CN111340701A - Circuit board image splicing method for screening matching points based on clustering method - Google Patents
Circuit board image splicing method for screening matching points based on clustering method Download PDFInfo
- Publication number
- CN111340701A CN111340701A CN202010111022.0A CN202010111022A CN111340701A CN 111340701 A CN111340701 A CN 111340701A CN 202010111022 A CN202010111022 A CN 202010111022A CN 111340701 A CN111340701 A CN 111340701A
- Authority
- CN
- China
- Prior art keywords
- circuit board
- matching
- image
- points
- matching points
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4038—Scaling the whole image or part thereof for image mosaicing, i.e. plane images composed of plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
Abstract
The invention discloses a circuit board image splicing method for screening matching points based on a clustering method, which comprises the steps of obtaining a large number of circuit board local images with overlapped areas through a high-definition camera; extracting SIFT feature points on the gray level image of the circuit board, and matching the feature points by using a rapid nearest neighbor matching algorithm; sorting the circuit board matching points by utilizing quick sorting, and preliminarily screening out the circuit board matching points with better matching degree; calculating the slope and the distance between each pair of circuit board matching points, and screening the circuit board matching points by using a K-means clustering method; acquiring a transformation matrix by using a RANSAC algorithm; and deforming and fusing the input image into the output image to complete splicing, thereby obtaining the global circuit board image with ultrahigh resolution. According to the invention, better matching points are screened out through a K-means clustering algorithm, the influence of bad matching points on affine transformation is reduced, and the accuracy of circuit board image splicing is improved.
Description
Technical Field
The invention belongs to the technical field of image splicing, and particularly relates to a splicing technology of a large number of high-resolution images of circuit boards.
Background
The image splicing technology is widely applied to actual scenes, such as remote sensing images, unmanned aerial vehicle aerial photography and the like, image splicing is a precursor step for further understanding the images, and the quality of the splicing effect directly influences the next work. The image stitching steps which are widely applied at present are divided into the following steps: firstly, preprocessing each image; extracting characteristic points of each image and matching the characteristic points; then carrying out image registration; then, copying the image; and finally, carrying out image fusion to obtain a final image.
The image preprocessing has an important influence on the subsequent image stitching process, the quality of the image preprocessing directly influences the image stitching result, and the currently commonly used image preprocessing methods generally include image noise suppression, graying processing, smoothing processing, image correction and the like.
The common feature point extraction method comprises the following steps: the SIFT algorithm searches image positions on all scales, identifies potential interest points invariant to scale and rotation through a gaussian differential function, determines position and scale through a fitting fine model on each candidate position, assigns one or more directions to each keypoint position based on the local gradient direction of the image, measures local gradients of the image on a selected scale row in a neighborhood around each keypoint, transforms the gradients into descriptions of the keypoints, and essentially searches the keypoints on different scale spaces and calculates the directions of the keypoints. The SURF algorithm is used for constructing a Hessian matrix to construct a Gaussian pyramid scale space, each pixel point can work out a Hessian matrix, non-maximum values are used for restraining and preliminarily determining key points, the values of the extreme points are accurately positioned, the main direction of the feature points is selected, and feature point descriptors are constructed.
Common feature matching methods include: the mutual information method is a method for matching features by comparing the entropy values of the information contents of two images mutually including each other, wherein the entropy value of the information contents is a representation of the probability distribution of the images. And (2) clustering characteristic matching, wherein the relative principal angles of characteristic point pairs are subjected to statistical analysis through clustering calculation, a plurality of local maximum values are searched, then the initial matching point set is clustered again according to each center to obtain a related pixel point pair set, and the pixel point set with the minimum average distance is selected and characteristic point matching is carried out according to the pixel point pair set. And the correlation coefficient method is used for determining the image matching degree by calculating the correlation value between the matched image and the search template.
A large number of local high-definition circuit board images have the characteristics of high background information similarity, large quantity and the like, and by adopting the conventional method, in order to ensure that enough feature points can be extracted for feature matching, a lower interest threshold value is usually set, and the condition that the proportion of invalid feature points is far higher than that of valid feature points is easy to occur in the circuit board images with high similarity, so that the problem of mismatching is easy to occur during feature point matching, the difficulty of eliminating wrong matching points is increased, the splicing confusion is finally caused, and the accuracy of splicing cannot be ensured. Only by overcoming the problem, the accurate splicing of the high-definition circuit board images can be completed.
Disclosure of Invention
The purpose of the invention is as follows: aiming at the prior art, a circuit board image splicing method for screening matching points based on a clustering method is provided, and the problem of poor circuit board image splicing effect in the prior art is solved.
The technical scheme is as follows: a circuit board image splicing method for screening matching points based on a clustering method comprises the following steps:
step 1: acquiring a plurality of circuit board local images with overlapping areas through a high-definition camera;
step 2: extracting SIFT feature points on the gray level image of the circuit board, and matching the feature points by using a rapid nearest neighbor matching algorithm;
and step 3: sorting the circuit board matching points by utilizing quick sorting, and preliminarily screening out the circuit board matching points with better matching degree;
and 4, step 4: calculating the slope and the distance between each pair of circuit board matching points, and screening the circuit board matching points by using a K-means clustering method;
and 5: acquiring a transformation matrix by using a RANSAC algorithm;
step 6: and deforming and fusing the input image into the output image to complete splicing, thereby obtaining the global circuit board image with ultrahigh resolution.
Further, in step 1, the image size is 2048 × 2048, one circuit board includes 30-120 partial images, the image naming order is sequentially from left to right and from top to bottom, which is marked as x rows and y columns, and the overlapping area occupies 1/3 of the width of each image.
Further, the step 2 comprises the following specific steps:
step 2-1: converting the three-channel color circuit board image into a single-channel gray circuit board image;
step 2-2: intercepting a right 1/2 part of a spliced image on the left side of the gray circuit board and a left 1/2 part of the spliced image on the right side of the gray circuit board;
step 2-3: respectively extracting feature points on the two intercepted circuit board images by utilizing an SIFT algorithm, wherein the upper limit of the number of the feature points is M;
step 2-4: and matching the characteristic points on the two circuit board images by using a fast nearest neighbor matching algorithm.
Further, the step 3 comprises the following specific steps:
obtaining distance D ═ D between circuit board matching points by using fast nearest neighbor matching algorithm1,d2,L,dmAfter the step is completed, the matching distance D is changed to { D } by adopting a quick sorting method1,d2,L,dmSorting the N matching points from small to large, and selecting the first N matching points with smaller distance as P1,p2,L,pn}。
Further, the step 4 comprises the following specific steps:
step 4-1: clustering the slopes by using a K-means clustering method, finding out the most numerous circuit board matching points, discarding other circuit board matching points, and taking the most numerous circuit board matching points as excellent circuit board matching points;
step 4-2: and performing secondary screening on the excellent circuit board matching points, clustering the distances of the circuit board matching points by using a K-means clustering method, taking the most numerous circuit board matching points as final excellent circuit board matching points, and excluding other circuit board matching points.
Further, the clustering the slope by using the K-means clustering method in the step 4-1 includes: calculating the slope of each pair of matching point connecting lines by using a slope calculation formula of the two point connecting linesRecording the data set as a K-means cluster, and randomly selecting K0Taking the data as a centroid, calculating the distance from each data point to each centroid, distributing each point in the data set to a cluster corresponding to the closest centroid, updating the centroid of each cluster to the average value of all the points of the cluster, and judging whether the centroid of each cluster is at the precision theta0If the iteration times are not reached or not, the algorithm is ended if the iteration times are not reached, otherwise, the random selection k is returned0The data serves as the centroid and the iteration continues.
Further, the clustering the distances of the matching points by using a K-means clustering method in the step 4-2 includes: calculating the distance between each pair of matching points by using a distance calculation formula between the two points, recording the distance as a data set of K-means clustering, and randomly selecting K1Taking the data as a centroid, calculating the distance from each data point to each centroid, distributing each point in the data set to a cluster corresponding to the closest centroid, updating the centroid of each cluster to the average value of all the points of the cluster, and judging whether the centroid of each cluster is at the precision theta1If the iteration times are not reached or not, the algorithm is ended if the iteration times are not reached, otherwise, the random selection k is returned1The data serves as the centroid and the iteration continues.
Further, the step 5 comprises the following specific steps:
step 5-1: firstly, randomly selecting 4 pairs of matching points from a data set, and determining a transformation matrix Hi；
Step 5-2: then each matching point p is calculatedjMatching point p after transformation of matrix Hij', transform formula:
pj′＝Hpj
step 5-3: calculating the distance from each matching point to the corresponding matching point after the transformation matrix;
step 5-4: then the matching point set is divided into inner points and outer points through a threshold value mu, and H is carried out if the number of the inner points is large enoughiReasonably, estimating a matrix H by using all the interior points, and stopping iteration, otherwise, continuing the 1-4 steps;
step 5-5: and finally, if the iteration times exceed the preset maximum iteration times L, exiting.
Further, the step 6 includes the following specific steps:
deforming and fusing the input image to a corresponding output image by using the matrix H, and taking 1/2 of the image overlapping area as a smooth area XsFusing the smoothed regions X using a weighted averagesThe pixel point of (2) is used for smoothly splicing gaps and avoiding image large-area blurring, and the smoothing formula is as follows:
wherein wiIs the abscissa of the current point and the smooth area XsThe quotient of the widths is,
the pixel values of the channels of the image are stitched for the left side of the smooth region,
splicing pixel values of all channels of the image on the right side of the smooth area;
and firstly splicing the rows of the circuit board image, and then splicing the columns of the circuit board image to obtain the final global circuit board image with ultrahigh resolution.
Has the advantages that: according to the invention, a high-definition camera is adopted to obtain a large number of high-resolution circuit board local images, the image fineness is increased, the number of extracted circuit board characteristic points is effectively increased, and circuit board characteristic point pairs which are well matched are screened and matched on the Euclidean distance and the slope of circuit board matching points by adopting a K-means clustering algorithm, so that the problem of characteristic point matching disorder caused by high similarity of circuit board image information is solved, and the splicing of a large number of high-definition circuit board images is accurately realized. The invention can provide the global circuit board image with ultrahigh resolution for the defect detection of the circuit board, and reduces the problems of splicing dislocation, deformation and the like of a large number of circuit board images.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a diagram illustrating the effect of image feature matching of the circuit board according to the present invention;
FIG. 3 is a diagram of the effect of the circuit board image on the preliminary screening of feature matching by the fast sorting method according to the present invention;
FIG. 4 is a graph of the effect of feature matching after the circuit board image is screened by the K-means clustering method;
FIG. 5 is an image of a completed circuit board being spliced in accordance with the present invention;
FIG. 6 is an enlarged detail effect view of the splice gap of the complete circuit board of the present invention;
Detailed Description
The invention is further explained below with reference to the drawings.
As shown in fig. 1, a circuit board image stitching method for screening matching points based on a clustering method includes the following steps:
s101: acquiring a large number of circuit board local images with overlapping areas through a high-definition camera;
s102: extracting SIFT feature points on the gray level image of the circuit board, and matching the feature points by using a rapid nearest neighbor matching algorithm;
s103: sorting the circuit board matching points by utilizing quick sorting, and preliminarily screening out the circuit board matching points with better matching degree;
s104: calculating the slope and the distance between each pair of circuit board matching points, and screening the circuit board matching points by using a K-means clustering method;
s105: acquiring a transformation matrix by using a RANSAC algorithm;
s106: and deforming and fusing the input image into the output image to complete splicing, thereby obtaining the global circuit board image with ultrahigh resolution.
As shown in fig. 2, the SIFT feature points on the circuit board grayscale image are extracted in step S102, and the feature point effect graph is matched by using the fast nearest neighbor matching algorithm, which includes the following steps:
2-1, converting the three-channel color circuit board image into a single-channel gray circuit board image;
step 2-2, intercepting the right 1/2 part of the spliced image on the left side of the gray circuit board and the left 1/2 part of the spliced image on the right side;
2-3, respectively extracting feature points on the two intercepted circuit board images by utilizing an SIFT algorithm, wherein the upper limit of the number of the feature points is M;
and 2-4, matching the characteristic points on the two circuit board images by using a rapid nearest neighbor matching algorithm.
As shown in fig. 3, the step S103 of sorting the circuit board matching points by fast sorting includes: obtaining the distance D ═ D between matching points by using a fast nearest neighbor matching algorithm1,d2,L,dmAfter the step is completed, the matching distance D is changed to { D } by adopting a quick sorting method1,d2,L,dmSorting from small to large, and selecting the first 500 matching points with smaller distance P ═ P1,p2,L,pnAnd fifthly, the smaller the matching distance is, the higher the matching degree is, and an effect diagram of a circuit board matching point with better matching degree is preliminarily screened out.
As shown in fig. 4, the effect diagram of calculating the slope and the distance between each pair of circuit board matching points in step S104 and screening the circuit board matching points by using the K-means clustering method includes:
step 4-1, clustering the slopes by using a K-means clustering method, finding out the most numerous circuit board matching points, discarding other circuit board matching points, and taking the most numerous circuit board matching points as excellent circuit board matching points;
and 4-2, performing secondary screening on the excellent circuit board matching points, clustering distances of the circuit board matching points by using a K-means clustering method, taking the most numerous circuit board matching points as final excellent circuit board matching points, and excluding other circuit board matching points.
In a preferred embodiment of the present invention, the obtaining of the transformation matrix in step S105 by using the RANSAC algorithm includes:
step 5-1, firstly, randomly selecting 4 pairs of matching points from the data set, and determining a transformation matrix Hi；
Step 5-2, then calculate each matching point pjMatching point p 'after transformation of matrix Hi'jTransforming the formula:
p′j＝Hpj
step 5-3, calculating the distance from each matching point to the corresponding matching point after the transformation matrix;
step 5-4, dividing the matching point set into inner points and outer points through a threshold value mu, and if the number of the inner points is large enough, HiReasonably, estimating a matrix H by using all the interior points, and stopping iteration, otherwise, continuing the 1-4 steps;
and 5-5, finally, if the iteration times exceed the preset maximum iteration times L, exiting.
As shown in fig. 5, the step S106 is to deform and fuse the input image into the output image, and complete the stitching to obtain the global circuit board image with ultrahigh resolution, including:
the input image is deformed and fused to a corresponding output image by using a matrix H, and 1/2 of the image overlapping region is taken as a smooth region XsFusing the smoothed regions X using a weighted averagesThe pixel point of (2) is used for smoothly splicing gaps and avoiding image large-area blurring, and the smoothing formula is as follows:
wherein wiIs the abscissa of the current point and the smooth area XsThe quotient of the widths is,
the pixel values of the channels of the image are stitched for the left side of the smooth region,
splicing pixel values of all channels of the image on the right side of the smooth area;
and firstly splicing the rows of the circuit board image, and then splicing the columns of the circuit board image to obtain the final global circuit board image with ultrahigh resolution.
As shown in fig. 6, it is a detail display diagram at the image stitching slit of the circuit board.
The technical solution of the present invention is further described with reference to the following specific examples.
In the embodiment of the invention, 72 local images of the high-definition circuit board are spliced, and the number of the local images is 6 rows and 12 columns.
In step S102, SIFT feature points on the circuit board grayscale image are extracted, the upper limit M of the number is 2000, and the feature points are matched by using a fast nearest neighbor matching algorithm.
In step S103, the circuit board matching points are sorted by fast sorting, and the preliminary screening of the circuit board matching points with better matching degree includes: obtaining distance D ═ D between circuit board matching points by using fast nearest neighbor matching algorithm1,d2,L,dmAfter the step is completed, the matching distance D is changed to { D } by adopting a quick sorting method1,d2,L,dmSorting from small to large, selecting the first N-500 matching points P-P with smaller distance1,p2,L,pnAnd fifthly, the smaller the matching distance is, the higher the matching degree is, and the circuit board matching point with better matching degree is preliminarily screened out.
In step S104, clustering the circuit board matching points with the calculated slope and distance by using a K-means clustering method, eliminating the matching points with poor matching, and randomly selecting K in the K-means clustering of the slope 015 data as centroids, precision range theta00.005; randomly selecting K in K-means clustering of distances 115 data as centroids, precision range theta1＝5。
In step S106, the input image is deformed and fused into a corresponding output image by using the transformation matrix H, and 1/2 of the image overlapping region is taken as a smooth region XsFusing the smoothed regions X using a weighted averagesThe pixel point of for the smooth joint gap with avoid the image large tracts of land fuzzy, from a left side to the right side respectively, from last order down, the concatenation x is 6 lines, and y is 12 high definition circuit board images of being listed as, acquires final complete circuit board image.
According to the invention, a high-definition camera is adopted to obtain a large number of high-resolution circuit board local images, the image fineness is increased, and the problem that excellent matching points are difficult to screen due to disordered matching of circuit board image feature points is solved, so that the splicing of a large number of high-definition circuit board images is accurately realized, and a good foundation is laid for subsequent work.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.
Claims (9)
1. A circuit board image splicing method for screening matching points based on a clustering method is characterized by comprising the following steps:
step 1: acquiring a plurality of circuit board local images with overlapping areas through a high-definition camera;
step 2: extracting SIFT feature points on the gray level image of the circuit board, and matching the feature points by using a rapid nearest neighbor matching algorithm;
and step 3: sorting the circuit board matching points by utilizing quick sorting, and preliminarily screening out the circuit board matching points with better matching degree;
and 4, step 4: calculating the slope and the distance between each pair of circuit board matching points, and screening the circuit board matching points by using a K-means clustering method;
and 5: acquiring a transformation matrix by using a RANSAC algorithm;
step 6: and deforming and fusing the input image into the output image to complete splicing, thereby obtaining the global circuit board image with ultrahigh resolution.
2. The circuit board image stitching method for screening matching points based on the clustering method as claimed in claim 1, wherein in the step 1, the image size is 2048 × 2048, one circuit board comprises 30-120 partial images, the image naming order is sequentially from left to right and from top to bottom, and is marked as x rows and y columns, and the overlapping area occupies 1/3 of the width of each image.
3. The circuit board image stitching method for screening matching points based on the clustering method as claimed in claim 1, wherein the step 2 comprises the following specific steps:
step 2-1: converting the three-channel color circuit board image into a single-channel gray circuit board image;
step 2-2: intercepting a right 1/2 part of a spliced image on the left side of the gray circuit board and a left 1/2 part of the spliced image on the right side of the gray circuit board;
step 2-3: respectively extracting feature points on the two intercepted circuit board images by utilizing an SIFT algorithm, wherein the upper limit of the number of the feature points is M;
step 2-4: and matching the characteristic points on the two circuit board images by using a fast nearest neighbor matching algorithm.
4. The circuit board image stitching method for screening matching points based on the clustering method as claimed in claim 1, wherein the step 3 comprises the following specific steps:
obtaining distance D ═ D between circuit board matching points by using fast nearest neighbor matching algorithm1,d2,L,dmAfter the step is completed, the matching distance D is changed to { D } by adopting a quick sorting method1,d2,L,dmSorting the N matching points from small to large, and selecting the first N matching points with smaller distance as P1,p2,L,pn}。
5. The circuit board image stitching method for screening matching points based on the clustering method as claimed in claim 1, wherein the step 4 comprises the following specific steps:
step 4-1: clustering the slopes by using a K-means clustering method, finding out the most numerous circuit board matching points, discarding other circuit board matching points, and taking the most numerous circuit board matching points as excellent circuit board matching points;
step 4-2: and performing secondary screening on the excellent circuit board matching points, clustering the distances of the circuit board matching points by using a K-means clustering method, taking the most numerous circuit board matching points as final excellent circuit board matching points, and excluding other circuit board matching points.
6. The circuit board image stitching method for screening matching points based on the clustering method as claimed in claim 5, wherein the clustering the slope by using the K-means clustering method in the step 4-1 comprises: calculating the slope of each pair of matching point connecting lines by using a slope calculation formula of the two point connecting lines, recording the slope as a data set of K-means clustering, and randomly selecting K0Taking the data as a centroid, calculating the distance from each data point to each centroid, distributing each point in the data set to a cluster corresponding to the closest centroid, updating the centroid of each cluster to the average value of all the points of the cluster, and judging whether the centroid of each cluster is at the precision theta0If the iteration times are not reached or not, the algorithm is ended if the iteration times are not reached, otherwise, the random selection k is returned0The data serves as the centroid and the iteration continues.
7. The circuit board image stitching method for screening matching points based on clustering method as claimed in claim 5,
the clustering the distances of the matching points by using the K-means clustering method in the step 4-2 comprises the following steps: calculating the distance between each pair of matching points by using a distance calculation formula between the two points, recording the distance as a data set of K-means clustering, and randomly selecting K1Taking the data as a centroid, calculating the distance from each data point to each centroid, distributing each point in the data set to a cluster corresponding to the closest centroid, updating the centroid of each cluster to the average value of all the points of the cluster, and judging whether the centroid of each cluster is at the precision theta1If the iteration times are not reached or not, the algorithm is ended if the iteration times are not reached, otherwise, the random selection k is returned1The data serves as the centroid and the iteration continues.
8. The circuit board image stitching method for screening matching points based on the clustering method as claimed in claim 1, wherein the step 5 comprises the following specific steps:
step 5-1: firstly, randomly selecting 4 pairs of matching points from a data set, and determining a transformation matrix Hi；
Step 5-2: then each matching point p is calculatedjMatching point p 'after transformation of matrix Hi'jTransforming the formula:
p′j＝Hpj
step 5-3: calculating the distance from each matching point to the corresponding matching point after the transformation matrix;
step 5-4: then the matching point set is divided into inner points and outer points through a threshold value mu, and H is carried out if the number of the inner points is large enoughiReasonably, estimating a matrix H by using all the interior points, and stopping iteration, otherwise, continuing the 1-4 steps;
step 5-5: and finally, if the iteration times exceed the preset maximum iteration times L, exiting.
9. The circuit board image stitching method for screening matching points based on the clustering method as claimed in claim 1, wherein the step 6 comprises the following specific steps:
deforming and fusing the input image to a corresponding output image by using the matrix H, and taking 1/2 of the image overlapping area as a smooth area XsFusing the smoothed regions X using a weighted averagesThe pixel point of (2) is used for smoothly splicing gaps and avoiding image large-area blurring, and the smoothing formula is as follows:
wherein wiIs the abscissa of the current point and the smooth area XsThe quotient of the widths is,
the pixel values of the channels of the image are stitched for the left side of the smooth region,
splicing pixel values of all channels of the image on the right side of the smooth area;
and firstly splicing the rows of the circuit board image, and then splicing the columns of the circuit board image to obtain the final global circuit board image with ultrahigh resolution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010111022.0A CN111340701B (en) | 2020-02-24 | 2020-02-24 | Circuit board image splicing method for screening matching points based on clustering method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010111022.0A CN111340701B (en) | 2020-02-24 | 2020-02-24 | Circuit board image splicing method for screening matching points based on clustering method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111340701A true CN111340701A (en) | 2020-06-26 |
| CN111340701B CN111340701B (en) | 2022-06-28 |
Family
ID=71181750
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010111022.0A Active CN111340701B (en) | 2020-02-24 | 2020-02-24 | Circuit board image splicing method for screening matching points based on clustering method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111340701B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112101183A (en) * | 2020-09-10 | 2020-12-18 | 深圳市商汤科技有限公司 | Vehicle identification method and device, electronic equipment and storage medium |
| CN112686302A (en) * | 2020-12-29 | 2021-04-20 | 科大讯飞股份有限公司 | Image feature point matching method, computer equipment and storage device |
| CN112767348A (en) * | 2021-01-18 | 2021-05-07 | 上海明略人工智能(集团)有限公司 | Method and device for determining detection information |
| CN113075648A (en) * | 2021-03-19 | 2021-07-06 | 中国舰船研究设计中心 | Clustering and filtering method for unmanned cluster target positioning information |
| CN113160048A (en) * | 2021-02-02 | 2021-07-23 | 重庆高新区飞马创新研究院 | Suture line guided image splicing method |
| CN113409251A (en) * | 2021-05-31 | 2021-09-17 | 广西格思克实业有限责任公司 | Data processing method for integrated circuit manufacturing |
| CN114022790A (en) * | 2022-01-10 | 2022-02-08 | 成都国星宇航科技有限公司 | Cloud layer detection and image compression method and device in remote sensing image and storage medium |
| CN116797463A (en) * | 2023-08-22 | 2023-09-22 | 佗道医疗科技有限公司 | Feature point pair extraction method and image stitching method |
| CN117670958A (en) * | 2024-01-31 | 2024-03-08 | 中国人民解放军国防科技大学 | Registration method, device and equipment for SAR image of small aperture sequence |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017193596A1 (en) * | 2016-05-13 | 2017-11-16 | 广州视源电子科技股份有限公司 | Image matching method and system for printed circuit board |
| CN107578376A (en) * | 2017-08-29 | 2018-01-12 | 北京邮电大学 | The fork division of distinguished point based cluster four and the image split-joint method of local transformation matrix |
| WO2019214568A1 (en) * | 2018-05-07 | 2019-11-14 | 清华大学深圳研究生院 | Depth-based light field splicing method |
| CN110544202A (en) * | 2019-05-13 | 2019-12-06 | 燕山大学 | parallax image splicing method and system based on template matching and feature clustering |
-
2020
- 2020-02-24 CN CN202010111022.0A patent/CN111340701B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017193596A1 (en) * | 2016-05-13 | 2017-11-16 | 广州视源电子科技股份有限公司 | Image matching method and system for printed circuit board |
| CN107578376A (en) * | 2017-08-29 | 2018-01-12 | 北京邮电大学 | The fork division of distinguished point based cluster four and the image split-joint method of local transformation matrix |
| WO2019214568A1 (en) * | 2018-05-07 | 2019-11-14 | 清华大学深圳研究生院 | Depth-based light field splicing method |
| CN110544202A (en) * | 2019-05-13 | 2019-12-06 | 燕山大学 | parallax image splicing method and system based on template matching and feature clustering |
Non-Patent Citations (6)
| Title |
|---|
| 张媛等: "运用特征点匹配的柱面全景图像快速拼接算法", 《中国图象图形学报》, no. 06, 15 June 2009 (2009-06-15) * |
| 张晓茹等: "基于角点匹配图像拼接方法的改进", 《河北农业大学学报》, no. 04, 15 July 2008 (2008-07-15) * |
| 穆柯楠: "基于特征点的图像拼接关键技术研究", 《中国优秀硕士学位论文全文数据库（电子期刊） 信息科技辑》 * |
| 穆柯楠: "基于特征点的图像拼接关键技术研究", 《中国优秀硕士学位论文全文数据库（电子期刊） 信息科技辑》, 15 February 2015 (2015-02-15), pages 4 - 5 * |
| 管士勇: "基于特征点的印刷电路板CT图像配准技术研究", 《中国优秀硕士学位论文全文数据库（电子期刊）信息科技辑》 * |
| 管士勇: "基于特征点的印刷电路板CT图像配准技术研究", 《中国优秀硕士学位论文全文数据库（电子期刊）信息科技辑》, 15 June 2013 (2013-06-15), pages 1 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112101183A (en) * | 2020-09-10 | 2020-12-18 | 深圳市商汤科技有限公司 | Vehicle identification method and device, electronic equipment and storage medium |
| CN112686302A (en) * | 2020-12-29 | 2021-04-20 | 科大讯飞股份有限公司 | Image feature point matching method, computer equipment and storage device |
| CN112686302B (en) * | 2020-12-29 | 2024-02-13 | 科大讯飞股份有限公司 | Image feature point matching method, computer equipment and storage device |
| CN112767348B (en) * | 2021-01-18 | 2023-11-24 | 上海明略人工智能(集团)有限公司 | Method and device for determining detection information |
| CN112767348A (en) * | 2021-01-18 | 2021-05-07 | 上海明略人工智能(集团)有限公司 | Method and device for determining detection information |
| CN113160048A (en) * | 2021-02-02 | 2021-07-23 | 重庆高新区飞马创新研究院 | Suture line guided image splicing method |
| CN113075648A (en) * | 2021-03-19 | 2021-07-06 | 中国舰船研究设计中心 | Clustering and filtering method for unmanned cluster target positioning information |
| CN113409251A (en) * | 2021-05-31 | 2021-09-17 | 广西格思克实业有限责任公司 | Data processing method for integrated circuit manufacturing |
| CN113409251B (en) * | 2021-05-31 | 2022-06-21 | 广西格思克实业有限责任公司 | Data processing method for integrated circuit manufacturing |
| CN114022790A (en) * | 2022-01-10 | 2022-02-08 | 成都国星宇航科技有限公司 | Cloud layer detection and image compression method and device in remote sensing image and storage medium |
| CN114022790B (en) * | 2022-01-10 | 2022-04-26 | 成都国星宇航科技有限公司 | Cloud layer detection and image compression method and device in remote sensing image and storage medium |
| CN116797463B (en) * | 2023-08-22 | 2023-11-21 | 佗道医疗科技有限公司 | Feature point pair extraction method and image stitching method |
| CN116797463A (en) * | 2023-08-22 | 2023-09-22 | 佗道医疗科技有限公司 | Feature point pair extraction method and image stitching method |
| CN117670958A (en) * | 2024-01-31 | 2024-03-08 | 中国人民解放军国防科技大学 | Registration method, device and equipment for SAR image of small aperture sequence |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111340701B (en) | 2022-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111340701B (en) | Circuit board image splicing method for screening matching points based on clustering method | |
| US8160366B2 (en) | Object recognition device, object recognition method, program for object recognition method, and recording medium having recorded thereon program for object recognition method | |
| US8666170B2 (en) | Computer system and method of matching for images and graphs | |
| CN111080529A (en) | Unmanned aerial vehicle aerial image splicing method for enhancing robustness | |
| CN106940876A (en) | A kind of quick unmanned plane merging algorithm for images based on SURF | |
| CN104751465A (en) | ORB (oriented brief) image feature registration method based on LK (Lucas-Kanade) optical flow constraint | |
| CN108961286B (en) | Unmanned aerial vehicle image segmentation method considering three-dimensional and edge shape characteristics of building | |
| US20170097944A1 (en) | Passenger vehicle make and model recognition system | |
| CN103902953B (en) | A kind of screen detecting system and method | |
| CN113052170B (en) | Small target license plate recognition method under unconstrained scene | |
| CN107862319B (en) | Heterogeneous high-light optical image matching error eliminating method based on neighborhood voting | |
| CN110569861A (en) | Image matching positioning method based on point feature and contour feature fusion | |
| CN109472770B (en) | Method for quickly matching image characteristic points in printed circuit board detection | |
| Ghassabi et al. | Colour retinal fundus image registration by selecting stable extremum points in the scale‐invariant feature transform detector | |
| CN111967337A (en) | Pipeline line change detection method based on deep learning and unmanned aerial vehicle images | |
| CN111369605A (en) | Infrared and visible light image registration method and system based on edge features | |
| CN110544202A (en) | parallax image splicing method and system based on template matching and feature clustering | |
| CN113793266A (en) | Multi-view machine vision image splicing method, system and storage medium | |
| CN116433733A (en) | Registration method and device between optical image and infrared image of circuit board | |
| CN110348307B (en) | Path edge identification method and system for crane metal structure climbing robot | |
| CN110929782A (en) | River channel abnormity detection method based on orthophoto map comparison | |
| CN113096016A (en) | Low-altitude aerial image splicing method and system | |
| CN110717910B (en) | CT image target detection method based on convolutional neural network and CT scanner | |
| CN104036494A (en) | Fast matching computation method used for fruit picture | |
| CN116883464A (en) | Registration method for large-viewing-angle difference optics and SAR remote sensing 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 |