CN115731155A - Monocrystalline silicon subsurface crack damage detection method based on machine vision - Google Patents
Monocrystalline silicon subsurface crack damage detection method based on machine vision Download PDFInfo
- Publication number
- CN115731155A CN115731155A CN202111017834.XA CN202111017834A CN115731155A CN 115731155 A CN115731155 A CN 115731155A CN 202111017834 A CN202111017834 A CN 202111017834A CN 115731155 A CN115731155 A CN 115731155A
- Authority
- CN
- China
- Prior art keywords
- crack damage
- image
- detection
- crack
- scale
- 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.)
- Pending
Links
Images
Landscapes
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The monocrystalline silicon subsurface crack damage detection method based on the machine vision is characterized by comprising a crack damage image acquisition module, a multi-light source imaging module, a crack damage extraction module and a crack damage detection module; the crack damage image acquisition module adopts an industrial CCD to acquire target information to be detected; the multi-light source imaging module realizes the uniformity of image characteristics; the crack damage extraction module is used for determining the crack damage characteristic position and scale, comprises a crack damage characteristic position algorithm and a crack damage scale algorithm and is used for realizing the detection of crack damage characteristic points, the maximum scale and the minimum scale of crack damage; and the crack damage detection module realizes high-precision detection and generates a high-goodness-of-fit detection curve.
Description
Technical Field
The invention belongs to the technical field of monocrystalline silicon precision detection, and particularly discloses a method for realizing efficient and precise detection of monocrystalline silicon subsurface damage.
Technical Field
In actual use, monocrystalline silicon is widely applied to the field of integrated circuits due to the characteristics of good radiation resistance and high temperature resistance; with the gradual development of integrated circuits towards the goals of low cost and high intelligence, the width of a scribed line of a silicon wafer is required to be thinner and thinner, and the requirements on the quality and specification of monocrystalline silicon are continuously improved. The processing process of the monocrystalline silicon mainly comprises mechanical and chemical processing, such as cutting, grinding, polishing and the like, and the damage such as scratches, cracks, dislocation, stacking faults and the like can be inevitably generated on the surface of a silicon wafer in the processing processes of cutting, grinding and the like; the damage of the surface layer of the silicon wafer is an important index for evaluating the processing quality of the silicon wafer; therefore, the effective detection of the surface crack damage of the monocrystalline silicon is explored, and the application quality of the monocrystalline silicon can be improved.
In recent years, researchers at home and abroad have rapid research and development on the detection of the sub-surface crack damage of the monocrystalline silicon, and the detection technology relates to a plurality of related subjects, such as a plurality of detection technologies based on mechanics, acoustics, spectroscopy, thermal imagery and the like, and the detection technologies are widely applied due to simple principle and intuitive result; the detection methods commonly used at present mainly include a transmission electron microscope method, an X-ray diffraction emission method and an etching method. The method has the defects of complex sample preparation, poor detection precision and the like to different degrees.
The existing single crystal silicon damage detection patents can be found as follows: in a method for detecting the thickness of a scratch damage layer on the surface of monocrystalline silicon, which is mentioned in a patent (CN 104034296A), an atomic force microscope is adopted for scanning to obtain the three-dimensional appearance of the surface of a monocrystalline silicon sample, and the thickness of the scratch damage layer on the surface of the monocrystalline silicon is determined by comparing the sample preparation recess depths before and after etching; in a method for detecting mechanical damage on the surface of a monocrystalline silicon based on conductivity change, which is mentioned in the patent (CN 10833333390A), an atomic force microscope is adopted to scan a damaged area to be detected, so as to obtain a surface topography and a current distribution diagram, and then the scanning result is contrasted and analyzed to find the mechanical damage position on the surface of the monocrystalline silicon, so as to obtain a detection result;
the two patents have certain limitations in detection precision and detection result display; the method for detecting the surface damage of the single crystal silicon based on machine vision is not reported, and the detection precision and speed can achieve ideal effects.
At present, a paper for detecting the sub-surface crack damage of the monocrystalline silicon by adopting machine vision is not retrieved, and papers on the sub-surface damage of the monocrystalline silicon published by scholars at home and abroad mainly focus on the damage mechanism research, the damaged surface appearance research, the surface quality research and the establishment and prediction of an analytical model; the detection of the damage of the sub-surface of the monocrystalline silicon is often only used as one aspect of the above research to verify the accuracy of the research, and the aforementioned conventional detection method is mostly adopted, and the detection of the crack damage of the sub-surface of the monocrystalline silicon based on machine vision is rarely reported.
Disclosure of Invention
In order to solve the problems that damage is caused to the surface of the monocrystalline silicon by destructive detection, the conventional nondestructive detection is complex in sample preparation, high in cost, complex in detection process and low in detection efficiency, the invention designs the monocrystalline silicon subsurface crack damage detection method based on machine vision, solves the problems in the conventional detection method by utilizing the high-level imaging technology of the machine vision, and provides help for the high-speed and accurate detection of the monocrystalline silicon subsurface damage; firstly, a CCD image acquisition module is designed to acquire a damage image, then crack damage features are extracted by using the characteristic that the crack damage feature scale changes, the damage scale and the position are determined according to the extracted damage features, and the sub-surface crack damage detection of the monocrystalline silicon based on machine vision is realized.
The technical scheme and the detection steps of the invention are as follows:
the monocrystalline silicon subsurface crack damage detection method based on machine vision is characterized by comprising a crack damage image acquisition module, a multi-light-source imaging module, a crack damage extraction module and a crack damage detection module; the crack damage image acquisition module adopts an industrial CCD to acquire target information to be detected; the multi-light source imaging module realizes the uniformity of image characteristics; the crack damage extraction module is used for determining the crack damage characteristic position and scale, comprises a crack damage characteristic position algorithm and a crack damage scale algorithm and is used for realizing the detection of crack damage characteristic points, the maximum scale and the minimum scale of crack damage; the crack damage detection module realizes high-precision detection and generates a high-goodness-of-fit detection curve.
The method comprises the following steps:
step 1, crack damage image acquisition, wherein an industrial camera CCD is mainly used for converting photoelectric signals into ordered electric signals and acquiring target information to be detected; considering the size of the target to be detected and the requirements of items such as an imaging area, a depth of field, a working distance and the like, a telecentric lens of XF-5MDT05X65 is selected to be matched with an industrial camera to acquire the image of the target to be detected.
Step 2, designing a multi-light-source imaging scheme, wherein the requirement of a camera for acquiring an image on a light source is very strict, and when a monocrystalline silicon subsurface crack damage image is acquired, a multi-light-source lighting scheme is selected according to specific requirements so as to obtain the optimal lighting effect and obtain a high-quality image; considering that the detection target is the monocrystalline silicon subsurface crack, a forward illumination mode is selected, and an LTLNC100-W linear light source is adopted for lighting the monocrystalline silicon subsurface crack; for a detection target with a complex structure and a large target, the condition of uneven characteristics is easy to occur in the acquisition process, and in order to ensure that the characteristics of different areas in the acquired image are even, a plurality of light sources are adopted to respectively illuminate the different areas, so that the gray value uniformity level of the image is improved; the plurality of light sources illuminate the object primarily by means of a superposition of light fields.
And 3, extracting crack damage characteristics, wherein the process comprises the following substeps:
step 3.1, determining the position of the characteristic point of the crack damage image;
calculating partial derivatives of Gaussian kernels H (x, y, alpha) in the neighborhood of a certain pixel point (x, y) on the image, wherein the directions are e and s, and performing convolution on Robert gradients of the neighborhood pixels; the gray level covariance matrix in the feature operator is:
wherein
In the formula (I), the compound is shown in the specification,
and
representing the Robert gradient in both directions; in the process of extracting the initial selection point, a 5 multiplied by 5 window pair is usedDetecting each pixel in the image one by one, simultaneously performing convolution operation once, removing noise in the image, and preliminarily determining candidate points; the specific candidate point is determined to be related to the central pixel of the image, and the central pixel is used for determining the specific candidate point
The gray values in the upper, lower, left and right directions are used as targets, and the gray difference absolute value in each direction is calculated; the calculation formula is as follows:
in the formula (I), the compound is shown in the specification,
respectively representing the gray difference absolute values of the image center pixel in four directions; according to the detection requirement, a threshold value is manually set, and if the calculation result is greater than the set threshold value, the threshold value is calculated
And determining the initial feature point, otherwise, if the calculation result is smaller than the set threshold value, recalculating until the initial feature point is determined.
Step 3.2, extracting the crack damage characteristic scale;
for the extraction of the feature scale, the image entropy is used for description; the method comprises the following specific steps:
in the formula (I), the compound is shown in the specification,
the entropy of the image is represented by,
a function representing the density of the grey scale distribution,
the size of the gray scale is represented,
represents the radius of a neighborhood circular window with a characteristic point as the center, represents the statistical region of entropy,
representing a position of a characteristic point; the value when the local entropy obtains the extreme value is taken as a characteristic scale, and is described as follows:
obtained by the above formula
Is the characteristic dimension of the target image
(ii) a And after obtaining the characteristic scale and position of the crack damage, detecting the sub-surface crack damage of the monocrystalline silicon.
Step 4, crack damage detection;
the obtained image crack damage characteristics are shown as the mutation of gray values through computer observation, the characteristic extraction is mainly realized by using a Gaussian function, and in order to ensure the invariance of the characteristic scale in the detection process, a scale related item is added in front of the second derivative of the Gaussian function and is expressed as follows:
wherein the content of the first and second substances,
express normalizationA factor. During the inspection process, the crack damage is characterized by
Convolution processing when
The method can generate positive response to the crack, and therefore, the method can be used for crack detection; and detecting the crack damage by using the opposite number, and taking the maximum value of crack detection under a plurality of characteristic scales as a final detection result in order to avoid the influence of the change of the crack width on the detection result, namely:
the minimum dimension and the maximum dimension in the formula are respectively set as half of the minimum distinguishable crack width and the maximum crack size; under extreme conditions, the monocrystalline silicon subsurface detection operator can generate relatively large response to the step edge, so that detection errors are caused; aiming at the situation, the step edges are identified by using different local structures of the step edges and the crack regions on the projection line, and only when the difference between the mean value of the projection line near the detected crack and the minimum value of the difference values of the left side and the right side of the projection line does not exceed a set threshold value, the step edges are regarded as crack damage, and the detection of the sub-surface crack damage of the monocrystalline silicon is further completed; and completing the design of the monocrystalline silicon ultrasonic vibration-assisted cutting subsurface crack detection algorithm based on machine vision.
The invention has the beneficial effects that:
according to the method, visual information is used as an information perception means of the monocrystalline silicon subsurface crack damage, the subsurface damage image is obtained by directly scanning the surface of the sample, the crack damage image feature point position and feature scale extraction algorithm provided by the method is accurate in calculation, a complex sample preparation process is not needed for detection, the monocrystalline silicon subsurface crack damage detection can be rapidly, simply and accurately realized, and the detection cost is low.
Drawings
FIG. 1 target multiple light source imaging scheme
FIG. 2 is a cross-sectional profile of single crystal silicon
FIG. 3 shows the experimental results of the detection method.
Detailed Description
The invention is described in detail with reference to the accompanying drawings and the embodiments; the system consists of a crack damage image acquisition module, a target multi-light-source imaging module, a crack damage characteristic extraction module and a crack damage detection module; the image acquisition module scans the surface of the monocrystalline silicon, acquires sub-surface damage image information, the multi-light-source imaging module ensures the uniformity of an acquired image, the crack damage feature extraction module determines the position and the feature scale of a feature point of a crack damage image in a damage image acquired by the acquisition module, and the crack damage detection module detects the information by using a VS2010+ OPENCVC 2.4.13 simulation system.
Designing a crack damage image acquisition module, wherein the crack damage image acquisition module mainly uses an industrial camera CCD (charge coupled device), converts a photoelectric signal into an ordered electric signal and acquires information of a target to be detected; considering the size of the target to be detected and the requirements of items such as an imaging area, a depth of field, a working distance and the like, a telecentric lens of XF-5MDT05X65 is selected to be matched with an industrial camera to acquire the image of the target to be detected.
Designing a target multi-light-source imaging module, using a camera to collect images, wherein the requirements on light sources are very strict, and selecting a multi-light-source lighting scheme according to specific requirements when collecting the sub-surface crack damage images of the monocrystalline silicon so as to obtain the optimal lighting effect and obtain high-quality images; considering that the detection target is the monocrystalline silicon subsurface crack, a forward illumination mode is selected, and an LTLNC100-W linear light source is adopted for lighting the monocrystalline silicon subsurface crack; for a detection target with a complex structure and a large target, the condition of uneven characteristics is easy to occur in the acquisition process, and in order to ensure that the characteristics of different areas in the acquired image are even, a plurality of light sources are adopted to respectively illuminate the different areas, so that the gray value uniformity level of the image is improved; the multiple light sources illuminate the target primarily by means of a superposition of light fields, as shown in fig. 1; the top light source displayed in the upper half part of the figure mainly aims at the central area of the target, the imaging of the edge area is realized by the bottom light source, and when the two light sources are used simultaneously, the aim of relatively uniform image gray value can be achieved, and the image quality is improved; when the image is obtained, a digital image is output through operations such as photoelectric conversion, signal amplification and the like, and crack damage characteristics are extracted in subsequent operations.
The design crack damage characteristic extraction module comprises two parts: 3.1, determining the position of the characteristic point of the crack damage image and 3.2 extracting the crack damage characteristic scale;
in the 3.1 part, calculating partial derivatives of Gaussian kernels H (x, y, alpha) in the neighborhood of a certain pixel point (x, y) on the image, wherein the directions are e and s, and convolving Robert gradients of the neighborhood pixels; the gray-scale covariance matrix in the feature operator is:
wherein
In the formula (I), the compound is shown in the specification,
and
representing the Robert gradient in both directions; in the process of extracting the initial selection point, detecting each pixel in the image one by using a 5 multiplied by 5 window, simultaneously performing convolution operation once, removing noise in the image, and preliminarily determining a candidate point; the specific candidate point is determined to be related to the central pixel of the image, so as to obtain the central pixel
The gray values of the four directions of up, down, left and right are taken as targets, and all directions are calculatedThe upper gray difference absolute value; the calculation formula is as follows:
in the formula (I), the compound is shown in the specification,
respectively representing the gray difference absolute values of the image center pixel in four directions; according to the detection requirement, a threshold value is manually set, and if the calculation result is greater than the set threshold value, the threshold value is calculated
And determining the initial feature point, otherwise, if the calculation result is smaller than the set threshold value, recalculating until the initial feature point is determined.
Extracting crack damage characteristic scale from the part 3.2;
for scale extraction of the feature scale, using image entropy to describe; the method comprises the following specific steps:
in the formula (I), the compound is shown in the specification,
the entropy of the image is represented by the entropy of the image,
a function representing the density of the grey scale distribution,
the size of the gray scale is represented,
represents the radius of a neighborhood circular window with a characteristic point as the center, represents the statistical region of entropy,
representing a feature point position; the value when the local entropy obtains the extreme value is taken as a characteristic scale, and is described as follows:
（7）
obtained by the above formula
Is the characteristic dimension of the target image
(ii) a And after obtaining the characteristic scale and position of the crack damage, detecting the sub-surface crack damage of the monocrystalline silicon.
Designing a crack damage detection module, observing and knowing that the obtained image crack damage characteristics are represented as the sudden change of a gray value by using a computer, wherein the characteristic extraction is mainly realized by using a Gaussian function, and in order to ensure the invariance of the characteristic scale in the detection process, a scale related item is added in front of the second derivative of the Gaussian function to represent as follows:
wherein, the first and the second end of the pipe are connected with each other,
representing a normalization factor. During the detection process, the crack damage is characterized by
Convolution processing when
Will generate positive reaction to cracks and can be used forDetecting cracks; and detecting the crack damage by using the inverse number of the crack damage, and taking the maximum value of crack detection under a plurality of characteristic scales as a final detection result in order to avoid the influence of the change of the crack width on the detection result, namely:
the minimum dimension and the maximum dimension in the formula are respectively set as half of the minimum distinguishable crack width and the maximum crack size; under extreme conditions, the monocrystalline silicon subsurface detection operator can generate relatively large response to the step edge, so that detection errors are caused; aiming at the situation, different local structures of the step edge and the crack area on the projection line are utilized to identify the step edge, and only when the difference between the mean value of the projection line near the detected crack and the minimum value of the difference value of the left side and the right side of the projection line does not exceed a set threshold value, the step edge is regarded as crack damage, and further the detection of the sub-surface crack damage of the monocrystalline silicon is completed.
Performing simulation experiment analysis in an experiment environment with VS2010+ OpenCV2.4.13 and Windows10 operating system, wherein the frame rate of a test image is 30fps, and the resolution is 800 x 600; the monocrystalline silicon surface processing mode adopts an ultrasonic vibration auxiliary cutting mode, and 12 groups of processing parameters and measurement data are extracted in total and are specifically shown in a table 1:
TABLE 1 ultrasonic vibration cutting of subsurface crack depth parameters for single crystal silicon
Considering that the detection method based on machine vision needs to process images, before an experiment, a target is scanned twice according to a window with the size of 60 multiplied by 60 pixels, the step length is set to be 30 pixels, and the condition that the characteristics of cracks are not obvious when the cracks are segmented at a certain time is avoided, so that the detection effect is influenced. For the setting of two times of scanning, the initial position of the first scanning is (0, 0), and the end position is (600 ); the starting position of the second scanning is (30, 30), the ending position is (570 ), the similarity among samples is reduced, and the overlapping among image blocks is ensured to be small; after the processing is finished, selecting an image block containing cracks from the divided image blocks as an experimental data set according to the proportion of 1.
Finishing a detection precision test, setting a judgment threshold value for the experiment of the detection precision of the sub-surface crack damage of the monocrystalline silicon, calculating by using artificial marking data, algorithm detection data and real data, and comprehensively analyzing the detection precision of the crack damage detection algorithm by comprehensively detecting the accuracy and recall ratio; the calculation formula is as follows:
in the formula (I), the compound is shown in the specification,
the accuracy of the detection is shown to be,
the detection recall ratio is shown to be indicated,
the crack image representing the artificial mark is the true value of the extracted crack feature,
indicates all the cracks in the test results,
indicating the part of the data true crack in the detection result,
a real crack that is indicative of an artificial mark,
the result of the crack detection of the algorithm is shown,
indicates a set threshold value and indicates a comprehensive evaluation of accuracy.
The threshold value calculated by the detection algorithm provided by the invention is shown in table 2, and the average level of the threshold value in randomly selected experimental data can reach more than 0.9; comprehensive analysis shows that the precision of the provided monocrystalline silicon subsurface crack damage detection algorithm based on machine vision is high;
TABLE 2 test results of precision
And comparing the crack detection results, randomly selecting a group of data in the experimental data in a centralized manner, simulating a real monocrystalline silicon section contour curve for comparison analysis with the experimental results, and obtaining the selected monocrystalline silicon section contour curve as shown in FIG. 2.
The test result of the provided detection method is basically consistent with the standard curve, and the only unmatched point is positioned in the upper half part and has little influence on the detection result; the experimental results and the detection precision experimental results are comprehensively analyzed, the detection precision based on the monocrystalline silicon subsurface crack damage algorithm is high, the detection result is basically consistent with the actual monocrystalline silicon section, and the monocrystalline silicon subsurface damage can be quickly, efficiently and nondestructively detected.
Claims (6)
1. The monocrystalline silicon subsurface crack damage detection method based on machine vision is characterized by comprising a crack damage image acquisition module, a multi-light-source imaging module, a crack damage extraction module and a crack damage detection module.
2. The crack damage image acquisition module adopts an industrial CCD to acquire target information to be detected; the multi-light source imaging module realizes the uniformity of image characteristics; the crack damage extraction module is used for determining the crack damage characteristic position and scale, comprises a crack damage characteristic position algorithm and a crack damage scale algorithm and is used for realizing the detection of crack damage characteristic points, the maximum scale and the minimum scale of crack damage; the crack damage detection module realizes high-precision detection and generates a high-goodness-of-fit detection curve.
3. The method comprises the following steps:
step 1, crack damage image acquisition, mainly using an industrial camera CCD to convert photoelectric signals into ordered electric signals and acquire target information to be detected; and selecting an XF-5MDT05X65 telecentric lens to be matched with an industrial camera to acquire an image of the target to be detected.
4. Step 2, designing a multi-light-source imaging scheme, wherein a forward illumination mode is selected in consideration of the fact that the detected target is the monocrystalline silicon subsurface crack, and an LTLNC100-W linear light source is adopted to polish the monocrystalline silicon subsurface crack; for a detection target with a complex structure and a large target, the condition of uneven characteristics is easy to occur in the acquisition process, and in order to ensure that the characteristics of different areas in the acquired image are even, a plurality of light sources are adopted to respectively illuminate the different areas, so that the gray value uniformity level of the image is improved; the plurality of light sources illuminate the object primarily by means of a superposition of light fields.
5. Step 3, extracting crack damage characteristics, wherein the process comprises the following substeps:
step 3.1, determining the position of the characteristic point of the crack damage image;
calculating partial derivatives of Gaussian kernels H (x, y, alpha) in the neighborhood of a certain pixel point (x, y) on the image, wherein the directions are e and s, and performing convolution on Robert gradients of the neighborhood pixels; the gray level covariance matrix in the feature operator is:
wherein
In the formula (I), the compound is shown in the specification,
and
representing Robert gradients in two directions, detecting each pixel in the image one by using a 5 multiplied by 5 window in the process of extracting the primary selection point, simultaneously performing convolution operation once, removing noise in the image, and primarily determining the candidate point; the specific candidate point is determined to be related to the central pixel of the image, so as to obtain the central pixel
The gray values in the upper, lower, left and right directions are used as targets, and the gray difference absolute value in each direction is calculated; the calculation formula is as follows:
in the formula (I), the compound is shown in the specification,
respectively representing the gray difference absolute values of the image center pixel in four directions; manually setting a threshold value according to the detection requirement, if calculating the knotIf the threshold value is larger than the set threshold value, the method will be used
And determining the initial feature point, otherwise, if the calculation result is smaller than the set threshold value, recalculating until the initial feature point is determined.
6. Step 3.2, extracting the crack damage characteristic scale;
for scale extraction of the feature scale, using image entropy to describe; the method comprises the following specific steps:
in the formula (I), the compound is shown in the specification,
the entropy of the image is represented by the entropy of the image,
a function representing the density of the grey scale distribution,
the size of the gray scale is represented,
represents the radius of a neighborhood circular window with a characteristic point as the center, represents the statistical region of entropy,
representing a position of a characteristic point; the value when the local entropy obtains the extreme value is taken as a characteristic scale, and is described as follows:
obtained by the above formula
Is the characteristic scale of the target image
(ii) a And after obtaining the characteristic size and position of the crack damage, detecting the sub-surface crack damage of the monocrystalline silicon.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111017834.XA CN115731155A (en) | 2021-09-01 | 2021-09-01 | Monocrystalline silicon subsurface crack damage detection method based on machine vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111017834.XA CN115731155A (en) | 2021-09-01 | 2021-09-01 | Monocrystalline silicon subsurface crack damage detection method based on machine vision |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115731155A true CN115731155A (en) | 2023-03-03 |
Family
ID=85291832
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111017834.XA Pending CN115731155A (en) | 2021-09-01 | 2021-09-01 | Monocrystalline silicon subsurface crack damage detection method based on machine vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115731155A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117351013A (en) * | 2023-12-05 | 2024-01-05 | 天津风霖物联网科技有限公司 | Intelligent detection system and method for building damage |
-
2021
- 2021-09-01 CN CN202111017834.XA patent/CN115731155A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117351013A (en) * | 2023-12-05 | 2024-01-05 | 天津风霖物联网科技有限公司 | Intelligent detection system and method for building damage |
| CN117351013B (en) * | 2023-12-05 | 2024-02-09 | 天津风霖物联网科技有限公司 | Intelligent detection system and method for building damage |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102032875B (en) | Image-processing-based cable sheath thickness measuring method | |
| CN102297660B (en) | Measuring method of shield tunnel lining segment seam open width and apparatus thereof | |
| CN108332689B (en) | Optical measurement system and method for detecting surface roughness and surface damage | |
| CN109859207B (en) | Defect detection method of high-density flexible substrate | |
| CN102288613B (en) | Surface defect detecting method for fusing grey and depth information | |
| Abouelatta | 3D surface roughness measurement using a light sectioning vision system | |
| CN108846397B (en) | Automatic detection method for cable semi-conducting layer based on image processing | |
| US20080292176A1 (en) | Pattern inspection method and pattern inspection apparatus | |
| CN107525768B (en) | Quality control method of DNA ploid analysis equipment | |
| CN112525879B (en) | In-situ identification and rapid quantification method for coal rock micro-components | |
| CN114910480A (en) | Wafer surface defect detection method based on machine vision | |
| CN110823097A (en) | Method for measuring size of optical element in automatic assembly of dense wavelength division multiplexer | |
| CN106770296B (en) | A kind of four ball friction tests mill spot image polishing scratch deflection automatic measuring method | |
| Landstrom et al. | Sub-millimeter crack detection in casted steel using color photometric stereo | |
| CN112489042A (en) | Metal product printing defect and surface damage detection method based on super-resolution reconstruction | |
| CN112730326A (en) | Intelligent identification device and method for rock slices | |
| CN115731155A (en) | Monocrystalline silicon subsurface crack damage detection method based on machine vision | |
| CN113034482A (en) | Surface roughness detection method based on machine vision and machine learning | |
| CN110426395B (en) | Method and device for detecting surface of solar EL battery silicon wafer | |
| CN109785290B (en) | Steel plate defect detection method based on local illumination normalization | |
| CN104515473A (en) | Online diameter detection method of varnished wires | |
| CN116740053B (en) | Management system of intelligent forging processing production line | |
| CN113075195A (en) | Method for depth positioning of microcracks in silicon carbide based on Raman residual stress detection | |
| CN112581424B (en) | Classification extraction method for surface and subsurface defects of optical element | |
| CN110411380B (en) | Non-contact surface roughness measurement method based on wavelet packet texture analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |