CN115274486B - Semiconductor surface defect identification method - Google Patents
Semiconductor surface defect identification method Download PDFInfo
- Publication number
- CN115274486B CN115274486B CN202211177695.1A CN202211177695A CN115274486B CN 115274486 B CN115274486 B CN 115274486B CN 202211177695 A CN202211177695 A CN 202211177695A CN 115274486 B CN115274486 B CN 115274486B
- Authority
- CN
- China
- Prior art keywords
- point
- processed
- points
- obtaining
- window
- 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
Links
- 230000007547 defect Effects 0.000 title claims abstract description 40
- 239000004065 semiconductor Substances 0.000 title claims abstract description 31
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000001914 filtration Methods 0.000 claims abstract description 15
- 238000003708 edge detection Methods 0.000 claims abstract description 11
- 230000014759 maintenance of location Effects 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 5
- 230000000694 effects Effects 0.000 description 6
- 239000003086 colorant Substances 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 239000000463 material Substances 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 239000013078 crystal Substances 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000009499 grossing Methods 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000002950 deficient Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 241001085205 Prenanthella exigua Species 0.000 description 1
- 230000003213 activating effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- -1 gray Substances 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 239000012212 insulator Substances 0.000 description 1
- 229910052751 metal Inorganic materials 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 229910021421 monocrystalline silicon Inorganic materials 0.000 description 1
- 238000005498 polishing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 239000002210 silicon-based material Substances 0.000 description 1
- 229910052709 silver Inorganic materials 0.000 description 1
- 239000004332 silver Substances 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 235000012431 wafers Nutrition 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L22/00—Testing or measuring during manufacture or treatment; Reliability measurements, i.e. testing of parts without further processing to modify the parts as such; Structural arrangements therefor
- H01L22/10—Measuring as part of the manufacturing process
- H01L22/12—Measuring as part of the manufacturing process for structural parameters, e.g. thickness, line width, refractive index, temperature, warp, bond strength, defects, optical inspection, electrical measurement of structural dimensions, metallurgic measurement of diffusions
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L22/00—Testing or measuring during manufacture or treatment; Reliability measurements, i.e. testing of parts without further processing to modify the parts as such; Structural arrangements therefor
- H01L22/20—Sequence of activities consisting of a plurality of measurements, corrections, marking or sorting steps
- H01L22/24—Optical enhancement of defects or not directly visible states, e.g. selective electrolytic deposition, bubbles in liquids, light emission, colour change
Landscapes
- Engineering & Computer Science (AREA)
- Manufacturing & Machinery (AREA)
- Computer Hardware Design (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Power Engineering (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention relates to the technical field of data processing, in particular to a semiconductor surface defect identification method. The method comprises the steps of analyzing a surface image of a semiconductor refrigerating sheet, calculating the gray degree of the semiconductor refrigerating sheet and the possibility of being a straight line to obtain a reserved point, a heavy point and a target point, setting different Gaussian filter variances for the reserved point, the heavy point and the target point, obtaining the weight of each pixel point in a corresponding filter window based on the Gaussian filter variances, and carrying out Gaussian filtering on the surface image according to the weight and the Gaussian filter variances to obtain an optimized image; and performing edge detection on the optimized image based on a Sobel operator to obtain defect pixel points in the surface image, and obtaining scratch defects in the surface image according to the defect pixel points, so that the accuracy of defect detection is improved.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a semiconductor surface defect identification method.
Background
A semiconductor is a material with conductivity between metal and insulator, and in the actual manufacturing process of semiconductor materials, due to the influence of various environments, temperatures, impurities and pulling rates, the semiconductor materials have some defects, such as obvious scratches on wafers during grinding or polishing.
The conventional image processing technology is used for detecting scratch defects by using edge detection, but a monocrystalline silicon material presents a situation of single crystal grains, and the brightness change of the crystal grains brings huge interference to the edge detection technology, so that the brightness change between the single crystal grains is detected into a plurality of short curves, and an error detection result is caused; the main reason that the edge detection effect is not good is that the effect of gaussian filtering is unsatisfactory, and when filtering is performed on the whole image by using a gaussian kernel with the same size and variance, the smoothing effect of pixel points of scratch pixels in the semiconductor material is the same as that of pixel points in a scratch area, so that scratches are not outstanding finally, but are smoothed together with other pixels, and edge detail information is lost, so that the scratch result obtained by edge detection is not accurate enough.
Disclosure of Invention
In order to solve the above technical problem, an object of the present invention is to provide a semiconductor surface defect identifying method, including the steps of:
acquiring a surface image of a semiconductor refrigerating sheet, and acquiring a corresponding brightness image based on the surface image;
acquiring the gray degree of each corresponding pixel point according to the brightness value of each pixel point in the brightness image, taking the pixel points with the gray degree larger than zero as points to be processed, and taking the pixel points with the gray degree not larger than zero as retention points;
acquiring neighborhood to-be-processed points in eight neighborhoods of each to-be-processed point, and obtaining a first possibility based on coordinate positions of the neighborhood to-be-processed points and the to-be-processed points, wherein the first possibility is the possibility that the to-be-processed points and the neighborhood to-be-processed points are distributed on a straight line;
when the first possibility is larger than a preset threshold value, constructing a plurality of windows by taking the point to be processed corresponding to the first possibility as a central point, wherein the windows are different in size; acquiring a point to be processed in each window, and obtaining a second possibility based on the coordinate position of the point to be processed in each window and the central point; obtaining confidence degrees according to the second possibility corresponding to all windows, recording the center point of the window with the confidence degree larger than zero as a key point, and recording the center point of the window with the confidence degree not larger than zero as a target point;
setting Gaussian filter variances with different sizes for the retention points, the target points and the important points, acquiring the weight in a filter window corresponding to each pixel point based on the Gaussian filter variances, and performing Gaussian filtering on the surface image according to the weight and the Gaussian filter variances to obtain an optimized image;
performing edge detection on the optimized image by using a Sobel operator to obtain defect pixel points in the surface image, and obtaining scratch defects in the surface image according to the defect pixel points;
obtaining a second possibility based on the coordinate position of the point to be processed and the central point in each window, including:
the second likelihood is calculated as:
wherein,
representing a second possibility;
representing a coordinate position of the center point within a window;
representing the coordinate position of the ith point to be processed in the window;
representing the coordinate position of the (i + 1) th point to be processed in the window;
representing the tangent function.
Preferably, the step of obtaining the grayness degree of the corresponding pixel point according to the brightness value of each pixel point in the brightness image includes:
the grayness was calculated as:
wherein,
indicating the degree of grayness;
expressing the brightness value of the pixel point;
representing a minimum luminance value in the luminance image;
representing a maximum luminance value in the luminance image;
is an activation function.
Preferably, the step of obtaining the first possibility based on the coordinate positions of the neighborhood point to be processed and the point to be processed includes:
obtaining coordinate differences between the neighborhood points to be processed and the points to be processed, calculating a difference value based on the coordinate differences corresponding to each neighborhood point to be processed, and obtaining the first possibility according to the difference value.
Preferably, the step of obtaining the confidence level according to the second likelihood corresponding to all the windows includes:
obtaining the difference between all the second possibilities, and obtaining the confidence coefficient based on the difference and the size of the window.
Preferably, the step of obtaining the weight in the corresponding filtering window of each pixel point based on the gaussian filtering variance includes:
the weight is calculated as:
wherein,
indicating the position as
The weight value of (1);
representing a gaussian filter variance;
represents a natural constant;
indicating the circumferential ratio.
The invention has the following beneficial effects: according to the embodiment of the invention, the surface pixel points of the semiconductor refrigerating sheet are divided into different categories, different Gaussian filter variances are set for the pixel points of different categories in a self-adaptive manner, corresponding weights are obtained based on the different Gaussian filter variances, finally, gaussian filtering is carried out to obtain an optimized image, scratch defect detection is carried out according to the optimized image, and the reliability and the accuracy of detection are improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flowchart of a semiconductor surface defect identification method according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a location distribution in a gaussian kernel according to an embodiment of the present invention.
Detailed Description
To further illustrate the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description of a semiconductor surface defect identification method according to the present invention, its specific implementation, structure, features and effects, with reference to the accompanying drawings and preferred embodiments, is provided below. In the following description, the different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
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.
The method is suitable for detecting the scratch defect caused in the production and processing process of the semiconductor refrigerating sheet; the following describes a specific scheme of the semiconductor surface defect identification method provided by the present invention in detail with reference to the accompanying drawings.
Referring to fig. 1, a flow chart of a semiconductor surface defect identification method according to an embodiment of the present invention is shown, the method includes the following steps:
and S100, acquiring a surface image of the semiconductor chilling plate, and acquiring a corresponding brightness image based on the surface image.
Horizontally placing the semiconductor chilling plate on a conveyor belt, vertically and downwards acquiring an initial image of the surface of the semiconductor chilling plate by using a CCD (charge coupled device) camera, performing semantic segmentation on the initial image, and reserving the area of the semiconductor chilling plate to obtain a surface image; the semantic segmentation is the prior known technology and is not described in detail.
Furthermore, because the semiconductor refrigerating sheet is close to gray and the RGB image effect of the surface image is not obvious enough, the surface image is converted into an HSV image of an HSV space, and the analysis of the semiconductor refrigerating sheet is more suitable based on the brightness space of the HSV space, so that the brightness image corresponding to the semiconductor refrigerating sheet is obtained based on the HSV image.
Step S200, obtaining the gray level of the corresponding pixel points according to the brightness value of each pixel point in the brightness image, taking the pixel points with the gray level larger than zero as the points to be processed, and taking the pixel points with the gray level not larger than zero as the reserved points.
The pixel points close to gray black in the brightness image are the colors of the normal area on the surface of the semiconductor refrigerating sheet, and the difference between the scratch defect pixel points and the scratch defect pixel points is larger; however, other white normal pixel points exist on the surface of the semiconductor chilling plate, the colors of the white normal pixel points are very close to the colors of the pixel points for dividing the defects, and therefore large influence can be generated when scratch defect analysis is carried out subsequently. Since there may be a plurality of white adjacent colors such as gray, silver, bright white, and the like in the luminance image, in the embodiment of the present invention, an area in which the empirical value indicates white and its adjacent colors is set, and the empirical value interval is [96,100].
Obtaining a maximum brightness value and a minimum brightness value in the brightness image, obtaining a gray level of a corresponding pixel point based on the brightness value corresponding to each pixel point and the maximum brightness value and the minimum brightness value in the brightness image, wherein the gray level is calculated specifically as follows:
wherein,
indicating the degree of grayness;
expressing the brightness value of the pixel point;
representing a minimum luminance value in the luminance image;
representing a maximum luminance value in the luminance image;
and setting the gray degree of the gray and black pixel points as 0 for activating the function.
When the pixel point approaches to white, the gray degree corresponding to the pixel point is larger; conversely, when the pixel point is farther away from white, the gray level degree corresponding to the pixel point is smaller; in order to facilitate subsequent analysis and calculation, the gray level smaller than zero is set to be 0 in the embodiment of the invention, that is, the gray level areas of all the pixel points are larger than zero or equal to zero.
Marking the pixel points with the gray degree of zero as retention points, and subsequently setting the same Gaussian filter parameters for filtering; and marking the pixel points with the gray degree larger than zero as the points to be processed, and carrying out subsequent analysis on the points to be processed.
Step S300, neighborhood to-be-processed points in eight neighborhoods of each to-be-processed point are obtained, and a first possibility is obtained based on coordinate positions of the neighborhood to-be-processed points and the to-be-processed points, wherein the first possibility is the possibility that the to-be-processed points and the neighborhood to-be-processed points are distributed on a straight line.
Because the pixels with the gray degree larger than zero are all close to white, and all the pixels with the gray degree larger than zero not only comprise the pixels in the normal area of the semiconductor refrigerating sheet, but also comprise the scratch pixels, the to-be-processed points with the gray degree larger than zero are analyzed, and the to-be-processed points in the scratch area are often approximately in an edge style and form a straight line with the to-be-processed points in the neighborhood, so that whether the to-be-processed points and the to-be-processed points in the neighborhood form the straight line is judged.
Specifically, a sliding window with any size is set, the size of the sliding window is set to be 3 × 3, and with any point to be processed as a central point of the sliding window, a pixel point in the sliding window is a pixel point in an eight-neighborhood corresponding to the point to be processed, all points to be processed in the sliding window are selected and marked as neighborhood points to be processed, a first possibility is obtained through a coordinate position between the point to be processed in the center of the sliding window and the neighborhood points to be processed, and then the first possibility is calculated as:
wherein,
representing a first possibility under a 3 x 3 sliding window;
coordinates of a point to be processed representing the center of the sliding window;
to show the second in the sliding window
Coordinates of each neighborhood point to be processed;
indicating first in the sliding window
Coordinates of each neighborhood point to be processed;
representing a tangent function.
When the first possibility
The more the value of (1) approaches to 1, the higher the possibility of forming a straight line is, that is, the higher the possibility that the point to be processed belongs to the point of the scratch defect area is; and by analogy, acquiring corresponding first possibility when all points to be processed in the brightness image are the central points of the sliding window.
Step S400, when the first possibility is larger than a preset threshold value, a plurality of windows are constructed by taking the point to be processed corresponding to the first possibility as a central point, and the windows are different in size; acquiring a point to be processed in each window, and obtaining a second possibility based on the coordinate position of the point to be processed in each window and the central point; and obtaining confidence degrees according to the second possibility corresponding to all the windows, recording the central point of the window with the confidence degree larger than zero as a key point, and recording the central point of the window with the confidence degree not larger than zero as a target point.
The first possibility corresponding to each point to be processed is obtained in step S300, when the first possibility is greater than the preset threshold, the point may be a defective point, and the point to be processed corresponding to the point with the first possibility greater than the preset threshold is analyzed again.
Taking any one to-be-processed point with the first probability greater than 0.9 as an example, expanding the sliding window from 3 × 3 to 5 × 5, then expanding to 7 × 7, and finally expanding to 9 × 9, thereby obtaining three windows with the to-be-processed point as a center, then marking the to-be-processed point in each window, that is, marking a pixel point with the gray level greater than zero in each window, and calculating the second probabilities corresponding to the to-be-processed points under different windows based on the same method for calculating the first probabilities in step S300, that is, the second probabilities include: 5 x 5 windowed
7 x 7 windowed
And 9 x 9 windows
(ii) a Whether the probability that a straight line is formed between the point to be processed and the point to be processed in the neighborhood of the point to be processed is increased along with the expansion of the window is analyzed, and the confidence coefficient that all the points to be processed in the window form a straight line is obtained according to the second probability corresponding to each window:
wherein,
representing a confidence level;
the size of the window is indicated and,
=3,5,7,9；
to represent
*
A second possibility of a window;
is an activation function.
If the probability that the points to be processed in the window form a straight line is increased along with the increase of the window, the corresponding confidence coefficient is larger than zero; conversely, if the probability that the point to be processed in the window constitutes a straight line is decreasing as the window increases, the confidence result is set to 0. And marking the point to be processed with the confidence coefficient being greater than zero as a key point, and marking the point to be processed with the confidence coefficient being zero as a target point.
And S500, setting Gaussian filter variances with different sizes for the retention point, the target point and the important point, acquiring the weight in a corresponding filter window of each pixel point based on the Gaussian filter variances, and performing Gaussian filtering on the surface image according to the weight and the Gaussian filter variances to obtain an optimized image.
Specifically, the size of all gaussian sliding windows is fixed to be 5 × 5, the positions in the gaussian core are distributed, please refer to fig. 2, and different variances need to be set according to different types of pixel points; the size of the core is fixed,
the larger the value is, the more gradual the weight distribution is, so that the influence of each point value of the neighborhood on the output value is larger, and the final result causes the image to be more fuzzy; otherwise, the size of the core is fixed,
the smaller the value is, the more prominent the weight distribution is, so the smaller the influence of each point value in the neighborhood on the output value is, and the smaller the image change is.
The embodiment of the invention reserves the pointCorresponding gaussian filter variance setting
Since it does not interfere with edge detection by itself; setting the Gaussian filter variance corresponding to the target point to be
Because the scratch edge detection is greatly influenced, the maximum smoothing is performed; setting the Gaussian filter variance corresponding to the important point as
The degree of smoothing thereof is small.
Therefore, the weight of each pixel point in the 5 x 5 window is calculated according to the Gaussian filtering variance of the pixel points of different categories as follows:
wherein,
the display position is
The weight value of (1);
representing a gaussian filter variance;
represents a natural constant;
indicating the circumferential ratio.
As the size of the Gaussian kernel in the whole image is fixed to be 5 x 5, each point in the sliding window is calculated according to the type of the central pixel point which slides to a certain central pixel point every time and the variance of different types and substituted into the formulaCorresponding weight value
。
Based on the set Gaussian filtering variance and the weight calculation, a Gaussian kernel with the size of 5 × 5 is used for sliding from left to right and from top to bottom in sequence from the upper left corner of the surface image, filtering operation is carried out on each pixel point, and the image after filtering is recorded as an optimized image.
And S600, performing edge detection on the optimized image by using a Sobel operator to obtain defect pixel points in the surface image, and obtaining scratch defects in the surface image according to the defect pixel points.
Specifically, the gradient of each pixel point is calculated by using a Sobel operator
And after the gradients of all the pixel points are obtained, non-maximum suppression is carried out, the aim is to enable the fuzzy boundary to be clear, the operation is to reserve the maximum in the gradient direction of each pixel point, and the rest values are removed.
Then, an appropriate high-low threshold value is set according to an empirical value
The subsequent meeting will
The pixel points of (2) are not considered as any edge points, namely points of non-defect areas; will be provided with
The pixel points of (2) are used as weak edge points, namely points of possible defect areas
The pixel points are used as strong edge points, namely the pixel points are determined as defect area pixel points; marking all detected defective pixel points, and setting the rest pixel points to be black so as to obtain a scratch defect area in the surface image.
In summary, in the embodiment of the present invention, the surface image of the semiconductor chilling plate is collected, and the corresponding brightness image is obtained based on the surface image; acquiring the gray degree of each corresponding pixel point according to the brightness value of each pixel point in the brightness image, taking the pixel points with the gray degree larger than zero as to-be-processed points, and taking the pixel points with the gray degree not larger than zero as to-be-reserved points; acquiring neighborhood to-be-processed points in eight neighborhoods of each to-be-processed point, and obtaining a first possibility based on the coordinate positions of the neighborhood to-be-processed points and the to-be-processed points, wherein the first possibility is the possibility that the to-be-processed points and the neighborhood to-be-processed points are distributed on a straight line; when the first possibility is larger than a preset threshold value, constructing a plurality of windows by taking the point to be processed corresponding to the first possibility as a central point, wherein the windows are different in size; acquiring a point to be processed in each window, and obtaining a second possibility based on the coordinate position of the point to be processed in each window and the central point; obtaining confidence degrees according to the second possibility corresponding to all the windows, recording the center point of the window with the confidence degree larger than zero as a key point, and recording the center point of the window with the confidence degree not larger than zero as a target point; setting Gaussian filter variances with different sizes for the retention points, the target points and the important points, acquiring the weight in the corresponding filter window of each pixel point based on the Gaussian filter variances, and performing Gaussian filtering on the surface image according to the weight and the Gaussian filter variances to obtain an optimized image; and carrying out edge detection on the optimized image by utilizing a Sobel operator to obtain defect pixel points in the surface image, and obtaining scratch defects in the surface image according to the defect pixel points to improve the accuracy of scratch defect detection.
It should be noted that: the precedence order of the above embodiments of the present invention is only for description, and does not represent the merits of the embodiments. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that are within the spirit of the present invention are intended to be included therein.
Claims (5)
1. A method for identifying surface defects of a semiconductor, the method comprising the steps of:
acquiring a surface image of a semiconductor refrigerating sheet, and acquiring a corresponding brightness image based on the surface image;
acquiring the gray degree of each corresponding pixel point according to the brightness value of each pixel point in the brightness image, taking the pixel points with the gray degree larger than zero as points to be processed, and taking the pixel points with the gray degree not larger than zero as retention points;
acquiring neighborhood to-be-processed points in the eight neighborhoods of each to-be-processed point, and obtaining a first possibility based on the coordinate positions of the neighborhood to-be-processed points and the to-be-processed points, wherein the first possibility is the possibility that the to-be-processed points and the neighborhood to-be-processed points are distributed on a straight line;
when the first possibility is larger than a preset threshold value, constructing a plurality of windows by taking the point to be processed corresponding to the first possibility as a central point, wherein the windows are different in size; acquiring a point to be processed in each window, and obtaining a second possibility based on the coordinate position of the point to be processed in each window and the central point; obtaining confidence degrees according to the second possibility corresponding to all windows, recording the center point of the window with the confidence degree larger than zero as a key point, and recording the center point of the window with the confidence degree not larger than zero as a target point;
setting Gaussian filter variances with different sizes for the retention points, the target points and the important points, acquiring the weight in a filter window corresponding to each pixel point based on the Gaussian filter variances, and performing Gaussian filtering on the surface image according to the weight and the Gaussian filter variances to obtain an optimized image;
utilizing a Sobel operator to carry out edge detection on the optimized image so as to obtain defect pixel points in the surface image, and obtaining scratch defects in the surface image according to the defect pixel points;
the step of obtaining a second possibility based on the coordinate position of the point to be processed and the center point in each window includes:
the second likelihood is calculated as:
wherein,
representing a second possibility;
representing a coordinate position of the center point within a window;
representing the coordinate position of the ith point to be processed in the window;
representing the coordinate position of the (i + 1) th point to be processed in the window;
representing a tangent function.
2. The method according to claim 1, wherein the step of obtaining the grayness of each pixel according to the brightness value of each pixel in the brightness image comprises:
the grayness was calculated as:
wherein,
indicating the degree of grayness;
expressing the brightness value of the pixel point;
representing a minimum luminance value in the luminance image;
representing a maximum luminance value in the luminance image;
is an activation function.
3. The method of claim 1, wherein the step of obtaining the first likelihood based on the coordinate positions of the neighborhood point to be processed and the point to be processed comprises:
obtaining coordinate differences between the neighborhood to-be-processed points and the to-be-processed points, calculating a difference value based on the coordinate differences corresponding to each neighborhood to-be-processed point, and obtaining the first possibility according to the difference value.
4. The method of claim 1, wherein the step of obtaining confidence levels according to the second likelihood corresponding to all the windows comprises:
obtaining the difference between all the second possibilities, and obtaining the confidence coefficient based on the difference and the size of the window.
5. The method of claim 1, wherein the step of obtaining the weight of each pixel point in the corresponding filter window based on the gaussian filter variance comprises:
the weight is calculated as:
wherein,
the display position is
The weight value of (1);
representing a gaussian filter variance;
represents a natural constant;
indicating the circumferential ratio.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211177695.1A CN115274486B (en) | 2022-09-27 | 2022-09-27 | Semiconductor surface defect identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211177695.1A CN115274486B (en) | 2022-09-27 | 2022-09-27 | Semiconductor surface defect identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115274486A CN115274486A (en) | 2022-11-01 |
| CN115274486B true CN115274486B (en) | 2022-12-13 |
Family
ID=83756808
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211177695.1A Active CN115274486B (en) | 2022-09-27 | 2022-09-27 | Semiconductor surface defect identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115274486B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116630315B (en) * | 2023-07-24 | 2023-09-29 | 山东东阿亿福缘阿胶制品有限公司 | Intelligent beverage packaging defect detection method based on computer vision |
| CN117115165A (en) * | 2023-10-24 | 2023-11-24 | 长沙康乾电子科技有限公司 | Method and device for scratching processing surface of mask |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110286126A (en) * | 2019-06-17 | 2019-09-27 | 浙江大学 | A kind of wafer surface defects subregion area detecting method of view-based access control model image |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5433201B2 (en) * | 2008-10-23 | 2014-03-05 | 株式会社Ｓｕｍｃｏ | Wafer backside evaluation method |
| CN112991332A (en) * | 2021-04-21 | 2021-06-18 | 视睿（杭州）信息科技有限公司 | Polycrystalline defect detection method and device for wafer defects |
| CN114235758A (en) * | 2021-12-10 | 2022-03-25 | 苏州凌云视界智能设备有限责任公司 | Defect detection method, device, equipment and storage medium |
| CN115020267B (en) * | 2022-08-05 | 2022-10-11 | 启东旺晟电子科技有限公司 | Semiconductor surface defect detection method |
-
2022
- 2022-09-27 CN CN202211177695.1A patent/CN115274486B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110286126A (en) * | 2019-06-17 | 2019-09-27 | 浙江大学 | A kind of wafer surface defects subregion area detecting method of view-based access control model image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115274486A (en) | 2022-11-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115274486B (en) | Semiconductor surface defect identification method | |
| CN117351008B (en) | Smart phone panel surface defect detection method | |
| CN107451999B (en) | Foreign matter detection method and device based on image recognition | |
| CN116596936B (en) | Solar wing sailboard quality detection method based on image data | |
| CN113963042B (en) | Metal part defect degree evaluation method based on image processing | |
| CN111415363B (en) | Image edge identification method | |
| CN116740070B (en) | Plastic pipeline appearance defect detection method based on machine vision | |
| CN115020267B (en) | Semiconductor surface defect detection method | |
| CN115290663B (en) | Mini LED wafer appearance defect detection method based on optical detection | |
| CN115861320B (en) | Intelligent detection method for automobile part machining information | |
| CN115797342B (en) | Industrial control capacitive touch LCD display assembly defect detection method | |
| CN110648330B (en) | Defect detection method for camera glass | |
| CN109378279B (en) | Wafer detection method and wafer detection system | |
| CN115690105B (en) | Milling cutter scratch detection method based on computer vision | |
| CN115330767A (en) | Method for identifying production abnormity of corrosion foil | |
| CN116773548B (en) | Wafer surface defect detection method and system | |
| CN113298809B (en) | Composite material ultrasonic image defect detection method based on deep learning and superpixel segmentation | |
| CN115100191A (en) | Metal casting defect identification method based on industrial detection | |
| CN115272335A (en) | Metallurgical metal surface defect detection method based on significance detection | |
| CN117274113B (en) | Broken silicon wafer cleaning effect visual detection method based on image enhancement | |
| CN116883408B (en) | Integrating instrument shell defect detection method based on artificial intelligence | |
| CN117152165B (en) | Photosensitive chip defect detection method and device, storage medium and electronic equipment | |
| CN115511907B (en) | Scratch detection method for LED screen | |
| CN115953398A (en) | Defect identification method for strip steel surface | |
| CN114937006A (en) | Hammerhead refinement defect detection method based on image enhancement |
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 |