CN106815832B - A kind of steel mesh automatic image registration method and system of surface mounting technology - Google Patents

A kind of steel mesh automatic image registration method and system of surface mounting technology Download PDF

Info

Publication number
CN106815832B
CN106815832B CN201611187643.7A CN201611187643A CN106815832B CN 106815832 B CN106815832 B CN 106815832B CN 201611187643 A CN201611187643 A CN 201611187643A CN 106815832 B CN106815832 B CN 106815832B
Authority
CN
China
Prior art keywords
registration
image
stencil
points
point
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201611187643.7A
Other languages
Chinese (zh)
Other versions
CN106815832A (en
Inventor
夏珉
杨克成
李微
章琦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huazhong University of Science and Technology
Original Assignee
Huazhong University of Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huazhong University of Science and Technology filed Critical Huazhong University of Science and Technology
Priority to CN201611187643.7A priority Critical patent/CN106815832B/en
Publication of CN106815832A publication Critical patent/CN106815832A/en
Application granted granted Critical
Publication of CN106815832B publication Critical patent/CN106815832B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G06T7/001—Industrial image inspection using an image reference approach
    • 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
    • G06T2207/30141—Printed circuit board [PCB]

Landscapes

  • Engineering & Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

本发明公开了一种表面贴装技术的钢网图像自动配准方法，属于图像处理与光学测量领域。该发明方法利用图像外接矩形和配准点间的距离不受钢网任意摆放位置和角度的影响的特型，利用设计配准点和设计点阵图和钢网成品图像进行配准，配准时将钢网与设计图像进行仿射变换配准所需要计算的平移、旋转、缩放3个参量的计算转换成了传输特征矩阵的计算，配准后进行配准自动测试，配准失败后进行二次配准。同时本发明还实现了一种表面贴装技术的钢网图像自动配准系统。本发明技术方案不受配准产品的摆放位置和摆放角度的影响，具有配准精确、快速和简单高效的特点，同时有效的提高了配准成功率。

The invention discloses a stencil image automatic registration method of surface mount technology, which belongs to the field of image processing and optical measurement. The inventive method utilizes the characteristic that the distance between the image circumscribed rectangle and the registration point is not affected by the arbitrary placement position and angle of the stencil, and uses the design registration point, the design bitmap and the finished stencil image for registration. The calculation of the three parameters of translation, rotation and scaling required for the affine transformation registration of the stencil and the design image is converted into the calculation of the transmission feature matrix. After the registration, the automatic registration test is performed. registration. At the same time, the invention also realizes a stencil image automatic registration system of surface mount technology. The technical scheme of the invention is not affected by the placement position and placement angle of the registration product, has the characteristics of accurate, fast, simple and efficient registration, and effectively improves the registration success rate.

Description

A kind of steel mesh automatic image registration method and system of surface mounting technology
Technical field
The invention belongs to image procossings and field of optical measurements, more particularly, to a kind of steel mesh of surface mounting technology Automatic image registration method.
Background technique
With the rapid development and progress of electronic technology, the integrated level and complexity of PCB circuit are also being continuously improved, patch Element has accounted for 50% to 90% in the component population of circuit board totality, and surface mounting technology (SMT) is current Electronic Assemblies A kind of most popular technology and technique in industry.Steel mesh is SMT particular manufacturing craft made of a kind of laser cutting, and major function is side Help the deposition of tin cream, it is therefore an objective to which the tin cream of accurate quantity is transferred to the accurate location on sky PCB.Now with processing technology with And the development of electronics industry is become better and better, the size of aperture is also smaller and smaller on SMT steel mesh, and quantity is more also at the increase of geometry Up to distribution of a pore sizes thousands of or even up to ten thousand from 0.1 micron to several centimeters.It is before this typically all by manually using putting Big mirror and other auxiliary tools with the naked eye carry out the quality in observation detection steel mesh hole one by one, and artificial vision's detection is a large amount of by consumption Time be difficult to ensure the accuracy of detection simultaneously.Therefore the automatic measurement technique based on machine vision will be the following steel mesh detection Inexorable trend.
Large format (600*600mm or more) steel mesh passes through single area array CCD and telecentric lens cooperate motor machine platform to carry out Snake scan imaging, image are spliced to form full width face image by precision and are registrated with vector design document by characteristic point, It calls detection algorithm to carry out defects detection and obtains quality inspection result.It is automatic and accurate how steel mesh image carries out with design document Before registration is the key that defects detection.Existing method for registering images usually needs the artificial characteristic point that clicks, and does not have automatic Property and high efficiency；Even if there is the method for autoregistration, existing autoregistration algorithm, which may be only available for steel mesh, to be put and lesser inclines Tiltedly and rotate.Be difficult to be suitable for steel mesh arbitrarily the stretcher strain of placement angle, steel mesh in process to registration success rate Influence, and do not have intelligent detection registration effect and Automatic Optimal adjustment is equipped with and improves the functions such as success rate.
Summary of the invention
In order to solve the above technological deficiency of the prior art and technical problem, the present invention provides a kind of surface mounting technologies Steel mesh automatic image registration method, its object is to choose matching characteristic point automatically；And the randomness of steel mesh placement angle is mentioned Reasonable usability methods are gone out；It joined internal automatic test matching effect and secondary registration.Thus existing steel mesh is solved to match The technical problems such as quasi- and defect inspection method low efficiency, poor for applicability and not smart enoughization.
To achieve the above object, according to one aspect of the present invention, a kind of steel mesh image of surface mounting technology is provided Autoegistration method, method includes the following steps:
(1) steel mesh design profile is obtained by steel mesh design document, seeks design profile minimum circumscribed rectangle, obtain rectangle four The coordinate on a vertex and the length of rectangle；
(2) area and barycentric coodinates of all intercommunicating pores in design profile are extracted, and generates design point according to barycentric coodinates System of battle formations picture；
(3) 4 design registration points successively are sought by 4 vertex of rectangle and design dot matrix image；
(4) finished product steel mesh high-contrast image is acquired, if finished product steel mesh is too big, the multiple topographies of steel mesh is acquired and carries out Splicing forms complete steel mesh image, reuses iterative method and carries out bilinear interpolation sub-pix Threshold segmentation processing image；It seeks locating Image-region minimum circumscribed rectangle and the coordinate on four vertex of rectangle and the length of rectangle are calculated after reason；
(5) area and barycentric coodinates in the steel mesh image connectivity hole after extraction process generates steel mesh by all barycentric coodinates Dot matrix image；
(6) 4 the first registration points successively are sought by 4 vertex of rectangle and steel mesh dot matrix image；
(7) it is found respectively in steel mesh dot matrix image away from 4 nearest focus points of 4 the first registration points as the second registration Point；
(8) 3 the first registration points for choosing 3 design registration points and corresponding position carry out affine transformation, and it is imitative to obtain feature Matrix is penetrated, design profile is covered on steel mesh image after converting by affine matrix；
(9) autoregistration test is carried out to the design profile converted by affine matrix, judges whether to be registrated successfully, if Then registration terminates；Otherwise the second registration point that failure registration point is changed to corresponding position is continued to be registrated.
Further, registration point finding process is divided into following sub-step in the step (3) and step (6):
(11) using a vertex of rectangle as the center of circle, with RdIt draws and justifies for radius, wherein Rd=kL, L are the length of rectangle, and k is First radius multiple；The value range of k is 5% < k < 20%, preferably k=10%；
(12) round and dot matrix image intersection is asked, judges whether there is focus point in intersection, has, obtain the center of gravity in intersection Coordinate, and execute step (34)；It is no to then follow the steps (33)；
(13) k=2k updates radius RdAnd circle is repainted, it executes step (32)；
(14) distance for seeking all barycentric coodinates in rectangle vertex and intersection takes apart from the smallest barycentric coodinates as registration Point.
Further, iterative method progress bilinear interpolation sub-pix Threshold segmentation processing image is specific in the step (4) It is divided into following sub-step:
(41) each pixel of the transition portion between steel mesh display foreground and background is selected to carry out 10 × 10 interpolation；
(42) threshold value that steel mesh image is acquired using iterative method carries out Threshold segmentation to the steel mesh image after interpolation.
Further, autoregistration test is divided into following sub-step in the step (9):
(91) the connection hole area and barycentric coodinates for extracting design profile after affine matrix converts, are given birth to by all barycentric coodinates At registration dot matrix image；
(92) using 3 selected the first registration points one of them as the center of circle, with RcIt draws and justifies for radius, wherein Rc=jl, j are Second radius multiple；The value range of j is 1% < j < 10%, preferably j=3%；L is lie farthest away in 4 the first registration points The distance between two registration points；The focus point coordinate in circle and registration dot matrix image intersection is acquired, the first intersection point set is denoted as It closes；The focus point coordinate in round and steel mesh dot matrix image intersection is acquired, the second intersection point set is denoted as；
(93) successively compare in the first intersection point set internal coordinate point and the second intersection point set corresponding position coordinate points it Between distance, distance be greater than maximum allowable offset coordinate points be denoted as registration failed point, if registration failed point number be greater than match Quasi- threshold value, then current the first registration point as the center of circle of judgement is registrated failure, and is denoted as unsuccessfully registration point；Otherwise judgement is current makees It is registrated successfully for first registration point in the center of circle；Wherein, registration threshold value value range is intersection point set internal coordinate point sum 30% to 50%, preferably the 40% of intersection point set internal coordinate point sum；The value range of maximum allowable offset arrives for 50um 150um, preferably 100um；
(94) it repeats step (92) and (93) first registration point selected by 3 and all completes autoregistration to test, if 3 institutes It selects the first registration point to be all registrated success, is then registrated success；Otherwise registration failure.
It is another aspect of this invention to provide that a kind of steel mesh automatic image registration system of surface mounting technology is provided, it should System includes following part:
Design profile boundary rectangle extraction module is sought designing for obtaining steel mesh design profile by steel mesh design document Profile minimum circumscribed rectangle obtains the coordinate on four vertex of rectangle and the length of rectangle；
Dot matrix image extraction module is designed, for extracting the area and barycentric coodinates of all intercommunicating pores in design profile, and Design dot matrix image is generated according to barycentric coodinates；
Registration point extraction module is designed, for successively being sought 4 designs by 4 vertex of rectangle and design dot matrix image and being matched On schedule；
Steel mesh image boundary rectangle extraction module, for acquiring finished product steel mesh high-contrast image, if finished product steel mesh is too big, It then acquires the multiple topographies of steel mesh and carries out the complete steel mesh image of splicing composition, reuse iterative method and carry out bilinear interpolation Asia picture Plain Threshold segmentation handles image；It seeks image-region minimum circumscribed rectangle after processing and calculates the coordinate and square on four vertex of rectangle The length of shape；
Steel mesh dot matrix image extraction module, for the area and barycentric coodinates in the steel mesh image connectivity hole after extraction process, Steel mesh dot matrix image is generated by all barycentric coodinates；
First registration point extraction module is matched for successively seeking 4 first by 4 vertex of rectangle and steel mesh dot matrix image On schedule；
Second registration point extraction module, for found respectively in steel mesh dot matrix image away from 4 the first registration points it is nearest 4 A focus point is as the second registration point；
Registration module, 3 the first registration points for choosing 3 design registration points and corresponding position carry out affine transformation, Feature affine matrix is obtained, design profile is covered on steel mesh image after converting by affine matrix；
It is registrated test module, for carrying out autoregistration test to the design profile converted by affine matrix, judgement is It is no to be registrated successfully, if then registration terminates；Otherwise the second registration point that failure registration point is changed to corresponding position is continued to be registrated.
Further, registration point finding process point in the design registration point extraction module and the first registration point extraction module For following submodule:
Dumpling made of glutinous rice flour module is drawn, for a vertex using rectangle as the center of circle, with RdIt draws and justifies for radius, wherein Rd=kL, L are square The length of shape, k are the first radius multiple；The value range of k is 5% < k < 20%, preferably k=10%；
Seek common ground submodule, for asking the intersection of round and dot matrix image, judges whether there is focus point in intersection, has, obtain Barycentric coodinates in intersection, and execute step (34)；It is no to then follow the steps (33)；
Dumpling made of glutinous rice flour module is updated, for updating k=2k, updates radius RdAnd circle is repainted, it executes step (32)；
Registration point submodule is sought, for seeking the distance of all barycentric coodinates in rectangle vertex and intersection, is taken apart from the smallest Barycentric coodinates are as registration point.
Further, iterative method carries out bilinear interpolation sub-pix threshold value in the steel mesh image boundary rectangle extraction module Dividing processing image is specifically divided into following submodule:
Interpolation submodule, for select the transition portion between steel mesh display foreground and background each pixel carry out 10 × 10 interpolation；
Threshold segmentation submodule, for acquiring the threshold value of steel mesh image using iterative method, to the steel mesh image after interpolation into Row threshold division.
Further, autoregistration test is divided into following submodule in the registration test module:
It is registrated dot matrix image and generates submodule, for extracting the connection hole area and again of design profile after affine matrix transformation Heart coordinate generates registration dot matrix image by all barycentric coodinates；
Be registrated intersection submodule, for using 3 selected the first registration points one of them as the center of circle, with RcFor radius picture Circle, wherein Rc=jl, j are the second radius multiple；The value range of j is 1% < j < 10%, preferably j=3%；L is 4 the The distance between two registration points of lie farthest away in one registration point；The focus point acquired in circle and registration dot matrix image intersection is sat Mark, is denoted as the first intersection point set；The focus point coordinate in round and steel mesh dot matrix image intersection is acquired, the second intersection point set is denoted as It closes；
Registration point judging submodule, for successively comparing in the first intersection point set internal coordinate point and the second intersection point set The distance between corresponding position coordinate points, the coordinate points that distance is greater than maximum allowable offset are denoted as registration failed point, if registration loses The number lost a little is greater than registration threshold value, then current the first registration point as the center of circle of judgement is registrated failure, and is denoted as and is unsuccessfully registrated Point；Otherwise current the first registration point as the center of circle of judgement is registrated successfully；Wherein, registration threshold value value range is all intersection points 30% to the 50% of sum, the 40% of preferably all intersection point sums；The value range of maximum allowable offset arrives for 50um 150um, preferably 100um；
It is registrated judging submodule, for repeating step registration intersection submodule and registration point judging submodule until 3 institutes It selects the first registration point all to complete autoregistration test, if the first registration point selected by 3 is all registrated success, is registrated success；Otherwise Registration failure.
In general, through the invention it is contemplated above technical scheme is compared with the prior art, have following technology special Sign and the utility model has the advantages that
(1) minimum circumscribed rectangle using figure of the invention can with arbitrary placement position and angle phase The variation answered, but registration features point and the constant property feature of the absolute minimum range in rectangle vertex, can effectively solve arbitrarily to revolve Turn the influence to registration effect.
(2) dividing between image and image with operation and the automatic search range that increases cleverly is utilized in the present invention Characteristic area greatly reduces the calculation amount and complexity for finding characteristic point.By using iterative method selected threshold and bilinearity The segmentation of interpolation sub-pix ensure that the precision and detection accuracy later of registration.It can be than classics under the premise of guaranteeing accuracy Sub-pixel Edge Detection faster completes contours extract.
(3) technical solution of the present invention in view of practical steel mesh in process there may be stretch and heating power deformation with Other influences for being difficult to expect, by testing the mechanism with adjust automatically registration point and secondary registration automatically, compared to existing Technology registration has higher success rate.
Detailed description of the invention
Fig. 1 is the method for the present invention flow chart；
Fig. 2 design profile file and its minimum circumscribed rectangle and 4 vertex schematic diagrames；
Fig. 3 is the minimum range registration point schematic diagram searched as the center of circle with dot chart intersection using the 1st vertex；
Fig. 4 is steel mesh image sub-pix Threshold segmentation schematic diagram；
Fig. 5 is the steel mesh image arbitrarily put and its minimum circumscribed rectangle and 4 vertex schematic diagrames；
Fig. 6 is the first registration point and potential interference the second registration point schematic diagram in image；
Fig. 7 is partial schematic diagram after steel mesh image is registrated with file；
Fig. 8 is that steel mesh image is registrated failure schematic diagram with file；
Fig. 9 be registrated unsuccessfully tested secondary registration automatically after success schematic diagram；
Figure 10 is that there are the schematic diagrames of hole location offset after steel mesh image is registrated with design document.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and It is not used in the restriction present invention.As long as in addition, technical characteristic involved in the various embodiments of the present invention described below Not constituting a conflict with each other can be combined with each other.
As shown in Figure 1, the method for the present invention process is as follows:
(1) it reads steel mesh and processes Gerber file, the information such as parsing hole site, shape size obtain steel mesh contour vector text Part is simultaneously shown as shown in Figure 2；
It seeks the minimum circumscribed rectangle of entire file area and rectangle upper left, upper right, lower-left, the top of bottom right four is calculated The coordinate of point and the length of rectangle, as shown in rectangle frame in Fig. 2 and 4 dots.The step of seeking the minimum circumscribed rectangle of image are as follows: Successively each column of scan image find the starting point coordinate (x1, y1) of each column target and terminal point coordinate (x2, y2) obtains in this column Barycentric coodinates be (x, y), calculate each column barycentric coodinates (xi, yi) using least square method carry out straight line fitting obtain horizontal master Axis linear equation y=k1x+b1.Row center of gravity and vertical major y=k2x+b2 similarly are asked to each row.Real center of gravity is horizontal main The intersection point of axis and vertical major.The constant downward translation in horizontal spindle direction is kept to find whether straight line intersects with target, last Lower boundary of the linear position of secondary intersection as boundary rectangle.It similarly can be in the hope of the upper and right boundary of boundary rectangle as just Beginning boundary rectangle.Then initial rectangular is rotated by certain direction of rotation and rotation angle interval, finds the smallest external square of area Shape is as preferred target.The mutually perpendicular four edges of rectangle are indicated with linear equation, and the intersection point of straight line is then external square 4 vertex of shape.
(2) intercommunicating pore is extracted according to the method for seeking center of gravity in step (1) to all connection bore regions of design profile file Area and center of gravity coordinate, and barycentric coodinates point generate a width design dot chart, as the dot in Fig. 3 show design Dot chart.
(3) using 4 vertex of rectangle as the center of circle, radius initial value is 10% long picture circle of rectangle, asks friendship with design dot chart Collection, and obtain the point of the barycentric coodinates in intersection.If one times of radius of circle automatic increase, duplicate picture circle asks friendship without intersection Collection is until having focus point in intersection.Barycentric coodinates are sought in intersection at a distance from current rectangle vertex, and take minimum range Corresponding coordinate points are a registration features point.4 registration points nearest with 4 vertex distances of rectangle are obtained, design is denoted as Registration point.As shown in figure 3, quadrant is the circle drawn using top left corner apex as the center of circle, cross corresponding point in figure For with top left corner apex apart from nearest registration point.
(4) bilinear interpolation sub-pix threshold value point is carried out using iterative method to the high contrast steel mesh image that acquisition has been spliced It cuts as shown in Figure 4.The transition portion between steel mesh display foreground and background is selected to carry out interpolation.Thus each pixel is carried out The interpolation of 10*10 carries out Threshold segmentation to the image after interpolation using the calculated threshold value of iterative method.Seek whole image region The length of interior minimum circumscribed rectangle and coordinate and rectangle that four vertex of rectangle are calculated.Such as rectangle frame in Fig. 5 and 4 dots It is shown,
(5) coordinate of the area to the image zooming-out hole after the segmentation of steel mesh sub-pix and center of gravity, all barycentric coodinates points The image for generating a width dot matrix, is denoted as steel mesh dot chart, as shown in Fig. 6 dot.
(6) 4 vertex of image boundary rectangle are the center of circle, and radius initial value is 10% long picture circle of rectangle, with steel mesh dot matrix Figure seeks common ground, and obtains the point of the barycentric coodinates in intersection.If without intersection, one times of radius of circle automatic increase, duplicate picture Circle seeks common ground until having intersection.Barycentric coodinates are sought in intersection at a distance from current rectangle vertex, and take minimum range institute Corresponding coordinate points are a registration features point.Obtain 4 first registration points nearest with 4 vertex distances of rectangle.In Fig. 6 The corresponding point of cross is with top left corner apex apart from nearest registration point.Steel mesh is put in the presence of randomness, and angle tilt is difficult to Avoid, the minimum circumscribed rectangle of image also can with the inclined degree put corresponding change, but most with rectangle vertex distance What the absolute property of small feature focus point was no variation in, thus the method can be adapted for the random inclination of placement angle.
(7) consider steel mesh during cutting processing as the number of open cell content and closeness increase can be cold due to heat expansion Contracting and stretching cause a degree of deformation of global shape.If having neighbor point and rectangle vertex distance near the first registration point It is very close with the first registration point minimum range, then it is backed up using neighbor point as potential interference point, as possible second Registration point.Conduct potential interference point as circle is irised out in Fig. 6 is backed up.
(8) coordinate for finding out 3 the first registration points of 3 design registration points and corresponding position carries out affine transformation, obtains Feature affine matrix.Design profile file by affine matrix convert after displacement, rotation angle and scaling all with steel mesh figure As consistent and coordinate unification.
(9) registration result is tested automatically.To the face of the design profile file drawing holes after step (8) affine transformation Long-pending and center of gravity coordinate, generates all barycentric coodinates points the image of the new dot matrix of one width, is denoted as registration dot chart.With step (8) 3 selected the first registration points are the center of circle in, and a length of radius of 3% rectangle bevel edge draws circle, acquire circle and registration dot matrix image Focus point coordinate in intersection is denoted as the first intersection point set；The focus point coordinate in round and steel mesh dot matrix image intersection is acquired, It is denoted as the second intersection point set；Corresponding position in the first intersection point set internal coordinate point and the second intersection point set is successively compared to sit The distance between punctuate, coordinate points of the distance greater than 100um are denoted as registration failed point, if the number of registration failed point is greater than intersection The 40% of point set internal coordinate point sum, then current the first registration point as the center of circle of judgement is registrated failure, and is denoted as failure and matches On schedule；Otherwise current the first registration point as the center of circle of judgement is registrated successfully；First registration point selected by 3 all completes autoregistration After test, if the first registration point selected by 3 is all registrated success, it is registrated success, as shown in Figure 8；If failure, as shown in fig. 7, First registration point of registration failure is changed to after the second registration point repeatedly autoregistration and tested and continues to be registrated.Images after registration office Portion is as shown in figure 9, obviously can see that steel mesh makes defect by registration result.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.

Claims (8)

1.一种表面贴装技术的钢网图像自动配准方法，其特征在于，该方法包括以下步骤：1. a stencil image automatic registration method of surface mount technology, is characterized in that, this method comprises the following steps: (1)由钢网设计文件得到钢网设计轮廓，求取设计轮廓最小外接矩形，得到矩形4个顶点的坐标和矩形的长；(1) Obtain the stencil design outline from the stencil design file, obtain the minimum circumscribed rectangle of the design outline, and obtain the coordinates of the four vertices of the rectangle and the length of the rectangle; (2)提取设计轮廓中所有连通孔的面积和重心坐标，并根据重心坐标生成设计点阵图像；(2) Extract the area and barycentric coordinates of all connected holes in the design outline, and generate a design lattice image according to the barycentric coordinates; (3)依次由矩形的4个顶点和设计点阵图像求取4个设计配准点；(3) Four design registration points are obtained from the four vertices of the rectangle and the design lattice image in turn; (4)采集成品钢网图像，再使用迭代法进行双线性插值亚像素阈值分割处理图像；求取处理后图像区域最小外接矩形并计算成品图像外接矩形4个顶点的坐标和矩形的长；(4) Collect the finished stencil image, and then use the iterative method to perform bilinear interpolation sub-pixel threshold segmentation to process the image; obtain the minimum circumscribed rectangle of the image area after processing and calculate the coordinates of the four vertices of the circumscribed rectangle of the finished image and the length of the rectangle; (5)提取处理后的钢网图像连通孔的面积和重心坐标，由所有重心坐标生成钢网点阵图像；(5) Extracting the area and barycentric coordinates of the connected holes in the processed stencil image, and generating a stencil lattice image from all the barycentric coordinates; (6)依次由成品图像外接矩形的4个顶点和钢网点阵图像求取4个第一配准点；(6) obtain 4 first registration points sequentially from 4 vertices of the rectangle circumscribing the finished image and the stencil lattice image; (7)在钢网点阵图像中分别找到距4个第一配准点最近的4个重心点作为第二配准点；(7) In the stencil lattice image, respectively find the 4 centroid points closest to the 4 first registration points as the second registration points; (8)选取3个设计配准点和对应位置的3个第一配准点进行仿射变换，得到特征仿射矩阵，设计轮廓通过仿射矩阵变换后覆盖在钢网图像上；(8) Selecting 3 design registration points and 3 first registration points at corresponding positions to carry out affine transformation to obtain a feature affine matrix, and the design outline is covered on the stencil image after being transformed by the affine matrix; (9)对通过仿射矩阵变换的设计轮廓进行自动配准测试，判断是否配准成功，若是则配准结束；否则将失败配准点更换为对应位置的第二配准点继续配准。(9) Perform an automatic registration test on the design contour transformed by the affine matrix to determine whether the registration is successful, and if so, the registration ends; otherwise, replace the failed registration point with the second registration point at the corresponding position to continue the registration. 2.根据权利要求1所述的一种表面贴装技术的钢网图像自动配准方法，其特征在于，所述步骤(3)和步骤(6)中配准点求取过程分为以下子步骤：2. the stencil image automatic registration method of a kind of surface mount technology according to claim 1, is characterized in that, in described step (3) and step (6), the registration point obtaining process is divided into following sub-steps : (11)以矩形的一个顶点为圆心，以Rd为半径画圆，其中，Rd＝kL，L为矩形的长，k为第一半径倍数；(11) Take a vertex of the rectangle as the center of the circle, and draw a circle with R d as the radius, wherein R d =kL, L is the length of the rectangle, and k is the multiple of the first radius; (12)求圆和点阵图像的交集，判断交集内是否有重心点，有则获取交集内的重心坐标，并执行步骤(14)；否则执行步骤(13)；(12) Find the intersection of the circle and the dot matrix image, determine whether there is a center of gravity in the intersection, and if so, obtain the coordinates of the center of gravity in the intersection, and execute step (14); otherwise, execute step (13); (13)k＝2k，更新半径Rd并重画圆，执行步骤(12)；(13) k=2k, update the radius R d and redraw the circle, and execute step (12); (14)求矩形顶点和交集内所有重心坐标的距离，取距离最小的重心坐标作为配准点。(14) Find the distance between the vertices of the rectangle and all the barycentric coordinates in the intersection, and take the barycentric coordinates with the smallest distance as the registration point. 3.根据权利要求1所述的一种表面贴装技术的钢网图像自动配准方法，其特征在于，所述步骤(4)中迭代法进行双线性插值亚像素阈值分割处理图像具体分为以下子步骤：3. the stencil image automatic registration method of a kind of surface mount technology according to claim 1, is characterized in that, in described step (4), iterative method carries out bilinear interpolation sub-pixel threshold segmentation processing image specific segmentation. for the following substeps: (41)选择钢网图像前景和背景之间的过渡部分的每个像素进行10×10的插值；(41) Select each pixel of the transition portion between the foreground and the background of the stencil image to perform 10×10 interpolation; (42)使用迭代法求得钢网图像的阈值，对插值后的钢网图像进行阈值分割。(42) The threshold value of the stencil image is obtained by an iterative method, and the threshold value is segmented on the stencil image after interpolation. 4.根据权利要求1所述的一种表面贴装技术的钢网图像自动配准方法，其特征在于，所述步骤(9)中自动配准测试分为以下子步骤：4. the stencil image automatic registration method of a kind of surface mount technology according to claim 1, is characterized in that, in described step (9), automatic registration test is divided into following sub-steps: (91)提取仿射矩阵变换后设计轮廓的连通孔面积和重心坐标，由所有重心坐标生成配准点阵图像；(91) extracting the connected hole area and barycentric coordinates of the design contour after affine matrix transformation, and generating a registration lattice image from all barycentric coordinates; (92)以所选的3个第一配准点其中一个为圆心，以Rc为半径画圆，其中，Rc＝jl，j为第二半径倍数；l为4个第一配准点中相距最远两个配准点之间的距离；求得圆和配准点阵图像交集内的重心点坐标，记为第一交集点集合；求得圆和钢网点阵图像交集内的重心点坐标，记为第二交集点集合；(92) Take one of the selected 3 first registration points as the center of the circle, and draw a circle with R c as the radius, where R c =jl, j is the second radius multiple; l is the distance between the 4 first registration points The distance between the two farthest registration points; find the coordinates of the center of gravity in the intersection of the circle and the registration lattice image, and record it as the first set of intersection points; find the coordinates of the center of gravity in the intersection of the circle and the lattice image of the steel mesh, record is the second intersection point set; (93)依次对比第一交集点集合内坐标点和第二交集点集合内对应位置坐标点之间的距离，距离大于最大允许偏差的坐标点记为配准失败点，若配准失败点的个数大于配准阈值，则判断当前作为圆心的第一配准点配准失败，并记为失败配准点；否则判断当前作为圆心的第一配准点配准成功；(93) Comparing the distances between the coordinate points in the first intersection point set and the corresponding position coordinate points in the second intersection point set in turn, the coordinate points whose distance is greater than the maximum allowable deviation are recorded as the registration failure points. If the number is greater than the registration threshold, it is judged that the registration of the first registration point currently serving as the center of the circle fails, and it is recorded as a failed registration point; otherwise, it is judged that the first registration point currently serving as the center of the circle is successfully registered; (94)重复步骤(92)和(93)直到3个所选第一配准点都完成自动配准测试，若3个所选第一配准点都配准成功，则配准成功；否则配准失败。(94) Repeat steps (92) and (93) until all three selected first registration points complete the automatic registration test, if all three selected first registration points are successfully registered, the registration is successful; otherwise, the registration is successful fail. 5.一种表面贴装技术的钢网图像自动配准系统，其特征在于，该系统包括以下部分：5. A stencil image automatic registration system of surface mount technology, characterized in that the system comprises the following parts: 设计轮廓外接矩形提取模块，用于由钢网设计文件得到钢网设计轮廓，求取设计轮廓最小外接矩形，得到矩形4个顶点的坐标和矩形的长；The design outline circumscribed rectangle extraction module is used to obtain the stencil design outline from the stencil design file, obtain the minimum circumscribed rectangle of the design outline, and obtain the coordinates of the four vertices of the rectangle and the length of the rectangle; 设计点阵图像提取模块，用于提取设计轮廓中所有连通孔的面积和重心坐标，并根据重心坐标生成设计点阵图像；Design a lattice image extraction module to extract the area and barycentric coordinates of all connected holes in the design outline, and generate a design lattice image according to the barycentric coordinates; 设计配准点提取模块，用于依次由矩形的4个顶点和设计点阵图像求取4个设计配准点；The design registration point extraction module is used to obtain 4 design registration points from the four vertices of the rectangle and the design lattice image in turn; 钢网图像外接矩形提取模块，用于采集成品钢网图像，再使用迭代法进行双线性插值亚像素阈值分割处理图像；求取处理后图像区域最小外接矩形并计算成品图像外接矩形4个顶点的坐标和矩形的长；The stencil image circumscribed rectangle extraction module is used to collect the finished stencil image, and then use the iterative method to perform bilinear interpolation sub-pixel threshold segmentation to process the image; find the minimum circumscribed rectangle of the processed image area and calculate the 4 vertices of the finished image circumscribed rectangle The coordinates of and the length of the rectangle; 钢网点阵图像提取模块，用于提取处理后的钢网图像连通孔的面积和重心坐标，由所有重心坐标生成钢网点阵图像；The stencil lattice image extraction module is used to extract the area and barycentric coordinates of the connected holes in the stencil image after processing, and generate a stencil lattice image from all the barycentric coordinates; 第一配准点提取模块，用于依次由成品图像外接矩形的4个顶点和钢网点阵图像求取4个第一配准点；The first registration point extraction module is used to sequentially obtain 4 first registration points from the 4 vertices of the circumscribed rectangle of the finished image and the stencil lattice image; 第二配准点提取模块，用于在钢网点阵图像中分别找到距4个第一配准点最近的4个重心点作为第二配准点；The second registration point extraction module is used to respectively find the 4 gravity center points closest to the 4 first registration points in the stencil lattice image as the second registration points; 配准模块，用于选取3个设计配准点和对应位置的3个第一配准点进行仿射变换，得到特征仿射矩阵，设计轮廓通过仿射矩阵变换后覆盖在钢网图像上；The registration module is used to select 3 design registration points and 3 first registration points at corresponding positions to perform affine transformation to obtain a feature affine matrix, and the design outline is covered on the stencil image after the affine matrix transformation; 配准测试模块，用于对通过仿射矩阵变换的设计轮廓进行自动配准测试，判断是否配准成功，若是则配准结束；否则将失败配准点更换为对应位置的第二配准点继续配准。The registration test module is used to perform an automatic registration test on the design contour transformed by the affine matrix, and judge whether the registration is successful, and if so, the registration is over; otherwise, the failed registration point is replaced with the second registration point at the corresponding position to continue the registration. allow. 6.根据权利要求5所述的一种表面贴装技术的钢网图像自动配准系统，其特征在于，所述设计配准点提取模块和第一配准点提取模块中配准点求取过程分为以下子模块：6. The stencil image automatic registration system of a surface mount technology according to claim 5, characterized in that, in the design registration point extraction module and the first registration point extraction module, the registration point extraction process is divided into: The following submodules: 画圆子模块，用于以矩形的一个顶点为圆心，以Rd为半径画圆，其中，Rd＝kL，L为矩形的长，k为第一半径倍数；The circle drawing submodule is used to draw a circle with a vertex of the rectangle as the center and R d as the radius, where R d =kL, L is the length of the rectangle, and k is a multiple of the first radius; 求交集子模块，用于求圆和点阵图像的交集，判断交集内是否有重心点，有则获取交集内的重心坐标，并执行求配准点子模块；否则执行更新圆子模块；The intersection sub-module is used to find the intersection of the circle and the dot matrix image, to determine whether there is a barycentric point in the intersection, and if so, obtain the barycentric coordinates in the intersection, and execute the registration point sub-module; otherwise, execute the update circle sub-module; 更新圆子模块，用于更新k＝2k，更新半径Rd并重画圆，执行求交集子模块；Update the circle submodule for updating k=2k, update the radius Rd and redraw the circle, and execute the intersection submodule; 求配准点子模块，用于求矩形顶点和交集内所有重心坐标的距离，取距离最小的重心坐标作为配准点。The sub-module for finding the registration point is used to find the distance between the vertex of the rectangle and all the barycentric coordinates in the intersection, and the barycentric coordinate with the smallest distance is taken as the registration point. 7.根据权利要求5所述的一种表面贴装技术的钢网图像自动配准系统，其特征在于，所述钢网图像外接矩形提取模块中迭代法进行双线性插值亚像素阈值分割处理图像具体分为以下子模块：7. The stencil image automatic registration system of a surface mount technology according to claim 5, wherein the iterative method in the stencil image circumscribed rectangle extraction module performs bilinear interpolation sub-pixel threshold segmentation processing Images are specifically divided into the following sub-modules: 插值子模块，用于选择钢网图像前景和背景之间的过渡部分的每个像素进行10×10的插值；The interpolation sub-module is used to select each pixel of the transition part between the foreground and background of the stencil image to perform 10×10 interpolation; 阈值分割子模块，用于使用迭代法求得钢网图像的阈值，对插值后的钢网图像进行阈值分割。The threshold segmentation sub-module is used to obtain the threshold value of the stencil image using the iterative method, and perform threshold segmentation on the interpolated stencil image. 8.根据权利要求5所述的一种表面贴装技术的钢网图像自动配准系统，其特征在于，所述配准测试模块中自动配准测试分为以下子模块：8. The stencil image automatic registration system of a surface mount technology according to claim 5, wherein the automatic registration test in the registration test module is divided into the following submodules: 配准点阵图像生成子模块，用于提取仿射矩阵变换后设计轮廓的连通孔面积和重心坐标，由所有重心坐标生成配准点阵图像；The registration lattice image generation sub-module is used to extract the connected hole area and the barycentric coordinates of the design contour after affine matrix transformation, and generate the registration lattice image from all the barycentric coordinates; 配准交集子模块，用于以所选的3个第一配准点其中一个为圆心，以Rc为半径画圆，其中，Rc＝jl，j为第二半径倍数；l为4个第一配准点中相距最远两个配准点之间的距离；求得圆和配准点阵图像交集内的重心点坐标，记为第一交集点集合；求得圆和钢网点阵图像交集内的重心点坐标，记为第二交集点集合；The registration intersection sub-module is used to draw a circle with one of the selected 3 first registration points as the center and R c as the radius, where R c =jl, j is the second radius multiple; l is the 4th The distance between the two registration points that are farthest apart in one registration point; obtain the coordinates of the center of gravity in the intersection of the circle and the registration lattice image, and record it as the first intersection point set; The coordinates of the center of gravity, recorded as the second set of intersection points; 配准点判断子模块，用于依次对比第一交集点集合内坐标点和第二交集点集合内对应位置坐标点之间的距离，距离大于最大允许偏差的坐标点记为配准失败点，若配准失败点的个数大于配准阈值，则判断当前作为圆心的第一配准点配准失败，并记为失败配准点；否则判断当前作为圆心的第一配准点配准成功；The registration point judgment sub-module is used to compare the distances between the coordinate points in the first intersection point set and the corresponding position coordinate points in the second intersection point set in turn, and the coordinate points whose distance is greater than the maximum allowable deviation are recorded as the registration failure points. If the number of registration failure points is greater than the registration threshold, it is judged that the registration of the first registration point currently serving as the center of the circle fails, and it is recorded as a failed registration point; otherwise, it is judged that the first registration point currently serving as the center of the circle is successfully registered; 配准判断子模块，用于重复配准交集子模块和配准点判断子模块直到3个所选第一配准点都完成自动配准测试，若3个所选第一配准点都配准成功，则配准成功；否则配准失败。The registration judgment sub-module is used to repeat the registration intersection sub-module and the registration point judgment sub-module until the three selected first registration points have completed the automatic registration test. If the three selected first registration points are all successfully registered, The registration is successful; otherwise, the registration fails.
CN201611187643.7A 2016-12-20 2016-12-20 A kind of steel mesh automatic image registration method and system of surface mounting technology Active CN106815832B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201611187643.7A CN106815832B (en) 2016-12-20 2016-12-20 A kind of steel mesh automatic image registration method and system of surface mounting technology

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201611187643.7A CN106815832B (en) 2016-12-20 2016-12-20 A kind of steel mesh automatic image registration method and system of surface mounting technology

Publications (2)

Publication Number Publication Date
CN106815832A CN106815832A (en) 2017-06-09
CN106815832B true CN106815832B (en) 2019-05-21

Family

ID=59109099

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201611187643.7A Active CN106815832B (en) 2016-12-20 2016-12-20 A kind of steel mesh automatic image registration method and system of surface mounting technology

Country Status (1)

Country Link
CN (1) CN106815832B (en)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108416773A (en) * 2018-03-08 2018-08-17 凌云光技术集团有限责任公司 A kind of the PCB aperture method of inspection and system of the over-determined systems based on registration point
CN108986181B (en) * 2018-06-15 2023-05-26 广东数相智能科技有限公司 Dot-based image processing method, device and computer readable storage medium
CN110766735B (en) 2019-10-21 2020-06-26 北京推想科技有限公司 Image matching method, device, equipment and storage medium
CN112861610B (en) * 2020-12-30 2025-01-10 深圳市路远智能装备有限公司 A visual recognition method for three-pack steel mesh
CN113242652B (en) * 2021-05-20 2022-04-01 上海望友信息科技有限公司 Spraying graph generation method and system, electronic equipment and storage medium
CN113888558B (en) * 2021-09-26 2024-05-28 武汉钢铁有限公司 Ladle slag surface image correction method
CN114429416B (en) * 2022-02-21 2024-06-21 合肥龙旗智能科技有限公司 Method, device, system and equipment for processing attachment of dial plate and bezel

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103177444A (en) * 2013-03-08 2013-06-26 中国电子科技集团公司第十四研究所 Automatic SAR (synthetic-aperture radar) image rectification method
CN103778626A (en) * 2013-12-31 2014-05-07 北京理工大学 Quick image registration method based on visual remarkable area
CN103854276A (en) * 2012-12-04 2014-06-11 株式会社东芝 Image registration device and method, image segmentation device and method and medical image device
CN104318568A (en) * 2014-10-24 2015-01-28 武汉华目信息技术有限责任公司 Method and system for image registration
CN104463866A (en) * 2014-12-04 2015-03-25 无锡日联科技有限公司 Local shape matching method based on outline random sampling
CN105389815A (en) * 2015-10-29 2016-03-09 武汉联影医疗科技有限公司 Breast image registration method and device

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103854276A (en) * 2012-12-04 2014-06-11 株式会社东芝 Image registration device and method, image segmentation device and method and medical image device
CN103177444A (en) * 2013-03-08 2013-06-26 中国电子科技集团公司第十四研究所 Automatic SAR (synthetic-aperture radar) image rectification method
CN103778626A (en) * 2013-12-31 2014-05-07 北京理工大学 Quick image registration method based on visual remarkable area
CN104318568A (en) * 2014-10-24 2015-01-28 武汉华目信息技术有限责任公司 Method and system for image registration
CN104463866A (en) * 2014-12-04 2015-03-25 无锡日联科技有限公司 Local shape matching method based on outline random sampling
CN105389815A (en) * 2015-10-29 2016-03-09 武汉联影医疗科技有限公司 Breast image registration method and device

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
基于机器视觉的SMT钢网质量检测技术研究;刘今越 等;《仪器仪表学报》;20090630;第30卷(第6期);第431-434页
基于轮廓特征的目标匹配算法研究;杨林杰;《中国优秀硕士学位论文全文数据库 信息科技辑》;20160815;第2016年卷(第8期);第I138-I340页

Also Published As

Publication number Publication date
CN106815832A (en) 2017-06-09

Similar Documents

Publication Publication Date Title
CN106815832B (en) A kind of steel mesh automatic image registration method and system of surface mounting technology
CN106340044B (en) Join automatic calibration method and caliberating device outside video camera
CN103047943B (en) Based on the door skin geomery detection method of single projection coded structured light
CN109754459B (en) Method and system for constructing human body three-dimensional model
CN106228528B (en) A Multi-focus Image Fusion Method Based on Decision Map and Sparse Representation
CN102496161B (en) Method for extracting contour of image of printed circuit board (PCB)
CN101493942B (en) Level set image segmentation method based on spacing moment
CN104359403A (en) Plane part size measurement method based on sub-pixel edge algorithm
CN109859226A (en) A kind of detection method of the X-comers sub-pix of figure segmentation
CN104751458B (en) A kind of demarcation angular-point detection method based on 180 ° of rotation operators
CN104376535A (en) A Sample-Based Fast Image Restoration Method
CN115170669A (en) Identification and positioning method and system based on edge feature point set registration and storage medium
CN108921864A (en) A kind of Light stripes center extraction method and device
CN110322403A (en) A kind of more supervision Image Super-resolution Reconstruction methods based on generation confrontation network
CN109727279B (en) Automatic registration method of vector data and remote sensing image
CN106296587B (en) Tire mold image stitching method
CN113920081A (en) A kind of tool wear detection method
CN115861274A (en) A Crack Detection Method Fused with 3D Point Cloud and 2D Image
CN111402330A (en) Laser line key point extraction method based on plane target
CN110807459A (en) License plate correction method and device and readable storage medium
CN105701776A (en) Lens distortion correcting method and system used for automatic optical detection
CN108182707A (en) Acquire it is imperfect under the conditions of gridiron pattern calibrating template and its automatic identifying method
CN106204564A (en) A kind of laser photocentre extracting method
CN109784297A (en) A 3D object recognition and optimal grasping method based on deep learning
CN114612412A (en) Processing method of three-dimensional point cloud data, application of processing method, electronic device and storage medium

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