CN112270329A - Accurate MARK point acquisition and identification algorithm based on multi-level algorithm fusion - Google Patents

Abstract

Translated fromChinese

本发明公开了一种基于多层次算法融合的精准MARK点采集和识别算法，S1、MARK点采集：选取一台相机并设置光源参数，通过相机来采集MARK图片，调节MARK轮廓搜寻的算法，直到MARK点的轮廓被清晰找出，当遍历了所有设定的MARK轮廓搜寻算法的参数都没有搜寻到MARK的清晰轮廓，就自动调节相机光源的预设参数和调整轮廓搜索视觉算法参数，本发明涉及SMT贴片机设备技术领域。该基于多层次算法融合的精准MARK点采集和识别算法，通过从MARK点采集开始就准备了MAKR识别时候的各种参数，包括相机光源，和轮廓搜索的参数，这样对抗了传统的MARK点识别受到光源等外界环境变化的影响，通过结合图像模板匹配率高和MARK点轮廓匹配精度高的优点，提高了MARK点识别的质量。

The invention discloses an accurate MARK point collection and recognition algorithm based on multi-level algorithm fusion. S1 and MARK point collection: select a camera and set light source parameters, collect MARK pictures through the camera, and adjust the MARK contour search algorithm until The contour of the MARK point is clearly found. When all the parameters of the set MARK contour search algorithm are traversed and no clear contour of the MARK is found, the preset parameters of the camera light source and the parameters of the contour search visual algorithm are automatically adjusted. The present invention It relates to the technical field of SMT placement machine equipment. This accurate MARK point collection and recognition algorithm based on multi-level algorithm fusion prepares various parameters for MAKR recognition, including camera light source and contour search parameters, from the beginning of MARK point collection, which resists traditional MARK point recognition. Affected by changes in the external environment such as light source, the quality of MARK point recognition is improved by combining the advantages of high image template matching rate and high matching accuracy of MARK point contour.

Description

Accurate MARK point acquisition and identification algorithm based on multi-level algorithm fusion

Technical Field

The invention relates to the technical field of SMT chip mounter equipment, in particular to an accurate MARK point acquisition and identification algorithm based on multi-level algorithm fusion.

Background

A mounter, also known as a "mounter" or a "surface mount system", is an apparatus that is disposed behind a dispenser or a screen printer and accurately places surface mount components on PCB pads by moving a mounting head in a production line. The method is divided into manual operation and full-automatic operation. The device is used for realizing high-speed and high-precision component placement, and is the most critical and complex device in the whole SMT and production. The chip mounter is a chip mounting device to be used in SMT production, and the chip mounter is developed from an early low-speed mechanical chip mounter to a high-speed optical centering chip mounter and is developed towards multifunctional and flexible connection modularization.

The chip mounter is in the first step of mounting chip component for the PCB and will position the true position of PCB, because there is slight deviation in the position of advancing the board each time, the chip mounter on the market basically all uses the way of MARK point identification, fixes the actual position and the angle deviation of PCB through two MARK points, two kinds of existing MARK point identification algorithms are specifically:

the first is an algorithm based on a template picture matching mode, a small template picture is directly stored when a picture is collected, and the template picture is used for directly matching the picture currently taken when the picture is actually matched. The disadvantage of this algorithm is that the matching accuracy is low, resulting in random overall shift of the mounted component position.

The second is based on finding dots of a specified size, since Mark points are mostly small dots. The disadvantage of this method is that the matching rate is low, and often no MARK point can be found, which results in reduced efficiency of production.

Disclosure of Invention

Aiming at the defects of the prior art, the invention provides an accurate MARK point acquisition and identification algorithm based on multi-level algorithm fusion, and solves the problems that the position of a mounted component is randomly and integrally deviated due to low identification rate and identification precision of MARK points in the prior art, and the production efficiency is reduced.

In order to achieve the purpose, the invention is realized by the following technical scheme: a multi-level algorithm fusion-based accurate MARK point acquisition and identification algorithm specifically comprises the following steps:

s1, MARK point acquisition: selecting a camera and setting light source parameters, acquiring a MARK picture through the camera, adjusting a MARK contour searching algorithm until the contour of a MARK point is clearly found out, automatically adjusting preset parameters of the camera light source and adjusting contour searching visual algorithm parameters when all set parameters of the MARK contour searching algorithm are traversed and the clear contour of the MARK is not searched, and searching the contour of the MARK point again until the clear contour of the MARK point is found, at the moment, saving the current light source and camera parameters, saving a template picture of the MARK point and contour searching parameters of the MARK point;

s2, matching and identifying MARK points: a1, in actual work, firstly moving a paster head of a chip mounter to be close to a MARK point, acquiring an image through a camera, judging whether the MARK point is close to a central point, performing rough matching by adopting a template image matching algorithm when the MARK point is not close to the central point, and obtaining XY deviation and moving if the matching is successful; a2, when the MARK point is close to the central point, adopting a MARK point contour matching recognition algorithm to perform accurate matching to obtain XY offset, and acquiring the image again to perform the next round of matching after moving the head.

Preferably, in step S2, if the XY offset finally converges to the tolerance range of the central point, the MARK point is considered to be recognized and the algorithm is ended.

Preferably, the template image matching algorithm in step S2 is a second matching error algorithm, where matching is performed twice in the second matching error algorithm, and the first matching is a rough matching. Taking interlaced array data of the template, namely quarter of the template data, carrying out interlaced array scanning matching on the searched image, namely matching in a quarter of the range of the original image, wherein the matching speed is obviously improved and the error threshold value E is greatly reduced due to the fact that the data volume is greatly reduced₀：

Wherein e₀The average maximum error of each point is generally 40-50, m and n are respectively the length and width of the template, and the second matching is precise matching. At the 1 st error minimum point (i)_min，j_min) In the field of (i), i.e. at a diagonal point of (i)_min-1，j_min-1)，(i_min+1，j_min+1) And searching and matching are carried out in the rectangle to obtain the final result. Preferably, in step S2, a MARK point template image matching algorithm and a MARK point contour matching algorithm are hierarchically fused, and dynamic matching and convergence are performed to the MARK center point.

Preferably, the algorithm needs to be integrated into the software of a control system of an upper computer of the chip mounter for use.

Preferably, before the system is used, a MARK point setting page needs to be opened through a software man-machine interaction interface of a chip mounter control system, an initial light source camera parameter is adjusted, and therefore a button for starting customizing a corresponding function of a MARK point is clicked.

Advantageous effects

The invention provides an accurate MARK point acquisition and identification algorithm based on multi-level algorithm fusion. Compared with the prior art, the method has the following beneficial effects:

(1) the accurate MARK point acquisition and recognition algorithm based on multi-level algorithm fusion is characterized in that various parameters including a camera light source and a contour search parameter during the identification of the MAKR are prepared from the beginning of MARK point acquisition, so that the influence of external environment changes such as the light source on the traditional MARK point recognition is resisted.

(2) The accurate MARK point acquisition and identification algorithm based on multi-level algorithm fusion improves the quality of MARK point identification, improves the production efficiency and also improves the mounting precision by combining the advantages of high matching rate of image templates and high matching precision of MARK point profiles.

(3) According to the accurate MARK point acquisition and identification algorithm based on multi-level algorithm fusion, the corresponding algorithm is used for searching and matching when identification is carried out in the MARK point acquisition stage, so that a user can quickly know whether the set MARK point is effective or not in the acquisition stage, and the existing method usually knows whether the initially set MARK point is effective or not at all in the real production time, so that the algorithm can more efficiently and quickly find out the reasonable MARK point compared with the existing method.

Drawings

FIG. 1 is a flow chart of the MARK point acquisition system framework of the present invention;

FIG. 2 is a flow chart of the MARK point acquisition algorithm body of the present invention;

FIG. 3 is a flow chart of the MARK point identification system framework of the present invention;

FIG. 4 is a flowchart of the main body of the MARK point matching identification algorithm of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Referring to fig. 1-4, the present invention provides a technical solution: a multi-level algorithm fusion-based accurate MARK point acquisition and identification algorithm specifically comprises the following steps:

s1, MARK point acquisition: selecting a camera and setting light source parameters, acquiring a MARK picture through the camera, adjusting a MARK contour searching algorithm until the contour of a MARK point is clearly found out, automatically adjusting preset parameters of the camera light source and adjusting contour searching visual algorithm parameters when all set parameters of the MARK contour searching algorithm are traversed and the clear contour of the MARK is not searched, and searching the contour of the MARK point again until the clear contour of the MARK point is found, at the moment, saving the current light source and camera parameters, saving a template picture of the MARK point and contour searching parameters of the MARK point;

s2, matching and identifying MARK points: a1, in actual work, firstly moving a paster head of a chip mounter to be close to a MARK point, acquiring an image through a camera, judging whether the MARK point is close to a central point, performing rough matching by adopting a template image matching algorithm when the MARK point is not close to the central point, and obtaining XY deviation and moving if the matching is successful; a2, when the MARK point is close to the central point, using the MARK point contour matching recognition algorithm to perform accurate matching to obtain XY offset, moving the head, and then collecting the image again to perform the next round of matching. Meanwhile, when one algorithm cannot be identified under certain conditions, the other algorithm is automatically switched to, and the robustness of the identification algorithm is guaranteed.

In the present invention, if the XY deviation finally converges within the tolerance range of the central point in step S2, it is determined that the MARK point has been identified, and the algorithm ends.

In the invention, the template image matching algorithm in the step S2 is a secondary matching error algorithm, the matching in the secondary matching error algorithm is carried out twice, and the first matching is rough matching. Taking interlaced array data of the template, namely quarter of the template data, carrying out interlaced array scanning matching on the searched image, namely matching in a quarter of the range of the original image, wherein the matching speed is obviously improved and the error threshold value E is greatly reduced due to the fact that the data volume is greatly reduced⁰：

Wherein e₀The average maximum error of each point is generally 40-50, m and n are respectively the length and width of the template, and the second matching is precise matching. At the 1 st error minimum point (i)_min，j_min) In the field of (i), i.e. at a diagonal point of (i)_min-1，j_min-1)，(i_min+1，j_min+1) to obtain the final result.

In the invention, a MARK point template picture matching algorithm and a MARK point contour matching algorithm are hierarchically fused in step S2, and dynamic matching and convergence are carried out to the MARK center point.

In the invention, the algorithm needs to be integrated in the upper computer control system software of the chip mounter for use.

In the invention, before the system is used, a MARK point setting page is opened through a software man-machine interaction interface of a chip mounter control system, and an initial light source camera parameter is adjusted, so that a button for starting customizing a corresponding function of the MARK point is clicked.

And those not described in detail in this specification are well within the skill of those in the art.

When the system is used, a MARK point setting page is opened through a man-machine interaction interface of a chip mounter control system software, an initial light source camera parameter is adjusted, a button corresponding to the MARK point is clicked to start customizing, the system can search the MARK profile, the clear MARK profile is known to stop, corresponding algorithm parameters, camera light source parameters, coordinate information and template pictures are stored, when the system is in actual production work, the head is attached to the position close to the MARK point coordinate, algorithm matching is carried out, the algorithm comprises multi-stage fusion of a template picture matching algorithm and a profile matching algorithm, and finally the position of the current MARK point is accurately located.

It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.

Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.

Claims (6)

Translated fromChinese

1.一种基于多层次算法融合的精准MARK点采集和识别算法，其特征在于：具体包括以下步骤：1. an accurate MARK point collection and identification algorithm based on multi-level algorithm fusion, is characterized in that: specifically comprise the following steps:S1、MARK点采集：选取一台相机并设置光源参数，通过相机来采集MARK图片，调节MARK轮廓搜寻的算法，直到MARK点的轮廓被清晰找出，当遍历了所有设定的MARK轮廓搜寻算法的参数都没有搜寻到MARK的清晰轮廓，就自动调节相机光源的预设参数和调整轮廓搜索视觉算法参数，再次进行MARK点的轮廓搜寻，直到找到MARK点的清晰轮廓，此时，保存当前的光源和相机参数，保存MARK点的模板图片，和MARK点的轮廓搜索参数；S1. MARK point collection: Select a camera and set the light source parameters, collect MARK pictures through the camera, and adjust the MARK contour search algorithm until the contour of the MARK point is clearly found. When all the set MARK contour search algorithms are traversed If no clear outline of MARK is found with all parameters, it will automatically adjust the preset parameters of the camera light source and the parameters of the contour search visual algorithm, and perform the outline search of the MARK point again until a clear outline of the MARK point is found. At this time, save the current Light source and camera parameters, save the template image of the MARK point, and the contour search parameters of the MARK point;S2、MARK点的匹配识别：a1、在实际工作中，首先贴片机贴头运动到MARK点附近，通过相机采集图像，并判断是否已经临近中心点，当MARK点没有临近中心点时，采用模板图像匹配算法，进行粗匹配，匹配如果成功，得到XY偏移，并移动；a2、当MARK点已经临近中心点时，采用MARK点轮廓匹配识别算法，进行精确匹配，得到XY的偏移，移动贴头之后，再次采集图像进行下一轮的匹配。S2. Matching and identification of MARK points: a1. In actual work, first the placement machine sticks to the vicinity of the MARK point, collects images through the camera, and judges whether it is close to the center point. When the MARK point is not close to the center point, use The template image matching algorithm performs rough matching. If the matching is successful, the XY offset is obtained and moved; a2. When the MARK point is close to the center point, the MARK point contour matching and recognition algorithm is used to perform precise matching to obtain the XY offset, After moving the sticker, collect images again for the next round of matching.2.根据权利要求1所述的一种基于多层次算法融合的精准MARK点采集和识别算法，其特征在于：所述步骤S2中如果XY偏移最终收敛在中心点容差范围之内，则认为MARK点已经识别完成，算法结束。2. a kind of accurate MARK point collection and identification algorithm based on multi-level algorithm fusion according to claim 1, is characterized in that: in described step S2, if XY offset finally converges within the center point tolerance range, then It is considered that the MARK point has been identified, and the algorithm ends.3.根据权利要求1所述的一种基于多层次算法融合的精准MARK点采集和识别算法，其特征在于：所述步骤S2中模板图像匹配算法为二次匹配误差算法，二次匹配误差算法中匹配分两次进行，第一次匹配是粗略匹配。取模板的隔行隔列数据，即四分之一的模板数据，在被搜索图上进行隔行隔列扫描匹配，即在原图的四分之一范围内匹配，由于数据量大幅度减少，匹配速度显著提高，误差阈值E₀：

其中e₀为各点平均的最大误差，一般取40-50即可，m和n分别为模板的长和宽，第二次匹配是精确匹配。在第1次误差最小点(i_min，j_min)的领域内，即在对角点为(i_min-1，j_min-1)，(i_min+1，j_min+1)的矩形内进行搜索匹配，得到最后结果。3. a kind of accurate MARK point collection and identification algorithm based on multi-level algorithm fusion according to claim 1, is characterized in that: in described step S2, template image matching algorithm is secondary matching error algorithm, secondary matching error algorithm The middle match is performed twice, and the first match is a rough match. Take the interlaced row and column data of the template, that is, a quarter of the template data, and perform interlaced and interlaced scanning matching on the searched image, that is, matching within a quarter of the original image. Due to the large reduction in the amount of data, the matching speed is reduced. Significantly improved, the error threshold E₀ :

Among them, e₀ is the average maximum error of each point, generally 40-50, m and n are the length and width of the template, respectively, and the second matching is exact matching. In the field of the first error minimum point (i_min , j_min ), that is, in the rectangle whose diagonal points are (i_min -1, j_min -1), (i_min +1, j_min +1) Do a search match to get the final result.4.根据权利要求1所述的一种基于多层次算法融合的精准MARK点采集和识别算法，其特征在于：所述步骤S2中分级融合了MARK点模板图片匹配算法和MARK点轮廓匹配算法，动态匹配和收敛至MARK中心点。4. a kind of accurate MARK point collection and identification algorithm based on multi-level algorithm fusion according to claim 1, it is characterized in that: in described step S2, hierarchical fusion MARK point template picture matching algorithm and MARK point contour matching algorithm, Dynamic matching and convergence to the MARK center point.5.根据权利要求1所述的一种基于多层次算法融合的精准MARK点采集和识别算法，其特征在于：所述该算法需要集成在贴片机上位机控制系统软件中使用。5. A kind of accurate MARK point acquisition and identification algorithm based on multi-level algorithm fusion according to claim 1, is characterized in that: described this algorithm needs to be integrated in the placement machine upper computer control system software to use.6.根据权利要求1所述的一种基于多层次算法融合的精准MARK点采集和识别算法，其特征在于：所述在使用前需通过贴片机控制系统软件人机交互界面打开MARK点设置页面，调整好一个初始的光源相机参数，点击开始定制MARK点相应功能按钮。6. a kind of accurate MARK point collection and identification algorithm based on multi-level algorithm fusion according to claim 1, is characterized in that: described before use, need to open MARK point setting by placement machine control system software man-machine interface page, adjust an initial light source camera parameters, and click the button to start customizing the corresponding function of the MARK point.

Images