where the nonlinear edge point is located; /(I)Representing the average value of the curvatures of all nonlinear edge points in the gray level image on the edge connected domain where the nonlinear edge points are positioned; /(I)Representing the total number of all non-linear edge points in the gray scale image; /(I)Representing a hyperbolic tangent function.
In the above-described calculation formula of the noise intensity characteristic value,The information entropy reflects the uncertainty degree of the included angle between the edge line segment and the set direction, and when the value is larger, the intensity of noise interference in the current gray image is larger. Meanwhile, considering that the chip gap line is relatively longer, so the total number of edge pixel points contained in the corresponding edge line segments is larger, the probability that the detected edge line segments with fewer edge pixel points are interfered by noise is larger, and the proportion of the edge line segments with fewer edge pixel points is higher when the uncertainty degree is determined, so the corresponding edge line segments pass/>Calculating the length parameter of the edge line segment corresponding to each type of included angle, adjusting the information entropy by taking the length parameter as a weight, and normalizing the adjusted information entropy through an entropy base line. /(I)The larger the value is, the larger the curvature variance of all the nonlinear edge points on the edge point connected domain where the nonlinear edge points are located is, and the larger the curvature phase difference of the nonlinear edge points on the edge point connected domain where the nonlinear edge points are located is, the larger the noise intensity in the current gray level image is.
Step S2: and acquiring a gray scale run matrix of the gray scale image, and determining texture information density parameters of the gray scale image according to the distribution condition of each element value in the gray scale run matrix and the noise intensity characteristic value of the gray scale image.
When the non-local mean filtering algorithm is adopted to filter and denoise the gray level image, the size of the search window and the size of the neighborhood block are selected and also need to be determined by combining the texture information of the whole image. For images with a higher density of texture information, a larger search window needs to be selected to capture texture information in the image. For the wafer image collected by the edge finder, the texture information mainly comprises texture information of chip arrangement on the wafer and the edges of the wafer, wherein the edges of wafers with different notches are different.
As the chips on the wafer image are orderly arranged, the texture information of the chips can be embodied by a gray scale run matrix. Therefore, the gray scale run matrix of the gray scale image is determined according to the gray scale value of each pixel point in the gray scale image. In the present embodiment, the gradation values 0 to 255 are set to 16 gradation, and a gradation run matrix of the gradation image is constructed with 0 degree as a run angle. Of course, since the wafer image has symmetry, as other embodiments, 90 degrees may be used as the run angle, which is not limited herein. Because the specific implementation process of constructing the gray scale run matrix of the gray scale image belongs to the prior art, the description is omitted here.
Analyzing the distribution condition of each element value in the gray scale run matrix to extract the texture density information of the gray scale image, thereby determining the texture information density parameter of the gray scale image, wherein the corresponding calculation formula is as follows:
；
；
Wherein, A texture information density parameter representing a gray scale image; /(I)Representing the/>, in a gray scale run-length matrixThe/>, in the row corresponding to the gray levelColumn element values; /(I)Representing the total number of gray levels in the gray level run matrix, in this embodiment/>；/>A noise intensity characteristic value representing a gray image; /(I)Representing a hyperbolic tangent function; /(I)Representing the/>, in a gray scale run-length matrixThe gray scales belong to probability parameters of the wafer chips; /(I)Representing the number of columns of the gray scale run matrix, i.e. the number of run values; representing the/>, in a gray scale run-length matrix The/>, in the row corresponding to the gray levelThe maximum difference between a column element value and its adjacent two columns, i.e. calculate the/>The absolute value of the difference between the column element value and each adjacent column element value is used as the maximum difference value; /(I)Representing the/>, in a gray scale run-length matrixThe first row corresponding to the gray levelThe derivative corresponding to the column element value is determined by deriving the row element value corresponding to each gray level in the gray level run matrix; /(I)Representing the/>, in a gray scale run-length matrixDerivative means of element values of each column in a row corresponding to each gray level; /(I)Representing the normalization function.
In the above calculation formula of the texture information density parameter, since a large number of chips are regularly arranged in the wafer image, the texture information density in the wafer image is mainly determined by the arrangement of the chips, and the gray scale of the chips should occupy a larger proportion in the calculation process. When the run direction corresponding to the set run angle is adopted to scratch on the rectangular chip, the sizes of the run values scratched from different positions of the rectangular chip are different, a maximum value exists in the run of each rectangle, the number of runs smaller than the maximum value in the rectangle is 1 to 2, and the number depends on the relation between the run direction and the rectangular angle. At this time, the change rate difference of the elements in each column in the gray scale of the gray scale run matrix is large, and part of the run value elements are 2 times of the other part of the run value elements.The derivative variance is used for representing the variation rate difference of the current line element on each column, meanwhile, when the fact that the running value is affected by noise is considered, the element value of each column does not strictly represent the double proportion difference, so that the accuracy of the derivative variance is interfered, the difference of the actual variation value needs to be combined to adjust the calculation process of the derivative variance, and particularly the part with larger variation of the actual value can be amplified. Thus pass/>And adjusting the derivative variance calculation process, and normalizing the adjusted derivative variance to obtain probability parameters of the current gray scale belonging to the wafer chip. Since the texture information density in the wafer image is mainly determined by the arrangement of the chips, the texture information density of the wafer image can be represented by the average value of the run value elements in the run matrix in the gray scale run matrix, and for the gray scale corresponding to the rectangular chip, the corresponding duty ratio should be larger when calculating the texture information density, so that the probability parameter/>, of the gray scale to the wafer chip is utilizedAdjusting the mean value calculation process of the run value elements of the corresponding rows of different gray scales, and when the probability parameter/>The larger the value of (2), the higher the ratio weight corresponding to the run value element of the corresponding row.
In the above calculation formula of the texture information density parameter, the shorter the run in the gray scale run matrix is, the larger the run is affected by noise, and the specific gravity of the low run value should be reduced in the process of calculating the texture information density. For example, originally, the run value of a rectangular chip at a set angle is y, but due to noise influence, y is divided into r and t (r+t=y), the run values of the two divided parts are lower, meanwhile, in the gray scale run matrix, the element added 1 part of the original run value y is added to the run values of r and t respectively, so that the element mean value in the gray scale run matrix is increased compared with the original one, and in order to compensate the increased part, the proportion occupied by the low run value should be reduced in the process of calculating the texture density. Therefore, the higher the noise level, the higher the probability of noise disturbance, and the lower the weight, the higher the run value, and the lower the weight, i.e., the number of columns in the gray run matrix, the higher the noise level, and the higher the gray level change in the image, which cannot reflect the real texture informationThe smaller the noise intensity characteristic value N is, the larger the corresponding specific gravity/>The smaller the value of (c).
Step S3: determining a wafer notch pixel point in the gray level image, and determining the size of a search window in the process of carrying out non-local mean filtering on the gray level image according to the position distribution of the wafer notch pixel point and the edge pixel point, the noise intensity characteristic value and the texture information density parameter of the gray level image.
Different types of notches may occur in the wafer edge portion, such as Flat grooves (Flat) and V grooves (Notch), in the wafer image, the wafer image with the Flat grooves is shown in fig. 3, and the wafer image with the V grooves is shown in fig. 4. When a non-local mean filtering algorithm is adopted to filter and denoise a gray level image, for a wafer image of a V groove, a certain blurring condition can occur at a notch, and blurring is caused by the fact that most of the searching window of background pixel points in the notch is wafer pixel points; but the blurring at the notch caused by the too large search window is not obvious for a wafer image of a flat slot. Moreover, the slot sizes and angles of slots of the same type may be different, and the size requirements for the search window may be different when they are disturbed by different levels of noise.
Since different types of notches may appear in the wafer edge portion in the wafer image and the notch area has different requirements on the size of the search window when the non-local mean filtering is performed due to the influence of noise, in order to facilitate the subsequent adaptive determination of the size of the search window when the non-local mean filtering algorithm is used for filtering and denoising the gray image, the type of the wafer notch in the wafer image needs to be adaptively identified. Considering that the type of the wafer notch can be represented by the number of the wafer notch pixel points under the condition of reducing the calculation cost, the number of the wafer notch pixel points is smaller in the wafer image with the V-shaped groove, and the number of the wafer notch pixel points is larger in the wafer image with the flat groove. Meanwhile, as the pixel points of the wafer notch belong to the edge pixel points, under the condition of noise interference, the angle dispersion condition of the edge line segments corresponding to the notch pixel points is required to be analyzed.
Based on the analysis, the wafer notch pixel points in the gray level image are identified by utilizing a Harris corner detection algorithm, template matching and the like, so that each wafer notch pixel point in the gray level image is obtained. And determining edge line segments containing the wafer slot pixel points according to the positions of the wafer slot pixel points, and taking the edge line segments as target edge line segments. According to the number of pixel points at the wafer slot opening and the difference of the included angles between the target edge line segments and the set direction, and combining the noise intensity characteristic value and the texture information density parameter of the gray level image and the maximum length of the edge line segments determined in the gray level image, determining the size of a search window in the process of carrying out non-local mean filtering on the gray level image, wherein the corresponding calculation formula is as follows:
；
Wherein, Representing the size of a search window in the process of carrying out non-local mean filtering on the gray level image; /(I)Representing the total number of wafer notch pixel points in the gray scale image; /(I)The/>, which represents the pixel point of the wafer slot included in the gray level imageAn included angle between the line segment of the item mark edge and the set direction; /(I)Representing the average value of included angles between all target edge line segments containing wafer slot pixel points and a set direction in the gray level image; /(I)Representing the total number of all target edge line segments containing the wafer notch pixel points in the gray level image; /(I)A noise intensity characteristic value representing a gray image; /(I)A texture information density parameter representing a gray scale image; /(I)Representing the maximum of the lengths of all edge line segments in the gray scale image.
In the above calculation formula of the search window size, the maximum length of the edge line segment in the gray image is selected as a basis to determine the search window size in the process of non-local mean filtering the gray image, so that the search window can contain more complete texture information to a certain extent.The characteristics of the wafer notch in the gray level image are reflected, when the total number of the pixel points of the wafer notch in the gray level image is smaller, the wafer notch in the gray level image is more likely to be a V-notch, and when the difference between the included angle between the target edge line segment containing the pixel point of the wafer notch in the gray level image and the set direction is larger, the situation that the wafer notch is interfered by noise is more serious is indicated, then when the non-local mean filtering algorithm is adopted to filter and denoise the gray level image, in order to avoid the blurring phenomenon at the notch, the adopted search window size should be smaller, otherwise, the search window size can be properly increased. Simultaneously, the noise intensity characteristic value/>, which is a parameter of the gray image, is combinedAnd texture information Density parameter/>When noise intensity eigenvalue/>When the noise intensity in the current gray level image is larger, the larger the noise intensity is, the more texture information of the local area where the pixel point is located can be acquired, namely, a plurality of different chip areas interfered by noise can be acquired, the neighborhood block comparison of the subsequent pixel point is more reference, and the size of the search window is increased. When texture information density parameter/>When the texture density of the current gray level image is larger, the corresponding chip area is smaller, the image noise is higher, and the search window size is required to be increased in order to obtain more comparison information in the neighborhood block comparison process.
After determining the size of the search window in the process of non-local mean filtering the gray image, in order to enable the neighborhood block to better represent the local features of the pixel points, for a larger search window, a larger neighborhood block size can be selected to contain more local features. Meanwhile, when the size of the neighborhood block is determined, the noise intensity of the gray image is considered, when the noise intensity is larger, the probability of internally containing noise pixels is increased because the oversized neighborhood block possibly contains picture information of other chips, the reliability of mean square error in neighborhood block comparison is reduced, and therefore the determined neighborhood block size is smaller.
In this embodiment, the calculation formula corresponding to the neighborhood block size in the process of determining the non-local mean value filtering of the gray image is:
；
Wherein, Representing the size of a neighborhood block in the process of carrying out non-local mean filtering on the gray level image; /(I)A noise intensity characteristic value representing a gray image; /(I)Representing natural constants; /(I)Representing the search window size during non-local mean filtering of the gray scale image.
It should be appreciated that the size of the search window size and the noise intensity of the gray scale image in the non-local mean filtering process of the gray scale image are referred to above to adaptively determine a more suitable neighborhood block size, thereby further improving the image filtering effect. Of course, as other embodiments, the neighborhood block size in the non-local mean filtering process of the gray scale image may also be determined by determining the neighborhood block size in the non-local mean filtering algorithm of the prior art.
Step S4: and carrying out non-local mean filtering on the gray level image according to the size of a search window and the size of a neighborhood block in the non-local mean filtering process of the gray level image, and carrying out wafer positioning according to the filtered gray level image.
After the search window size and the neighborhood block size in the process of carrying out non-local mean filtering on the gray image are determined in the mode, the non-local mean filtering algorithm is utilized to carry out non-local mean filtering on the gray image based on the search window size and the neighborhood block size, so that the gray image after filtering is obtained. Because the specific process of non-local mean filtering the gray image by using the non-local mean filtering algorithm belongs to the prior art, the description is omitted here.
And carrying out wafer positioning based on the filtered gray level image. When the wafer positioning is performed, the wafer position can be identified and marked through secondary development of a machine vision library (such as OpenCV, HALCON) and combination with the existing nonlinear wafer notch fitting algorithm, so that the intelligent wafer positioning is realized. For example, qu Dongsheng et al, in volume 3, 7 th, of nanotechnology and precision engineering, publication 12-1351/O3, provide a wafer prealignment method based on a high-precision micrometer that can achieve wafer positioning. The embodiment of the invention does not limit the specific implementation means for carrying out wafer positioning based on the gray level image after filtering.
According to the visual positioning method for the edge finder wafer positioning, the degree that the edge information of the wafer image is influenced by noise, the texture characteristics of the whole wafer image and the characteristics of the wafer notch are comprehensively considered by acquiring the wafer image acquired by the edge finder, and the reasonable search window size when the image is subjected to non-local mean filtering can be adaptively determined, so that the effect of filtering the wafer image is improved, and the accuracy of wafer positioning is finally improved.
An embodiment of an image processing method for locating a wafer of an edge finder:
In the process of positioning the wafer, the edge finder samples the wafer for multiple times through three cameras, processes the images to acquire other position information such as the circle center and the edge of the wafer, and finally realizes the wafer positioning. Because of the inevitable image noise generated during image acquisition and transmission, denoising of the wafer image is often required before wafer positioning.
When non-local mean filtering is performed on a wafer image, important parameters such as search window size, neighborhood block size, similarity weight and the like need to be determined. The size of the search window determines the range considered when pixel denoising is performed, and the size of the neighborhood block determines local context information of the pixel point, so that in order to ensure the denoising effect, the texture characteristics of the image and the denoising needs need to be comprehensively considered to determine the proper size of the search window and the size of the neighborhood block. However, in the prior art, the fixed search window size and the neighborhood block size are usually determined empirically, at this time, because different wafer images are affected by different noise, and the wafer areas are usually provided with notches, the different types and sizes of the notches of different wafers are different, the situation that the search window size and the neighborhood block size are not matched with the wafer images easily occurs, and thus the denoising effect is not ideal.
Aiming at the technical problem that the denoising effect is poor due to the fact that the size of the search window is not adapted, the non-local mean value filtering is performed on the wafer image, the embodiment provides an image processing method for locating the wafer of the edge finder, which comprises the following steps:
Acquiring a gray level image of a wafer image, performing edge detection on the gray level image to obtain each edge pixel point, and determining a noise intensity characteristic value of the gray level image according to the position distribution of the edge pixel points;
acquiring a gray scale run matrix of the gray scale image, and determining texture information density parameters of the gray scale image according to the distribution condition of each element value in the gray scale run matrix and the noise intensity characteristic value of the gray scale image;
Determining a wafer notch pixel point in the gray level image, and determining a search window size in the process of carrying out non-local mean filtering on the gray level image according to the position distribution of the wafer notch pixel point and the edge pixel point, the noise intensity characteristic value and the texture information density parameter of the gray level image;
And according to the size of a search window in the process of carrying out non-local mean filtering on the gray level image, carrying out non-local mean filtering on the gray level image to obtain a filtered gray level image.
Because the implementation process of each step in the image processing method for locating the edge finder wafer is described in detail in the embodiment of the visual locating method for locating the edge finder wafer, the description is omitted here.
According to the image processing method for locating the edge finder wafer, the degree that the edge information of the wafer image is affected by noise, the texture characteristics of the whole wafer image and the notch characteristics of the wafer are comprehensively considered, so that the reasonable search window size when the image is subjected to non-local mean filtering can be adaptively determined, and the effect of filtering the wafer image is improved.
It should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; although the application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application, and are intended to be included in the scope of the present application.

Claims (5)

1. The visual positioning method for positioning the edge finder wafer is characterized by comprising the following steps of:
Acquiring a gray level image of a wafer image, performing edge detection on the gray level image to obtain each edge pixel point, and determining a noise intensity characteristic value of the gray level image according to the position distribution of the edge pixel points;
acquiring a gray scale run matrix of the gray scale image, and determining texture information density parameters of the gray scale image according to the distribution condition of each element value in the gray scale run matrix and the noise intensity characteristic value of the gray scale image;
Determining a wafer notch pixel point in the gray level image, and determining a search window size in the process of carrying out non-local mean filtering on the gray level image according to the position distribution of the wafer notch pixel point and the edge pixel point, the noise intensity characteristic value and the texture information density parameter of the gray level image;
according to the size of a search window in the process of carrying out non-local mean filtering on the gray level image, and carrying out wafer positioning according to the gray level image after filtering;
Determining a noise intensity characteristic value of the gray image comprises:
According to the position distribution of the edge pixel points, carrying out linear detection on the edge pixel points to obtain each edge line segment, and determining the total number of the edge pixel points contained in each edge line segment;
determining the included angles between the edge line segments and the set direction, counting the included angles, and determining the occurrence frequency of each type of included angle;
Determining edge pixel points which are not positioned on the edge line segment in all the edge pixel points as nonlinear edge points, detecting connected domains of the nonlinear edge points, obtaining edge connected domains where each nonlinear edge point is positioned, and determining the curvature of the edge connected domains;
determining a noise intensity characteristic value of the gray image according to the total number of edge pixel points contained in each edge line segment, the occurrence frequency of each type of included angle and the curvature of an edge connected domain where each nonlinear edge point is located;
determining a noise intensity characteristic value of the gray image, wherein a corresponding calculation formula is as follows:
；
Wherein, A noise intensity characteristic value representing the gray image; /(I)Representing the/>, in the included angles of all edge line segments and the set direction, in the gray level imageThe frequency of occurrence of seed angles; /(I)The number of kinds of included angles which appear in included angles between all edge line segments and a set direction in the gray level image is represented; /(I)Representing the/>, in the included angles of all edge line segments and the set direction, in the gray level imageThe average total number of edge pixel points contained in the edge line segment corresponding to the included angle; /(I)Representing the total number of edge pixel points contained in all edge line segments in the gray image; /(I)Representing the base of the logarithmic function; /(I)Representing the/>, in the gray scale imageCurvature of each nonlinear edge point on the edge connected domain where the nonlinear edge point is located; /(I)Representing the average value of the curvature of all nonlinear edge points in the gray level image on the edge connected domain where the nonlinear edge points are positioned; /(I)Representing the total number of all non-linear edge points in the gray scale image; /(I)Representing a hyperbolic tangent function;
determining texture information density parameters of the gray level image, wherein the corresponding calculation formula is as follows:
；
；
Wherein, A texture information density parameter representing the gray scale image; /(I)Representing the/>, in the gray run matrixThe/>, in the row corresponding to the gray levelColumn element values; /(I)Representing the total number of gray levels in the gray scale run matrix; /(I)Representing the/>, in the gray run matrixThe gray scales belong to probability parameters of the wafer chips; /(I)Representing the number of columns of the gray run matrix; Representing the/>, in the gray run matrix The/>, in the row corresponding to the gray levelA maximum difference value between a column element value and its adjacent two column element values; /(I)Representing the/>, in the gray run matrixThe/>, in the row corresponding to the gray levelA derivative corresponding to the column element value; /(I)Representing the/>, in the gray run matrixDerivative means of element values of each column in a row corresponding to each gray level; /(I)Representing a normalization function;
Determining a search window size in a non-local mean filtering process of the gray scale image comprises:
determining an edge line segment containing the wafer slot pixel points, and determining the edge line segment containing the wafer slot pixel points as a target edge line segment;
Determining the size of a search window in the process of carrying out non-local mean filtering on the gray level image according to the total number of pixel points at the wafer slot opening in the gray level image, the difference of included angles between each target edge line segment and a set direction, the length of the edge line segment, the noise intensity characteristic value and the texture information density parameter of the gray level image;
determining the size of a search window in the process of carrying out non-local mean filtering on the gray level image, wherein the corresponding calculation formula is as follows:
；
Wherein, Representing the size of a search window in the process of carrying out non-local mean filtering on the gray level image; /(I)Representing the total number of wafer notch pixel points in the gray scale image; /(I)Representing the/>, including the wafer notch pixel points, in the gray scale imageAn included angle between the line segment of the item mark edge and the set direction; /(I)Representing the average value of included angles between all target edge line segments containing wafer slot pixel points and a set direction in the gray level image; /(I)Representing the total number of all target edge line segments containing wafer notch pixel points in the gray level image; /(I)Representing the maximum value of the lengths of all the edge line segments in the gray scale image.
2. The method for visual positioning of an edge finder wafer of claim 1, further comprising, prior to non-local mean filtering the gray scale image:
Determining the size of a neighborhood block in the process of carrying out non-local mean filtering on the gray image according to the noise intensity characteristic value of the gray image and the size of the search window, wherein the noise intensity characteristic value and the size of the neighborhood block form a negative correlation, and the size of the search window and the size of the neighborhood block form a positive correlation;
and carrying out non-local mean filtering on the gray level image according to the size of the search window and the size of the neighborhood block in the non-local mean filtering process of the gray level image.
3. The visual positioning method for positioning an edge finder wafer according to claim 2, wherein the neighborhood block size in the process of performing non-local mean filtering on the gray level image is determined according to the following calculation formula:
；
Wherein, Representing the size of a neighborhood block in the process of carrying out non-local mean filtering on the gray level image; /(I)A noise intensity characteristic value representing the gray image; /(I)Representing natural constants; /(I)Representing the search window size during non-local mean filtering of the gray scale image.
4. The visual positioning method for positioning an edge finder wafer according to claim 1, wherein performing edge detection on the gray level image to obtain each edge pixel point comprises:
Performing edge detection on the gray level image by adopting a Canny edge detection method to obtain each edge pixel point;
And eliminating discrete edge pixel points in the edge pixel points, and taking the edge pixel points after eliminating the rest edge pixel points as final edge pixel points.
5. The method for locating a wafer by an edge finder of claim 4, wherein if the eight neighboring pixels of the edge pixel do not include any other edge pixel, determining the edge pixel as a discrete edge pixel.
CN202410123196.7A 2024-01-30 2024-01-30 Visual positioning method for positioning edge finder wafer Active CN117649451B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202410123196.7A CN117649451B (en) 2024-01-30 2024-01-30 Visual positioning method for positioning edge finder wafer

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202410123196.7A CN117649451B (en) 2024-01-30 2024-01-30 Visual positioning method for positioning edge finder wafer

Publications (2)

Publication Number Publication Date
CN117649451A CN117649451A (en) 2024-03-05
CN117649451B true CN117649451B (en) 2024-04-30

Family

ID=90046445

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202410123196.7A Active CN117649451B (en) 2024-01-30 2024-01-30 Visual positioning method for positioning edge finder wafer

Country Status (1)

Country Link
CN (1) CN117649451B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117934459B (en) * 2024-03-21 2024-06-28 锦诚实业科技(深圳)有限公司 Food material encapsulation opening defect detection method

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108171688A (en) * 2017-12-19 2018-06-15 浙江大学 A kind of wafer surface defects detection method based on Gabor characteristic Yu random dimensionality reduction
CN115100199A (en) * 2022-08-24 2022-09-23 宁波鑫芯微电子科技有限公司 Method for detecting wafer low-texture defects
CN116740058A (en) * 2023-08-11 2023-09-12 深圳市金胜电子科技有限公司 Quality detection method for solid state disk matched wafer
CN116993718A (en) * 2023-09-25 2023-11-03 深圳市东陆科技有限公司 TFT array substrate defect detection method based on machine vision
CN117115130A (en) * 2023-09-12 2023-11-24 长川科技(苏州)有限公司 Wafer edge defect detection method and device

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108171688A (en) * 2017-12-19 2018-06-15 浙江大学 A kind of wafer surface defects detection method based on Gabor characteristic Yu random dimensionality reduction
CN115100199A (en) * 2022-08-24 2022-09-23 宁波鑫芯微电子科技有限公司 Method for detecting wafer low-texture defects
CN116740058A (en) * 2023-08-11 2023-09-12 深圳市金胜电子科技有限公司 Quality detection method for solid state disk matched wafer
CN117115130A (en) * 2023-09-12 2023-11-24 长川科技(苏州)有限公司 Wafer edge defect detection method and device
CN116993718A (en) * 2023-09-25 2023-11-03 深圳市东陆科技有限公司 TFT array substrate defect detection method based on machine vision

Also Published As

Publication number Publication date
CN117649451A (en) 2024-03-05

Similar Documents

Publication Publication Date Title
CN109035276B (en) Image edge extraction method and device and automatic driving system
CN117649451B (en) Visual positioning method for positioning edge finder wafer
CN104574347B (en) Satellite in orbit image geometry positioning accuracy evaluation method based on multi- source Remote Sensing Data data
WO2021109697A1 (en) Character segmentation method and apparatus, and computer-readable storage medium
WO2023070593A1 (en) Line width measurement method and apparatus, computing processing device, computer program, and computer readable medium
CN115619793B (en) Power adapter appearance quality detection method based on computer vision
CN116977329B (en) Photovoltaic grid line detection method based on machine vision
CN115290663B (en) Mini LED wafer appearance defect detection method based on optical detection
CN109359604B (en) Method for identifying instrument under shadow interference facing inspection robot
CN116228768B (en) Method for detecting scratches on surface of electronic component
CN115661143B (en) Rapid detection system for MiniLED wafer defects
CN115018835B (en) Automobile starter gear detection method
CN116228780A (en) Silicon wafer defect detection method and system based on computer vision
CN113673515A (en) Computer vision target detection algorithm
CN117557820A (en) Quantum dot optical film damage detection method and system based on machine vision
CN113899349B (en) Sea wave parameter detection method, equipment and storage medium
CN116883408A (en) Integrating instrument shell defect detection method based on artificial intelligence
CN117314901B (en) Scale-adaptive chip detection neural network system
CN114970590A (en) Bar code detection method
CN116862907B (en) Motor accessory quality detection method based on image features
CN111553927B (en) Checkerboard corner detection method, detection system, computer device and storage medium
CN111161264A (en) Method for segmenting TFT circuit image with defects
CN114742849B (en) Leveling instrument distance measuring method based on image enhancement
CN110349169B (en) Linear measuring method
CN113554695A (en) Intelligent part hole site identification and positioning method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant