CN116612113B - Multi-image stitching detection method based on wafer - Google Patents
Multi-image stitching detection method based on wafer Download PDFInfo
- Publication number
- CN116612113B CN116612113B CN202310869003.8A CN202310869003A CN116612113B CN 116612113 B CN116612113 B CN 116612113B CN 202310869003 A CN202310869003 A CN 202310869003A CN 116612113 B CN116612113 B CN 116612113B
- Authority
- CN
- China
- Prior art keywords
- image
- grain
- arrangement
- product
- images
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 48
- 238000000034 method Methods 0.000 claims abstract description 25
- 230000007547 defect Effects 0.000 claims abstract description 15
- 230000011218 segmentation Effects 0.000 claims abstract description 14
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 8
- 238000003786 synthesis reaction Methods 0.000 claims abstract description 8
- 239000013078 crystal Substances 0.000 claims description 63
- 238000009826 distribution Methods 0.000 claims description 22
- 230000001502 supplementing effect Effects 0.000 claims description 21
- 238000005259 measurement Methods 0.000 claims description 13
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000010586 diagram Methods 0.000 claims description 6
- 238000013461 design Methods 0.000 claims description 5
- 239000000284 extract Substances 0.000 claims description 4
- 230000002194 synthesizing effect Effects 0.000 claims description 4
- 238000005520 cutting process Methods 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 abstract description 7
- 238000000605 extraction Methods 0.000 abstract description 2
- 235000012431 wafers Nutrition 0.000 description 16
- 238000007689 inspection Methods 0.000 description 3
- 229940095676 wafer product Drugs 0.000 description 3
- 230000010354 integration Effects 0.000 description 1
- 239000000523 sample Substances 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- 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
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Testing Or Measuring Of Semiconductors Or The Like (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a wafer-based multi-image stitching detection method, which comprises the following steps: and (3) image acquisition and splicing, namely, based on image grain and mark point segmentation and extraction, calibrating the coordinates of the image grain and mark point to the global coordinates of the product, detecting grain defects and integrating detection data. The problem that the field of view of a camera cannot completely cover the image acquisition of the whole product is solved by an image distributed splicing acquisition method, and the integrity of grain images in an acquisition image area is ensured by longitudinal overlapping splicing between upper and lower images and repeated area guarantee between left and right images; the image detection information and the position relation of the whole product are established through the mapping relation, the global data synthesis of the product can be realized by directly executing single image operation one by one, the problem of image acquisition of high-precision detection of large-size products is solved, the detection requirement of small-grain small-size defects can be met, the operation of splicing images into a complete product image is avoided, and the detection efficiency is effectively improved.
Description
Technical Field
The invention relates to detection of wafer products, in particular to a multi-image stitching detection method based on wafers.
Background
Because the number of the crystal grains on the wafer product is large, the crystal grains are generally different from tens of thousands to hundreds of thousands, the number of the crystal grains arranged on the wafer product is more and more, the size of the crystal grains is smaller and less, and one image cannot completely represent one product; the surface of the crystal grain is often provided with problems of shielding, unfilled corner, deformation and the like, and all crystal grain information is difficult to accurately extract by adopting a template matching or threshold segmentation method.
The chinese patent application No. 2022104779334 discloses a wafer inspection system and a wafer inspection method, the wafer inspection system comprising: computer equipment, photographing equipment, driving equipment, probe card and action executing equipment. The scheme only discloses automatic detection of the display chips in the wafer and screening of bad display chips.
Based on the reasons of larger number of crystal grains, small size and defects on the wafer product, it is necessary to provide a method for dividing and arranging crystal grains, detecting surface defects and synthesizing detection data.
Disclosure of Invention
The invention aims to solve the technical problems that: in order to overcome the defects in the prior art, the multi-image stitching detection method based on the wafer is provided.
The technical scheme adopted by the invention is as follows: a multi-image stitching detection method based on a wafer comprises the following steps:
s1, image acquisition and splicing: acquiring and arranging images through distribution arrangement to generate a complete product image set;
s2, dividing and extracting grains and mark points based on the arrangement image: extracting the coordinate positions of all grains and mark points on each arrangement image in the S1, and giving out the coordinate position (x, y) of each grain based on the corresponding arrangement image, the grain mark and mark point coordinates according to the sequence from top to bottom and from left to right; wherein mark points are position identification points selected on the product;
s3, calibrating the coordinates of the grain and mark points of the image to global coordinates of the product: calculating global coordinate positions of the crystal grains and mark points on the complete product image according to the coordinate positions of all the crystal grains and mark points on each arrangement image extracted in the step S2;
s4, detecting grain defects: cutting a grain image based on the coordinate position in the S2, extracting and detecting the surface defects of the grains on the grain image, and giving out defect information and detection results;
s5, detecting data synthesis: and based on the full-measurement document of the template product and the die arrangement data of the template product, synthesizing the detection result of the actual data, removing the repeated coordinate die, and generating the die coordinate data of the corresponding full-measurement document, wherein the full-measurement document comprises a mark point distribution schematic diagram and a global coordinate distribution schematic diagram.
Further, in S1, the method includes the following steps:
s11, image acquisition: generating a first row of arrangement images by scanning from top to bottom from the left side of a product, translating the product leftwards after finishing the up-down scanning, generating a second row of arrangement images by scanning from bottom to top, continuously translating the product leftwards after finishing the up-down scanning to scan from top to bottom or from bottom to top until finishing the translation, dividing the complete product image into an array arrangement image with the number of rows of M and the number of columns of N, dividing each of the divided arrangement images into an array arrangement image with the same size, wherein the width of each of the arrangement images is W, the height of each of the arrangement images is H, and the arrangement images are marked according to the acquisition sequence and are respectively 1, 2, 3, …, M x N-1 and M x N; after the number of rows, the number of columns and the travel direction of the scanning are determined, the coordinate mapping relation between the arrangement image of each scanning and the whole image of the product is determined, and data statistics and synthesis are carried out through the mapping relation.
S12, image stitching:
s121, longitudinally splicing images: the collected arrangement images are spliced in the longitudinal direction,
the splicing mode in the longitudinal direction of the arrangement images collected by the odd columns is as follows: splicing the tail fixed line number image of the previous collected distribution image to the head of the current image by the collected distribution image;
the splicing mode in the longitudinal direction of the arrangement image acquired by the dual-array is as follows: the head fixed line number image of the previous collected distribution image is spliced to the tail of the current image by the collected distribution image,
the height h of the fixed line number spliced in the longitudinal direction is larger than the height h0 of one grain;
s122, transversely splicing the images: and (3) transversely splicing left and right images of the acquired arrangement images, wherein the splicing mode is finished by hardware design, namely, when the product transversely moves, the overlapping part of fixed columns exists between two adjacent columns of the acquired arrangement images, and the width w of the fixed columns transversely spliced reaches the width w0 of one crystal grain.
The method of splicing front and back images in fixed line numbers during up and down scanning is adopted during the generation of the acquired images so as to ensure the integrity of the crystal grains in the height direction, and in the transverse direction, the method of overlapping and acquiring fixed line numbers at the transverse connection of the product is realized through the design of a mechanical structure so as to ensure the integrity of the crystal grains in the width direction.
Further, in S2, the method includes the following steps:
s21, manufacturing a grain image template in advance: creating a grain image template by using the region of interest, and storing template data;
s22, reading template data, executing a template matching algorithm on the spliced image, performing global matching according to row coordinate positions from left to right, and recording that each row of grains extracts coordinates according to a matching result to generate a grain region of interest;
s23, optimizing a matching result: combining the template matching result with the known information of the grain size, the inter-grain transverse spacing w1 and the inter-grain longitudinal spacing h1, and performing error segmentation grain removal and missing segmentation grain alignment operation;
and S24, after the mark points are positioned, outputting the coordinate position (x, y) and the grain number of each grain based on the corresponding arrangement image, the corresponding product number and the mark point coordinates.
Based on the characteristic that crystal grains on the surface of the wafer are distributed row by row at equal intervals, a transverse filling method is designed, and a segmentation result is optimized on the basis of a template matching algorithm.
Further, in S23, the method includes the following steps:
s231, calculating a difference value c1 between the x-direction coordinate value of the leftmost crystal grain of each row and the x-direction coordinate value of the left edge of the wafer of the row, if c1 is more than w1, supplementing c1/w1 crystal grain coordinates to the left according to the crystal grain size, otherwise, not supplementing, and if the supplementing is invalid crystal grain, removing the crystal grain through subsequent data comprehensive operation;
s232, calculating a difference value c2 between the x-direction coordinate value of the rightmost crystal grain on each row and the x-direction coordinate value of the right edge of the wafer on each row, if c2 is more than w1, supplementing c2/w1 crystal grain coordinates to the right according to the crystal grain size, otherwise, not supplementing, and if the supplementing is invalid crystal grains, removing the crystal grains through subsequent data comprehensive operation;
s233, judging the distance c3 between adjacent grains, if c3 is more than w1, supplementing c3/w1 grain coordinates among the adjacent grains according to the grain size, and if invalid grains are supplemented, removing the grains through subsequent data comprehensive operation; if c3 is less than w1, comparing the template matching scores of the two crystal grains, and eliminating the crystal grains with smaller matching scores.
Further, in S3,
the position of the image in the global product can be obtained through the image label acquired in the S2, wherein the calculation formulas of row coordinates m and column coordinates n of the arranged image with the label of i in the product position are as follows:
,
,
wherein M is the number of lines of the arrangement image obtained after the complete product image is segmented, N is the number of columns of the arrangement image obtained after the complete product image is segmented, and i is the corresponding sequence label when the arrangement image is acquired;
the calculation formula of the coordinate position (X, Y) on the arrangement image with the reference number i corresponding to the coordinate position (X, Y) on the product is as follows:
,
wherein W is the width of the arranged images, H is the height of the arranged images, H is the height of the fixed line number spliced in the longitudinal direction of the arranged images, and W is the width of the fixed column number spliced in the transverse direction of the arranged images.
Further, in S5, the method includes the following steps:
s51, analyzing the full-test document of the template product and the grain arrangement data of the template product;
s52, searching the coordinate position of the mark point at the origin according to the result of the S2;
s53, resetting the global coordinate position of the crystal grain on the complete product image according to the coordinate position of the origin;
s54, comparing the actual detection result with template arrangement data;
and S55, generating grain coordinate data corresponding to the full-measurement document by the conforming grains, otherwise, eliminating the repeated coordinate grains.
Further, in S5, the comprehensive rule of the grain detection result is:
(1) If the actual detection result has data and the template is arranged with data, the actual classification result is taken as the reference;
(2) If the actual detection result has data and the template arrangement has no data, the actual classification result is taken as the reference;
(3) If the actual detection result has no data, the template is provided with data, and the classification result is fixed.
And (3) carrying out full product synthesis on the detection data in the multi-image acquisition and splicing mode, so as to ensure the correctness and the integrity of the detection data of the product: designing a coordinate calibration algorithm from a single arranged picture to a global product picture, mapping the coordinates of the crystal grains and mark points segmented based on the image to the global product coordinates, and resetting the original point coordinates according to the data of the full-measurement document; and a data comprehensive algorithm is designed to verify the actual detection data and the template arrangement data, and a final comprehensive detection result is output.
Compared with the prior art, the invention has the following advantages:
the distributed splicing and collecting method for the images effectively ensures the detection precision of large-size products, solves the problem that the field of view of a camera cannot completely cover the image collection of the whole product, and ensures the integrity of grain images in the area of the collected images by longitudinally overlapping and splicing the upper and lower images and guaranteeing the repeated areas between left and right images;
the position relation between the image detection information and the whole product is established through the mapping relation, and the global data synthesis of the product can be realized by directly executing single image operation one by one, so that the problem of image acquisition of high-precision detection of large-size products is solved, the detection requirement of small-grain small-size defects can be met, the operation of splicing images into a complete product image is avoided, and the detection efficiency is effectively improved;
the transverse alignment method provided by the invention ensures the accuracy and stability of the segmentation of the grain image, and simultaneously improves the anti-interference capability of the segmentation algorithm: extracting the coordinates of the full-image grains by a shape template matching method, and carrying out the alignment operation on the left-hand, right-hand and middle-missing segmentation conditions of the grains in the same row by combining the information of the grain size, the grain horizontal-longitudinal spacing and the like, and carrying out the rejection operation on the middle-missing segmentation conditions of the grains in the same row; and under the condition of multiple divisions caused by the filling operation, the subsequent data comprehensive flow is regarded as invalid divisions by comparing with the full-measured document data, and the result statistics is not recorded.
Drawings
FIG. 1 is a schematic view of an image acquisition distribution of the present invention;
FIG. 2 is a camera scan roadmap of the invention;
FIG. 3 is a flow chart of the present invention;
FIG. 4 is a flow chart of the segmentation extraction of the present invention;
fig. 5 is a flow chart of the present invention for detecting data integration.
Detailed Description
The following describes the embodiments of the present invention in detail, and the embodiments and specific operation procedures are given on the premise of the technical solution of the present invention, but the scope of protection of the present invention is not limited to the following embodiments.
A multi-image stitching detection method based on a wafer, as shown in FIG. 3, comprises the following steps:
s1, image acquisition and splicing: acquiring and arranging images through distribution arrangement to generate a complete product image set;
s11, image acquisition: as shown in fig. 2, a first column arrangement image is generated by scanning from top to bottom from the left side of a product, the product translates leftwards after the upper and lower scanning is finished, a second column arrangement image is generated by scanning from bottom to top, the product translates leftwards after the upper and lower scanning is finished, the product continues to scan from top to bottom or from bottom to top until the translation is finished, the complete product image is divided into an array arrangement image with the number of rows of M and the number of columns of N, the size of each divided arrangement image is consistent, the width of the arrangement image is W, the height of the arrangement image is H, and the arrangement images are marked according to the acquisition sequence, namely 1, 2, 3, …, M x N-1 and M x N; after the number of rows, the number of columns and the travel direction of the scanning are determined, the coordinate mapping relation between the arrangement image of each scanning and the whole image of the product is determined, and data statistics and synthesis are carried out through the mapping relation;
s12, image stitching:
s121, longitudinally splicing images: the collected arrangement images are spliced in the longitudinal direction,
the splicing mode in the longitudinal direction of the arrangement images collected by the odd columns is as follows: splicing the tail fixed line number image of the previous collected distribution image to the head of the current image by the collected distribution image;
the splicing mode in the longitudinal direction of the arrangement image acquired by the dual-array is as follows: the head fixed line number image of the previous collected distribution image is spliced to the tail of the current image by the collected distribution image,
the height h of the fixed line number spliced in the longitudinal direction is larger than the height h0 of one grain;
s122, transversely splicing the images: the method comprises the steps of transversely splicing left and right images of the acquired arrangement images, wherein the splicing mode is completed by hardware design, namely, when a product transversely moves, overlapping parts of fixed columns exist between two adjacent columns of the acquired arrangement images, and the width w of the fixed columns transversely spliced reaches the width w0 of one crystal grain;
the method of splicing front and back images in fixed line numbers during up and down scanning is adopted during the generation of the acquired images so as to ensure the integrity of the crystal grains in the height direction, and in the transverse direction, the method of overlapping and acquiring fixed line numbers at the transverse connection of the product is realized through the design of a mechanical structure so as to ensure the integrity of the crystal grains in the width direction.
S2, dividing and extracting grains and mark points based on the arrangement image: extracting the coordinate positions of all grains and mark points on each arrangement image in the S1, and giving out the coordinate position (x, y) of each grain based on the corresponding arrangement image, the grain mark and mark point coordinates according to the sequence from top to bottom and from left to right; wherein mark points are position identification points selected on the product;
as shown in fig. 4, the method specifically comprises the following steps:
s21, manufacturing a grain image template in advance: creating a grain image template by using the region of interest, and storing template data;
s22, reading template data, executing a template matching algorithm on the spliced image, performing global matching according to row coordinate positions from left to right, and recording that each row of grains extracts coordinates according to a matching result to generate a grain region of interest;
s23, optimizing a matching result: combining the template matching result with the known information of the grain size, the inter-grain transverse spacing w1 and the inter-grain longitudinal spacing h1, and performing error segmentation grain removal and missing segmentation grain alignment operation; based on the characteristic that crystal grains on the surface of the wafer are distributed row by row at equal intervals, a transverse filling method is designed, and a division result is optimized on the basis of a template matching algorithm;
s231, calculating a difference value c1 between the x-direction coordinate value of the leftmost crystal grain of each row and the x-direction coordinate value of the left edge of the wafer of the row, if c1 is more than w1, supplementing c1/w1 crystal grain coordinates to the left according to the crystal grain size, otherwise, not supplementing, and if the supplementing is invalid crystal grain, removing the crystal grain through subsequent data comprehensive operation;
s232, calculating a difference value c2 between the x-direction coordinate value of the rightmost crystal grain on each row and the x-direction coordinate value of the right edge of the wafer on each row, if c2 is more than w1, supplementing c2/w1 crystal grain coordinates to the right according to the crystal grain size, otherwise, not supplementing, and if the supplementing is invalid crystal grains, removing the crystal grains through subsequent data comprehensive operation;
s233, judging the distance c3 between adjacent grains, if c3 is more than w1, supplementing c3/w1 grain coordinates among the adjacent grains according to the grain size, and if invalid grains are supplemented, removing the grains through subsequent data comprehensive operation; if c3 is less than w1, comparing the template matching scores of the two crystal grains, and eliminating the crystal grains with smaller matching scores;
and S24, after the mark points are positioned, outputting the coordinate position (x, y) and the grain number of each grain based on the corresponding arrangement image, the corresponding product number and the mark point coordinates.
S3, calibrating the coordinates of the grain and mark points of the image to global coordinates of the product: calculating global coordinate positions of the crystal grains and mark points on the complete product image according to the coordinate positions of all the crystal grains and mark points on each arrangement image extracted in the step S2;
the position of the image in the global product can be obtained through the image label acquired in the S2, wherein the calculation formulas of row coordinates m and column coordinates n of the arranged image with the label of i in the product position are as follows:
,
,
wherein M is the number of lines of the arrangement image obtained after the complete product image is segmented, N is the number of columns of the arrangement image obtained after the complete product image is segmented, and i is the corresponding sequence label when the arrangement image is acquired;
the calculation formula of the coordinate position (X, Y) on the arrangement image with the reference number i corresponding to the coordinate position (X, Y) on the product is as follows:
,
wherein W is the width of the arranged images, H is the height of the arranged images, H is the height of the fixed line number spliced in the longitudinal direction of the arranged images, and W is the width of the fixed column number spliced in the transverse direction of the arranged images.
S4, detecting grain defects: and (3) cutting the grain image based on the coordinate position in the S2, extracting and detecting the surface defects of the grains on the grain image, and giving out defect information and detection results.
S5, detecting data synthesis: based on the full-measurement document of the template product and the die arrangement data of the template product, synthesizing the detection result of the actual data, removing the repeated coordinate die, and generating die coordinate data corresponding to the full-measurement document, wherein the full-measurement document comprises a mark point distribution schematic diagram and a global coordinate distribution schematic diagram;
as shown in fig. 5, the method specifically comprises the following steps:
s51, analyzing the full-test document of the template product and the grain arrangement data of the template product;
s52, searching the coordinate position of the mark point at the origin according to the result of the S2;
s53, resetting the global coordinate position of the crystal grain on the complete product image according to the coordinate position of the origin;
s54, comparing the actual detection result with template arrangement data;
and S55, generating grain coordinate data corresponding to the full-measurement document by the conforming grains, otherwise, eliminating the repeated coordinate grains.
The comprehensive rule of the grain detection result is as follows:
(1) If the actual detection result has data and the template is arranged with data, the actual classification result is taken as the reference;
(2) If the actual detection result has data and the template arrangement has no data, the actual classification result is taken as the reference;
(3) If the actual detection result has no data, the template is provided with data, and the classification result is fixed.
Claims (6)
1. The multi-image stitching detection method based on the wafer is characterized by comprising the following steps of:
s1, image acquisition and splicing: acquiring and arranging images through distribution arrangement to generate a complete product image set;
s11, image acquisition: generating a first row of arrangement images by scanning from top to bottom from the left side of a product, translating the product leftwards after finishing the up-down scanning, generating a second row of arrangement images by scanning from bottom to top, continuously translating the product leftwards after finishing the up-down scanning to scan from top to bottom or from bottom to top until finishing the translation, dividing the complete product image into an array arrangement image with the number of rows of M and the number of columns of N, dividing each of the divided arrangement images into an array arrangement image with the same size, wherein the width of each of the arrangement images is W, the height of each of the arrangement images is H, and the arrangement images are marked according to the acquisition sequence and are respectively 1, 2, 3, …, M x N-1 and M x N;
s12, image stitching:
s121, longitudinally splicing images: the collected arrangement images are spliced in the longitudinal direction,
the splicing mode in the longitudinal direction of the arrangement images collected by the odd columns is as follows: splicing the tail fixed line number image of the previous collected distribution image to the head of the current image by the collected distribution image;
the splicing mode in the longitudinal direction of the arrangement image acquired by the dual-array is as follows: the head fixed line number image of the previous collected distribution image is spliced to the tail of the current image by the collected distribution image,
the height h of the fixed line number spliced in the longitudinal direction is larger than the height h0 of one grain;
s122, transversely splicing the images: the method comprises the steps of transversely splicing left and right images of the acquired arrangement images, wherein the splicing mode is completed by hardware design, namely, when a product transversely moves, overlapping parts of fixed columns exist between two adjacent columns of the acquired arrangement images, and the width w of the fixed columns transversely spliced reaches the width w0 of one crystal grain;
s2, dividing and extracting grains and mark points based on the arrangement image: extracting the coordinate positions of all grains and mark points on each arrangement image in the S1, and giving out the coordinate position (x, y) of each grain based on the corresponding arrangement image, the grain mark and mark point coordinates according to the sequence from top to bottom and from left to right; wherein mark points are position identification points selected on the product;
s3, calibrating the coordinates of the grain and mark points of the image to global coordinates of the product: calculating global coordinate positions of the crystal grains and mark points on the complete product image according to the coordinate positions of all the crystal grains and mark points on each arrangement image extracted in the step S2;
s4, detecting grain defects: cutting a grain image based on the coordinate position in the S2, extracting and detecting the surface defects of the grains on the grain image, and giving out defect information and detection results;
s5, detecting data synthesis: and based on the full-measurement document of the template product and the die arrangement data of the template product, synthesizing the detection result of the actual data, removing the repeated coordinate die, and generating the die coordinate data of the corresponding full-measurement document, wherein the full-measurement document comprises a mark point distribution schematic diagram and a global coordinate distribution schematic diagram.
2. The multi-image stitching detection method according to claim 1, wherein S2 comprises the steps of:
s21, manufacturing a grain image template in advance: creating a grain image template by using the region of interest, and storing template data;
s22, reading template data, executing a template matching algorithm on the spliced image, performing global matching according to row coordinate positions from left to right, and recording that each row of grains extracts coordinates according to a matching result to generate a grain region of interest;
s23, optimizing a matching result: combining the template matching result with the known information of the grain size, the inter-grain transverse spacing w1 and the inter-grain longitudinal spacing h1, and performing error segmentation grain removal and missing segmentation grain alignment operation;
and S24, after the mark points are positioned, outputting the coordinate position (x, y) and the grain number of each grain based on the corresponding arrangement image, the corresponding product number and the mark point coordinates.
3. The multi-image stitching detection method according to claim 2, wherein in S23, comprising the steps of:
s231, calculating a difference value c1 between the x-direction coordinate value of the leftmost crystal grain of each row and the x-direction coordinate value of the left edge of the wafer of the row, if c1 is more than w1, supplementing c1/w1 crystal grain coordinates to the left according to the crystal grain size, otherwise, not supplementing, and if the supplementing is invalid crystal grain, removing the crystal grain through subsequent data comprehensive operation;
s232, calculating a difference value c2 between the x-direction coordinate value of the rightmost crystal grain on each row and the x-direction coordinate value of the right edge of the wafer on each row, if c2 is more than w1, supplementing c2/w1 crystal grain coordinates to the right according to the crystal grain size, otherwise, not supplementing, and if the supplementing is invalid crystal grains, removing the crystal grains through subsequent data comprehensive operation;
s233, judging the distance c3 between adjacent grains, if c3 is more than w1, supplementing c3/w1 grain coordinates among the adjacent grains according to the grain size, and if invalid grains are supplemented, removing the grains through subsequent data comprehensive operation; if c3 is less than w1, comparing the template matching scores of the two crystal grains, and eliminating the crystal grains with smaller matching scores.
4. The multi-image stitching method as recited in claim 1, wherein in S3,
the position of the image in the global product can be obtained through the image label acquired in the S2, wherein the calculation formulas of row coordinates m and column coordinates n of the arranged image with the label of i in the product position are as follows:
wherein M is the number of lines of the arrangement image obtained after the complete product image is segmented, N is the number of columns of the arrangement image obtained after the complete product image is segmented, and i is the arrangementCorresponding sequence labels during image acquisition;
the calculation formula of the coordinate position (X, Y) on the arrangement image with the reference number i corresponding to the coordinate position (X, Y) on the product is as follows:
wherein W is the width of the arranged images, H is the height of the arranged images, H is the height of the fixed line number spliced in the longitudinal direction of the arranged images, and W is the width of the fixed column number spliced in the transverse direction of the arranged images.
5. The multi-image stitching method according to claim 1, wherein in S5, comprising the steps of:
s51, analyzing the full-test document of the template product and the grain arrangement data of the template product;
s52, searching the coordinate position of the mark point at the origin according to the result of the S2;
s53, resetting the global coordinate position of the crystal grain on the complete product image according to the coordinate position of the origin;
s54, comparing the actual detection result with template arrangement data;
and S55, generating grain coordinate data corresponding to the full-measurement document by the conforming grains, otherwise, eliminating the repeated coordinate grains.
6. The multi-image stitching method according to any one of claims 1 and 5, wherein in S5, the overall rule of the grain detection result is:
(1) If the actual detection result has data and the template is arranged with data, the actual classification result is taken as the reference;
(2) If the actual detection result has data and the template arrangement has no data, the actual classification result is taken as the reference;
(3) If the actual detection result has no data, the template is provided with data, and the classification result is fixed.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310869003.8A CN116612113B (en) | 2023-07-17 | 2023-07-17 | Multi-image stitching detection method based on wafer |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310869003.8A CN116612113B (en) | 2023-07-17 | 2023-07-17 | Multi-image stitching detection method based on wafer |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116612113A CN116612113A (en) | 2023-08-18 |
CN116612113B true CN116612113B (en) | 2023-09-15 |
Family
ID=87678574
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310869003.8A Active CN116612113B (en) | 2023-07-17 | 2023-07-17 | Multi-image stitching detection method based on wafer |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116612113B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2011090003A (en) * | 2010-12-06 | 2011-05-06 | Hitachi Ltd | Image display method of defective candidates |
CN113288429A (en) * | 2021-05-07 | 2021-08-24 | 山东商业职业技术学院 | Space registration and real-time navigation method of breast minimally invasive interventional operation robot |
CN113670223A (en) * | 2021-07-20 | 2021-11-19 | 温州职业技术学院 | 3D vision detection system and method thereof |
CN114387433A (en) * | 2022-01-14 | 2022-04-22 | 佛山市南海区广工大数控装备协同创新研究院 | High-precision positioning method for large-amplitude FPC |
CN114627072A (en) * | 2022-03-14 | 2022-06-14 | 苏州中巨苏智能科技有限公司 | Wafer defect extraction method based on neural network |
CN114638796A (en) * | 2022-03-08 | 2022-06-17 | 中南大学湘雅医院 | Deep tumour accurate positioning image processing system in brain based on ultrasonic medicine |
WO2022237029A1 (en) * | 2021-05-11 | 2022-11-17 | 杭州众硅电子科技有限公司 | Apparatus for calibrating positions of polishing head and loading/unloading platform, polishing device, and calibration method |
CN115661148A (en) * | 2022-12-26 | 2023-01-31 | 视睿(杭州)信息科技有限公司 | Wafer crystal grain arrangement detection method and system |
-
2023
- 2023-07-17 CN CN202310869003.8A patent/CN116612113B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2011090003A (en) * | 2010-12-06 | 2011-05-06 | Hitachi Ltd | Image display method of defective candidates |
CN113288429A (en) * | 2021-05-07 | 2021-08-24 | 山东商业职业技术学院 | Space registration and real-time navigation method of breast minimally invasive interventional operation robot |
WO2022237029A1 (en) * | 2021-05-11 | 2022-11-17 | 杭州众硅电子科技有限公司 | Apparatus for calibrating positions of polishing head and loading/unloading platform, polishing device, and calibration method |
CN113670223A (en) * | 2021-07-20 | 2021-11-19 | 温州职业技术学院 | 3D vision detection system and method thereof |
CN114387433A (en) * | 2022-01-14 | 2022-04-22 | 佛山市南海区广工大数控装备协同创新研究院 | High-precision positioning method for large-amplitude FPC |
CN114638796A (en) * | 2022-03-08 | 2022-06-17 | 中南大学湘雅医院 | Deep tumour accurate positioning image processing system in brain based on ultrasonic medicine |
CN114627072A (en) * | 2022-03-14 | 2022-06-14 | 苏州中巨苏智能科技有限公司 | Wafer defect extraction method based on neural network |
CN115661148A (en) * | 2022-12-26 | 2023-01-31 | 视睿(杭州)信息科技有限公司 | Wafer crystal grain arrangement detection method and system |
Non-Patent Citations (3)
Title |
---|
Multi-image mosaic with SIFT and vision measurement for microscale structures processed by femtosecond laser;Fu-Bin Wang等;《Optics and Lasers in Engineering》;第100卷;124-130 * |
晶圆表面缺陷自动检测技术的研究;倪天宇;《中国优秀硕士学位论文全文数据库 (信息科技辑)》(第1期);I135-460 * |
边缘检测法在晶圆缺陷检测中的应用;方宝英等;《计算机与数字工程》;第38卷(第6期);132-134 * |
Also Published As
Publication number | Publication date |
---|---|
CN116612113A (en) | 2023-08-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113077453B (en) | Circuit board component defect detection method based on deep learning | |
CN107423501B (en) | Array automatic detection and numbering method based on photovoltaic power station logic diagram | |
US7646908B2 (en) | Defect detection apparatus and defect detection method | |
US20080232715A1 (en) | Image processing apparatus | |
TWI533383B (en) | Design-based inspection using repeating structures | |
CN105989588B (en) | Special-shaped material cutting image correction method and system | |
CN114152627B (en) | Chip circuit defect detection method and device, electronic equipment and storage medium | |
CN111461133B (en) | Express delivery surface single item name identification method, device, equipment and storage medium | |
CN111274961B (en) | Character recognition and information analysis method for flexible IC substrate | |
CN116433672A (en) | Silicon wafer surface quality detection method based on image processing | |
CN116612113B (en) | Multi-image stitching detection method based on wafer | |
CN113112396B (en) | Method for detecting conductive particles | |
KR101653861B1 (en) | Drawing data generating method, drawing method, drawing data generating apparatus and drawing apparatus | |
CN116309573B (en) | Defect detection method for printed characters of milk packaging box | |
CN116189212A (en) | Intelligent table identification and scoring method based on image | |
CN116385347A (en) | Deformation analysis-based visual inspection method for aircraft skin curved surface pattern | |
CN111860332B (en) | Dual-channel electrokinetic diagram part detection method based on multi-threshold cascade detector | |
CN114463544A (en) | Irregular object semantic segmentation quick labeling method | |
CN113191351A (en) | Reading identification method and device of digital electric meter and model training method and device | |
CN113792721B (en) | Instrument detection method based on one-shot mechanism | |
CN117132603B (en) | Wafer map generation method, device, equipment and storage medium | |
CN118097121B (en) | Target recognition and counting method and device based on image segmentation and deep learning | |
CN117314906B (en) | Wafer grain detection method based on array distribution information guidance | |
CN117974601B (en) | Method and system for detecting surface defects of silicon wafer based on template matching | |
CN114612915B (en) | Method and device for extracting patient information of film image |
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 |