WO2018068415A1 - 元件错件检测方法和系统 - Google Patents

元件错件检测方法和系统 Download PDF

Info

Publication number
WO2018068415A1
WO2018068415A1 PCT/CN2016/113146 CN2016113146W WO2018068415A1 WO 2018068415 A1 WO2018068415 A1 WO 2018068415A1 CN 2016113146 W CN2016113146 W CN 2016113146W WO 2018068415 A1 WO2018068415 A1 WO 2018068415A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
feature
pixel
component
tested
Prior art date
Application number
PCT/CN2016/113146
Other languages
English (en)
French (fr)
Inventor
李红匣
Original Assignee
广州视源电子科技股份有限公司
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 广州视源电子科技股份有限公司 filed Critical 广州视源电子科技股份有限公司
Publication of WO2018068415A1 publication Critical patent/WO2018068415A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/22Matching criteria, e.g. proximity measures
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection
    • G06T7/0004Industrial image inspection
    • G06T7/001Industrial image inspection using an image reference approach
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/70Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75Organisation 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
    • G06V10/751Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30108Industrial image inspection
    • G06T2207/30141Printed circuit board [PCB]

Definitions

  • the present invention relates to the field of automatic optical detection technology, and in particular, to a method and system for detecting a component wrong component.
  • AOI Automatic Optic Inspection
  • common defect detection includes missing parts detection, wrong part detection, reverse part detection, multi-piece detection, and the like.
  • the wrong component detection refers to extracting the features of the component to be detected and comparing with the template to determine whether the components inserted into the circuit board are correct.
  • the wrong component detection of components is mainly detected by manual, but this detection method is inefficient, and the detection result is easy to make mistakes, and the detection accuracy is low.
  • a method for detecting a component wrong component includes the following steps:
  • the component to be tested is determined to be in error.
  • a component fault detection system includes:
  • a positioning module configured to acquire an original image of the device to be tested on the circuit board, and locate a feature region image of the device to be tested from the original image; wherein the feature region image includes feature information of the device to be tested The feature information is used to distinguish the device under test from other components;
  • a comparison module configured to compare pixel values of respective pixels in the feature area image with pixel values of corresponding pixel points in the pre-stored feature area template image, and acquire the feature area image and the feature area template image Pixel similarity
  • the determining module is configured to determine that the component to be tested is faulty if the pixel similarity is less than a preset similarity threshold.
  • the method and system for detecting the wrong component of the above component by detecting the feature region of the component to be tested, when the pixel similarity between the feature region image and the feature region template image is small, determining the wrong component of the component to be tested, and realizing the automation of component fault detection. Can effectively improve detection efficiency and accuracy.
  • 1 is a flow chart of a method for detecting a defective component of an embodiment
  • Figure 3 is a corrected image of the character area
  • FIG. 4 is a schematic structural view of a component misdetection detecting system of one embodiment.
  • the component error detecting method may include the following steps:
  • S1 acquiring an original image of the device to be tested on the circuit board, and positioning a feature region image of the device to be tested from the original image; wherein the feature region image includes feature information of the device to be tested, Feature information is used to distinguish the device under test from other components;
  • the features of the present invention may include information such as the color and shape of the component to be tested and character information on the body region of the component to be tested, such as to facilitate distinguishing the component under test from other components.
  • the character information may be a character, a symbol, a pattern, or the like.
  • the original image of the device to be tested may be first located from the image of the entire circuit board, and the image of the feature region of the device to be tested is located from the original image.
  • the original images of the components to be tested are respectively obtained, and the image of the feature region of each component to be tested is respectively located from each original image.
  • each test is to be tested.
  • the feature area images corresponding to the components are sequentially stored to facilitate the execution of subsequent detection operations.
  • each feature area image may also be sequentially numbered to facilitate execution of subsequent detection operations.
  • the feature area image may be enlarged before being compared.
  • the image of the feature area may be subjected to noise reduction processing.
  • the feature region image may be subjected to gradation processing to obtain a grayscale image, and the grayscale image is binarized according to a preset pixel threshold.
  • the grayscale processing shown can be performed according to the following formula:
  • R, G, and B are the three color components of the RGB color space, respectively, and Gray is the binarized gray value.
  • the binarization may set a pixel point whose gradation value is greater than a preset gradation threshold value to a certain gradation value, and set a pixel point whose pixel value is less than or equal to a preset gradation threshold value to another gradation value.
  • the grayscale threshold may be a grayscale value that maximizes a function value of the following objective function:
  • g(t) ⁇ 0 *( ⁇ 0 - ⁇ ) 2 + ⁇ 1 *( ⁇ 1 - ⁇ ) 2 ;
  • ⁇ 0 is the ratio of the pixel points corresponding to the feature information in the feature area image
  • ⁇ 0 is the mean value of the pixel values of the pixel points corresponding to the feature information
  • ⁇ 1 is the pixel point of the background image
  • the ratio in the feature area image, ⁇ 1 is the mean value of the pixel values of the pixel points of the background image.
  • the image noise reduction processing can reduce the noise interference, it does not completely eliminate the noise interference.
  • the noise is fine and discrete in the image, and the character area is continuous. Therefore, in order to more accurately locate the character region, the binarized image can be subjected to morphological processing, the character regions are connected together to form an area, and contour extraction is performed, and the portion with the largest contour area is the text region.
  • the contour extracted character area image is shown in Fig. 2.
  • the image can be corrected. Specifically, after performing contour extraction on the morphologically processed feature region image, a minimum matrix fit may be performed on the contour image to obtain a fitted image; and coordinate values of three vertices in the fitted image are obtained; Calculating a transfer matrix for rotating the contour image according to the coordinate value and the pre-stored original coordinate value; performing coordinate transformation on the pixel point on the contour image according to the transfer matrix; setting the coordinate image of the coordinate transformation to Feature area image.
  • the transfer matrix can be recorded as:
  • Coordinate transformation can be performed on points on the image of the feature area according to the following formula:
  • (x, y) is the coordinate value of the pixel point on the image extracted by the contour before the coordinate transformation
  • (x', y') is the coordinate value of the pixel point on the image extracted by the contour after the coordinate transformation.
  • the coordinates of the upper left corner, the lower left corner, and the upper right corner of the rectangle are (x 1 , y 1 ), (x 2 , y 2 ), (x 3 , y 3 ), respectively, and the coordinates of the corrected rectangle are (0, 0). ), (0, h), (w, 0), where w and h represent the width and height of the corrected rectangle, and in order to ensure that the corrected rectangle is similar or identical to the original image, it should satisfy:
  • (x, y) represents a coordinate value of a pixel point in the image of the feature area
  • (x+d, y+d) represents a coordinate value in the image of the feature area template image and the feature area image
  • I(x, y) represents the pixel value of the pixel point whose coordinate value is (x, y) in the image of the feature area
  • M(x+d, y+d) represents a pixel value of a pixel point whose coordinate value is (x+d, y+d) in the feature region template image
  • C represents the similarity.
  • each feature region template image may be stored locally in advance, and the storage order may be set to be the same or corresponding to the storage order of the respective feature region images.
  • each feature region template image may be numbered, and the number may be set to be the same as or corresponding to the storage order of the respective feature region images.
  • the pixel similarity between the feature area image of the device to be tested and the reference image of the feature area is less than the preset similarity threshold, it indicates that the feature area image of the device to be tested differs greatly from the character area of the reference image of the feature region. Therefore, the wrong component of the device to be tested can be determined; otherwise, if the pixel similarity between the feature region image of the device to be tested and the feature region reference image is greater than or equal to a preset similarity threshold, the feature region image of the device to be tested is indicated. It is similar to the character area of the feature area reference image, so that it can be determined that the device under test is not wrong.
  • the similarity threshold may be set according to actual conditions. Generally, the greater the value of the similarity threshold, the higher the detection accuracy.
  • the component fault detection method of the invention realizes the automation of the component fault detection, and can effectively improve the detection efficiency and accuracy.
  • the color and/or shape characteristics of the elements are relatively similar, it is possible to distinguish whether or not the similarity is obtained by extracting the character information on the element to be tested and comparing it with the template element.
  • the present invention also provides a component wrong component detecting system.
  • the component error detection system may include:
  • a positioning module 10 configured to acquire an original image of the device to be tested on the circuit board, and locate a feature area image of the device to be tested from the original image; wherein the feature area image includes characteristics of the device to be tested Information, the feature information is used to distinguish the device under test from other components;
  • the features of the present invention may include information such as the color and shape of the component to be tested and character information on the body region of the component to be tested, such as to facilitate distinguishing the component under test from other components.
  • the character information may be a character, a symbol, a pattern, or the like.
  • the original image of the device to be tested may be first located from the image of the entire circuit board, and the image of the feature region of the device to be tested is located from the original image.
  • the original images of the components to be tested are respectively obtained, and the image of the feature region of each component to be tested is respectively located from each original image.
  • each test is to be tested.
  • the feature area images corresponding to the components are sequentially stored to facilitate the execution of subsequent detection operations.
  • each feature area image may also be sequentially numbered to facilitate execution of subsequent detection operations.
  • the comparison module 20 is configured to compare the pixel values of the pixels in the feature area image with the pixel values of the corresponding pixel points in the pre-stored feature area template image, and acquire the feature area image and the feature area template image. Pixel similarity;
  • the feature area image may be enlarged before being compared.
  • the image of the feature area may be subjected to noise reduction processing.
  • the feature region image may be subjected to gradation processing to obtain a grayscale image, and the grayscale image is binarized according to a preset pixel threshold.
  • the grayscale processing shown can be performed according to the following formula:
  • R, G, and B are the three color components of the RGB color space, respectively, and Gray is the binarized gray value.
  • the binarization may set a pixel point whose gradation value is greater than a preset gradation threshold value to a certain gradation value, and set a pixel point whose pixel value is less than or equal to a preset gradation threshold value to another gradation value.
  • the grayscale threshold may be a grayscale value that maximizes a function value of the following objective function:
  • g(t) ⁇ 0 *( ⁇ 0 - ⁇ ) 2 + ⁇ 1 *( ⁇ 1 - ⁇ ) 2 ;
  • ⁇ 0 is the ratio of the pixel points corresponding to the feature information in the feature area image
  • ⁇ 0 is the mean value of the pixel values of the pixel points corresponding to the feature information
  • ⁇ 1 is the pixel point of the background image
  • the ratio in the feature area image, ⁇ 1 is the mean value of the pixel values of the pixel points of the background image.
  • the image noise reduction processing can reduce the noise interference, it does not completely eliminate the noise interference.
  • the noise is fine and discrete in the image, and the character area is continuous. Therefore, in order to more accurately locate the character region, the binarized image can be subjected to morphological processing, the character regions are connected together to form an area, and contour extraction is performed, and the portion with the largest contour area is the text region.
  • the contour extracted character area image is shown in Fig. 2.
  • the image can be corrected. Specifically, after performing contour extraction on the morphologically processed feature region image, a minimum matrix fit may be performed on the contour image to obtain a fitted image; and coordinate values of three vertices in the fitted image are obtained; Calculating a transfer matrix for rotating the contour image according to the coordinate value and the pre-stored original coordinate value; performing coordinate transformation on the pixel point on the contour image according to the transfer matrix; setting the coordinate image of the coordinate transformation to Feature area image.
  • the transfer matrix can be recorded as:
  • Coordinate transformation can be performed on points on the image of the feature area according to the following formula:
  • (x, y) is the coordinate value of the pixel point on the image extracted by the contour before the coordinate transformation
  • (x', y') is the coordinate value of the pixel point on the image extracted by the contour after the coordinate transformation.
  • the coordinates of the upper left corner, the lower left corner, and the upper right corner of the rectangle are (x 1 , y 1 ), (x 2 , y 2 ), (x 3 , y 3 ), respectively, and the coordinates of the corrected rectangle are (0, 0). ), (0, h), (w, 0), where w and h represent the width and height of the corrected rectangle, and in order to ensure that the corrected rectangle is similar or identical to the original image, it should satisfy:
  • (x, y) represents a coordinate value of a pixel point in the image of the feature area
  • (x+d, y+d) represents a coordinate value in the image of the feature area template image and the feature area image
  • I(x, y) represents the pixel value of the pixel point whose coordinate value is (x, y) in the image of the feature area
  • M(x+d, y+d) represents a pixel value of a pixel point whose coordinate value is (x+d, y+d) in the feature region template image
  • C represents the similarity.
  • each feature region template image may be stored locally in advance, and the storage order may be set to be the same or corresponding to the storage order of the respective feature region images.
  • each feature region template image may be numbered, and the number may be set to be the same as or corresponding to the storage order of the respective feature region images.
  • the determining module 30 is configured to determine that the component to be tested is faulty if the pixel similarity is less than a preset similarity threshold.
  • the pixel similarity between the feature area image of the device to be tested and the feature area reference image is smaller than the preset similarity threshold, it indicates that the feature area image of the device to be tested differs greatly from the character area of the reference image of the feature region, so that the The wrong component of the component to be tested is described; otherwise, if the pixel similarity between the feature region image of the device to be tested and the feature region reference image is greater than or equal to the preset similarity threshold, the feature region image and the feature region reference image of the device to be tested are indicated.
  • the character areas are relatively similar, so that it can be determined that the device under test is not wrong.
  • the similarity threshold may be set according to actual conditions. Generally, the greater the value of the similarity threshold, the higher the detection accuracy.
  • the component fault detection system of the invention realizes the automation of component fault detection, and can effectively improve the detection efficiency and accuracy.
  • the color and/or shape characteristics of the elements are relatively similar, it is possible to distinguish whether or not the similarity is obtained by extracting the character information on the element to be tested and comparing it with the template element.
  • the component error detecting system of the present invention has a one-to-one correspondence with the component wrong component detecting method of the present invention, and the technical characters and the advantageous effects thereof described in the embodiment of the component wrong component detecting method are applicable to the embodiment of the component wrong component detecting system. In this regard, hereby declare.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Multimedia (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Quality & Reliability (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Image Analysis (AREA)

Abstract

一种元件错件检测方法和系统,其中,方法包括以下步骤:获取待测元件在电路板上的原始图像,从所述原始图像中定位所述待测元件的特征区域图像;其中,所述特征区域图像包含所述待测元件的特征信息,所述特征信息用于对所述待测元件与其他元件进行区分(S1);将所述特征区域图像中各个像素点的像素值分别与预存的特征区域模板图像中对应像素点的像素值进行比较,获取所述特征区域图像与所述特征区域模板图像的像素相似度(S2);若所述像素相似度小于预设的相似度阈值,判定所述待测元件错件(S3)。上述元件错件检测方法和系统,实现了元件错件检测的自动化,能够有效提高检测效率和准确性。

Description

元件错件检测方法和系统 技术领域
本发明涉及自动光学检测技术领域,特别是涉及一种元件错件检测方法和系统。
背景技术
AOI(Automatic Optic Inspection,自动光学检测),是利用光学原理对电路板焊接生产中出现的常见缺陷进行检测的设备。对于插件的电路板来说,常见的缺陷检测包括漏件检测、错件检测、反件检测、多件检测等。其中,错件检测是指提取待检测元件的特征,并与模板进行比较,从而判断插入电路板的元件是否正确。
目前,元件的错件检测主要由人工进行检测,但是,这种检测方式效率较低,而且,检测结果容易出错,检测正确率较低。
发明内容
基于此,有必要针对现有技术检测效率低、正确率低的问题,提供一种元件错件检测方法和系统。
一种元件错件检测方法,包括以下步骤:
获取待测元件在电路板上的原始图像,从所述原始图像中定位所述待测元件的特征区域图像;其中,所述特征区域图像包含所述待测元件的特征信息,所述特征信息用于对所述待测元件与其他元件进行区分;
将所述特征区域图像中各个像素点的像素值分别与预存的特征区域模板图像中对应像素点的像素值进行比较,获取所述特征区域图像与所述特征区域模板图像的像素相似度;
若所述像素相似度小于预设的相似度阈值,判定所述待测元件错件。
一种元件错件检测系统,包括:
定位模块,用于获取待测元件在电路板上的原始图像,从所述原始图像中定位所述待测元件的特征区域图像;其中,所述特征区域图像包含所述待测元件的特征信息,所述特征信息用于对所述待测元件与其他元件进行区分;
比较模块,用于将所述特征区域图像中各个像素点的像素值分别与预存的特征区域模板图像中对应像素点的像素值进行比较,获取所述特征区域图像与所述特征区域模板图像的像素相似度;
判断模块,用于若所述像素相似度小于预设的相似度阈值,判定所述待测元件错件。
上述元件错件检测方法和系统,通过检测待测元件的特征区域,当特征区域图像与特征区域模板图像的像素相似度较小时,判定待测元件错件,实现了元件错件检测的自动化,能够有效提高检测效率和准确性。
附图说明
图1为一个实施例的元件错件检测方法流程图;
图2为经轮廓提取的字符区域图像;
图3为校正后的字符区域图像;
图4为一个实施例的元件错件检测系统的结构示意图。
具体实施方式
下面结合附图对本发明的技术方案进行说明。
图1为一个实施例的元件错件检测方法流程图。如图1所示,所述元件错件检测方法可包括以下步骤:
S1,获取待测元件在电路板上的原始图像,从所述原始图像中定位所述待测元件的特征区域图像;其中,所述特征区域图像包含所述待测元件的特征信息,所述特征信息用于对所述待测元件与其他元件进行区分;
本发明所述的特征可以包括待测元件的颜色、形状和待测元件的主体区域上的字符信息等便于使待测元件与其他元件区分开的信息。下面以所述特征为字符信息为例进行说明。所述字符信息可以是文字、符号、图案等。
在检测前,可以首先从整个电路板的图像中定位出所述待测元件的原始图像,再从所述原始图像中定位出所述待测元件的特征区域图像。当一块电路板上有多个待测元件都需要进行错件检测时,可以分别获取各个待测元件的原始图像,再分别从各个原始图像中定位出各个待测元件的特征区域图像。可以根据各个待测元件在电路板上的位置对各个待测 元件对应的特征区域图像进行顺序存储,以便于后续检测操作的执行。在其中一个实施例中,还可以为各个特征区域图像顺序编号,以便于后续检测操作的执行。
S2,将所述特征区域图像中各个像素点的像素值分别与预存的特征区域模板图像中对应像素点的像素值进行比较,获取所述特征区域图像与所述特征区域模板图像的像素相似度;
在本步骤中,为了防止特征区域太小,不利于后续操作,在进行比较之前,还可以对特征区域图像进行放大处理。所述放大处理是指尺寸的放大,即将图像的长宽分别放大到原来的n倍,n可以根据实际的需求设置,一般情况n=2即可。
为了消除电路板上的污点以及背景颜色和图案等因素对检测结果的影响,可以对所述特征区域图像进行降噪处理。具体地,可以对所述特征区域图像进行灰度处理,得到灰度图像,并根据预设的像素阈值对所述灰度图像进行二值化处理。所示灰度处理可以根据如下公式进行:
Gray=0.299*R+0.587*G+0.114*B;
式中,R、G和B分别为RGB颜色空间的三个颜色分量,Gray为二值化后的灰度值。所述二值化可以将灰度值大于预设的灰度阈值的像素点设为某一灰度值,将像素值小于或等于预设的灰度阈值的像素点设为另一灰度值。其中,所述灰度阈值可以是使以下目标函数的函数值最大的灰度值:
g(t)=ω0*(μ0-μ)21*(μ1-μ)2
其中,μ=ω0011
式中,ω0为所述特征信息对应的像素点在所述特征区域图像中的比例,μ0为所述特征信息对应的像素点的像素值的均值,ω1为背景图像的像素点在所述特征区域图像中的比例,μ1为背景图像的像素点的像素值的均值。通过这种方式,可以将灰度图像的灰度值分成两个部分,且两部分之间的灰度值差异最大、每个部分之间的灰度差异最小。
图像的降噪处理虽然能够减少噪声的干扰,但是并不能完全消除掉噪声的干扰。通常情况下,噪声在图像中较为细小、离散,而字符区域较为连续。所以为了更加精确的定位字符区域,可以对二值化的图像进行形态学处理,将字符区域连在一起形成一片区域,并进行轮廓提取,轮廓面积最大的部分即为文字区域。经轮廓提取的字符区域图像如图2所示。
由于文字区域会有一定角度的旋转,因此可以对图像进行校正。具体地,在对经形态学处理的特征区域图像进行轮廓提取之后,还可以对所述轮廓图像进行最小矩阵拟合,得到拟合图像;获取所述拟合图像中三个顶点的坐标值;根据所述坐标值与预存的原始坐标值计算对所述轮廓图像进行旋转的转移矩阵;根据所述转移矩阵对所述轮廓图像上的像素点进行坐标变换;将经坐标变换的轮廓图像设为特征区域图像。
其中,所述转移矩阵可记为:
Figure PCTCN2016113146-appb-000001
可根据如下公式对所述特征区域图像上的点进行坐标变换:
Figure PCTCN2016113146-appb-000002
式中,(x,y)为坐标变换前经轮廓提取的图像上的像素点的坐标值,(x',y')为坐标变换后经轮廓提取的图像上的像素点的坐标值。假设矩形左上角、左下角和右上角的坐标分别为(x1,y1)、(x2,y2)、(x3,y3),经过校正后的矩形的坐标为(0,0)、(0,h)、(w,0),其中w和h表示校正后矩形的宽和高,且为了保证校正后的矩形与原始图像相近或相同,应满足:
Figure PCTCN2016113146-appb-000003
由此,设原始图像变换到校正后的图像的仿射变换矩阵M的六个元素。校正后的字符区域图像如图3所示。
比较时,可以根据如下公式计算所述相似度:
Figure PCTCN2016113146-appb-000004
式中,(x,y)表示所述特征区域图像中的像素点的坐标值,(x+d,y+d)表示所述特征区域模板图像中与所述特征区域图像中坐标值为(x,y)的像素点相对应的像素点的坐标值,I(x,y)表示所述特征区域图像中坐标值为(x,y)的像素点的像素值,M(x+d,y+d)表示所述特征区域模板图像中坐标值为(x+d,y+d)的像素点的像素值,C表示所述相似度。
若步骤S1中存储了多个特征区域图像,本步骤可以分别将各个特征区域图像与对应 的特征区域模板图像进行像素值的比较。在一个实施例中,各个特征区域模板图像可以预先顺序存储在本地,期存储顺序可以设置为与各个特征区域图像的存储顺序相同或相应。或者,可以为各个特征区域模板图像编号,其编号可以设置为与各个特征区域图像的存储顺序相同或相应。通过顺序存储特征区域图像和/或特征区域模板图像的方式,可以便于并行地对多个待测元件进行比较,从而提高元件错件检测效率。
S3,若所述像素相似度小于预设的相似度阈值,判定所述待测元件错件。
在本步骤中,若待测元件的特征区域图像与特征区域参考图像的像素相似度小于预设的相似度阈值,则表明待测元件的特征区域图像与特征区域参考图像的字符区域相差较大,从而可以判定所述待测元件错件;反之,若待测元件的特征区域图像与特征区域参考图像的像素相似度大于或等于预设的相似度阈值,则表明待测元件的特征区域图像与特征区域参考图像的字符区域较为相似,从而可以判定所述待测元件未错件。
所述相似度阈值可以根据实际情况自行设定,一般来说,所述相似度阈值的值越大,检测准确性越高。
本发明的元件错件检测方法实现了元件错件检测的自动化,能够有效提高检测效率和准确性。尤其是在元件的颜色和/或形状特征比较相似时,通过提取待测元件上的字符信息,并与模板元件进行比较,能够很好地判别出是否相似。
与上述元件错件检测方法相对应地,本发明还提供一种元件错件检测系统。如图2所示,所述元件错件检测系统可包括:
定位模块10,用于获取待测元件在电路板上的原始图像,从所述原始图像中定位所述待测元件的特征区域图像;其中,所述特征区域图像包含所述待测元件的特征信息,所述特征信息用于对所述待测元件与其他元件进行区分;
本发明所述的特征可以包括待测元件的颜色、形状和待测元件的主体区域上的字符信息等便于使待测元件与其他元件区分开的信息。下面以所述特征为字符信息为例进行说明。所述字符信息可以是文字、符号、图案等。
在检测前,可以首先从整个电路板的图像中定位出所述待测元件的原始图像,再从所述原始图像中定位出所述待测元件的特征区域图像。当一块电路板上有多个待测元件都需要进行错件检测时,可以分别获取各个待测元件的原始图像,再分别从各个原始图像中定位出各个待测元件的特征区域图像。可以根据各个待测元件在电路板上的位置对各个待测 元件对应的特征区域图像进行顺序存储,以便于后续检测操作的执行。在其中一个实施例中,还可以为各个特征区域图像顺序编号,以便于后续检测操作的执行。
比较模块20,用于将所述特征区域图像中各个像素点的像素值分别与预存的特征区域模板图像中对应像素点的像素值进行比较,获取所述特征区域图像与所述特征区域模板图像的像素相似度;
为了防止特征区域太小,不利于后续操作,在进行比较之前,还可以对特征区域图像进行放大处理。所述放大处理是指尺寸的放大,即将图像的长宽分别放大到原来的n倍,n可以根据实际的需求设置,一般情况n=2即可。
为了消除电路板上的污点以及背景颜色和图案等因素对检测结果的影响,可以对所述特征区域图像进行降噪处理。具体地,可以对所述特征区域图像进行灰度处理,得到灰度图像,并根据预设的像素阈值对所述灰度图像进行二值化处理。所示灰度处理可以根据如下公式进行:
Gray=0.299*R+0.587*G+0.114*B;
式中,R、G和B分别为RGB颜色空间的三个颜色分量,Gray为二值化后的灰度值。所述二值化可以将灰度值大于预设的灰度阈值的像素点设为某一灰度值,将像素值小于或等于预设的灰度阈值的像素点设为另一灰度值。其中,所述灰度阈值可以是使以下目标函数的函数值最大的灰度值:
g(t)=ω0*(μ0-μ)21*(μ1-μ)2
其中,μ=ω0011
式中,ω0为所述特征信息对应的像素点在所述特征区域图像中的比例,μ0为所述特征信息对应的像素点的像素值的均值,ω1为背景图像的像素点在所述特征区域图像中的比例,μ1为背景图像的像素点的像素值的均值。通过这种方式,可以将灰度图像的灰度值分成两个部分,且两部分之间的灰度值差异最大、每个部分之间的灰度差异最小。
图像的降噪处理虽然能够减少噪声的干扰,但是并不能完全消除掉噪声的干扰。通常情况下,噪声在图像中较为细小、离散,而字符区域较为连续。所以为了更加精确的定位字符区域,可以对二值化的图像进行形态学处理,将字符区域连在一起形成一片区域,并进行轮廓提取,轮廓面积最大的部分即为文字区域。经轮廓提取的字符区域图像如图2所示。
由于文字区域会有一定角度的旋转,因此可以对图像进行校正。具体地,在对经形态学处理的特征区域图像进行轮廓提取之后,还可以对所述轮廓图像进行最小矩阵拟合,得到拟合图像;获取所述拟合图像中三个顶点的坐标值;根据所述坐标值与预存的原始坐标值计算对所述轮廓图像进行旋转的转移矩阵;根据所述转移矩阵对所述轮廓图像上的像素点进行坐标变换;将经坐标变换的轮廓图像设为特征区域图像。
其中,所述转移矩阵可记为:
Figure PCTCN2016113146-appb-000005
可根据如下公式对所述特征区域图像上的点进行坐标变换:
Figure PCTCN2016113146-appb-000006
式中,(x,y)为坐标变换前经轮廓提取的图像上的像素点的坐标值,(x',y')为坐标变换后经轮廓提取的图像上的像素点的坐标值。假设矩形左上角、左下角和右上角的坐标分别为(x1,y1)、(x2,y2)、(x3,y3),经过校正后的矩形的坐标为(0,0)、(0,h)、(w,0),其中w和h表示校正后矩形的宽和高,且为了保证校正后的矩形与原始图像相近或相同,应满足:
Figure PCTCN2016113146-appb-000007
由此,设原始图像变换到校正后的图像的仿射变换矩阵M的六个元素。校正后的字符区域图像如图3所示。
比较时,可以根据如下公式计算所述相似度:
Figure PCTCN2016113146-appb-000008
式中,(x,y)表示所述特征区域图像中的像素点的坐标值,(x+d,y+d)表示所述特征区域模板图像中与所述特征区域图像中坐标值为(x,y)的像素点相对应的像素点的坐标值,I(x,y)表示所述特征区域图像中坐标值为(x,y)的像素点的像素值,M(x+d,y+d)表示所述特征区域模板图像中坐标值为(x+d,y+d)的像素点的像素值,C表示所述相似度。
若定位模块10中存储了多个特征区域图像,比较模块20可以分别将各个特征区域图 像与对应的特征区域模板图像进行像素值的比较。在一个实施例中,各个特征区域模板图像可以预先顺序存储在本地,期存储顺序可以设置为与各个特征区域图像的存储顺序相同或相应。或者,可以为各个特征区域模板图像编号,其编号可以设置为与各个特征区域图像的存储顺序相同或相应。通过顺序存储特征区域图像和/或特征区域模板图像的方式,可以便于并行地对多个待测元件进行比较,从而提高元件错件检测效率。
判断模块30,用于若所述像素相似度小于预设的相似度阈值,判定所述待测元件错件。
若待测元件的特征区域图像与特征区域参考图像的像素相似度小于预设的相似度阈值,则表明待测元件的特征区域图像与特征区域参考图像的字符区域相差较大,从而可以判定所述待测元件错件;反之,若待测元件的特征区域图像与特征区域参考图像的像素相似度大于或等于预设的相似度阈值,则表明待测元件的特征区域图像与特征区域参考图像的字符区域较为相似,从而可以判定所述待测元件未错件。
所述相似度阈值可以根据实际情况自行设定,一般来说,所述相似度阈值的值越大,检测准确性越高。
本发明的元件错件检测系统实现了元件错件检测的自动化,能够有效提高检测效率和准确性。尤其是在元件的颜色和/或形状特征比较相似时,通过提取待测元件上的字符信息,并与模板元件进行比较,能够很好地判别出是否相似。
本发明的元件错件检测系统与本发明的元件错件检测方法一一对应,在上述元件错件检测方法的实施例阐述的技术字符及其有益效果均适用于元件错件检测系统的实施例中,特此声明。
以上所述实施例的各技术字符可以进行任意的组合,为使描述简洁,未对上述实施例中的各个技术字符所有可能的组合都进行描述,然而,只要这些技术字符的组合不存在矛盾,都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本发明的几种实施方式,其描述较为具体和详细,但并不能因此而理解为对发明专利范围的限制。应当指出的是,对于本领域的普通技术人员来说,在不脱离本发明构思的前提下,还可以做出若干变形和改进,这些都属于本发明的保护范围。因此,本发明专利的保护范围应以所附权利要求为准。

Claims (10)

  1. 一种元件错件检测方法,其特征在于,包括以下步骤:
    获取待测元件在电路板上的原始图像,从所述原始图像中定位所述待测元件的特征区域图像;其中,所述特征区域图像包含所述待测元件的特征信息,所述特征信息用于对所述待测元件与其他元件进行区分;
    将所述特征区域图像中各个像素点的像素值分别与预存的特征区域模板图像中对应像素点的像素值进行比较,获取所述特征区域图像与所述特征区域模板图像的像素相似度;
    若所述像素相似度小于预设的相似度阈值,判定所述待测元件错件。
  2. 根据权利要求1所述的元件错件检测方法,其特征在于,在将所述特征区域图像中各个像素点的像素值分别与预存的特征区域模板图像中对应像素点的像素值进行比较之前,还包括以下步骤:
    对所述特征区域图像进行放大处理。
  3. 根据权利要求1所述的元件错件检测方法,其特征在于,从所述原始图像中定位所述待测元件的特征区域图像的步骤包括:
    对所述特征区域图像进行形态学处理;
    对经形态学处理的特征区域图像进行轮廓提取,得到轮廓图像;
    将所述轮廓图像中轮廓面积最大的部分设为特征区域图像。
  4. 根据权利要求3所述的元件错件检测方法,其特征在于,对所述特征区域图像进行形态学处理的步骤包括:
    对所述特征区域图像进行灰度处理,得到灰度图像;
    根据预设的像素阈值对所述灰度图像进行二值化处理;
    对二值化的灰度图像进行形态学处理。
  5. 根据权利要求4所述的元件错件检测方法,其特征在于,所述预设的灰度阈值为使以下目标函数的函数值最大的灰度值:
    g(t)=ω0*(μ0-μ)21*(μ1-μ)2
    其中,μ=ω0011
    式中,ω0为所述特征信息对应的像素点在所述特征区域图像中的比例,μ0为所述特 征信息对应的像素点的像素值的均值,ω1为背景图像的像素点在所述特征区域图像中的比例,μ1为背景图像的像素点的像素值的均值。
  6. 根据权利要求3所述的元件错件检测方法,其特征在于,在对经形态学处理的特征区域图像进行轮廓提取之后,还包括以下步骤:
    对所述轮廓图像进行最小矩阵拟合,得到拟合图像;
    获取所述拟合图像中三个顶点的坐标值;
    根据所述坐标值与预存的原始坐标值计算对所述轮廓图像进行旋转的转移矩阵;
    根据所述转移矩阵对所述轮廓图像上的像素点进行坐标变换;
    将经坐标变换的轮廓图像设为特征区域图像。
  7. 根据权利要求6所述的元件错件检测方法,其特征在于,根据所述转移矩阵对所述轮廓图像上的像素点进行坐标变换的步骤包括:
    根据如下公式对所述特征区域图像上的点进行坐标变换:
    Figure PCTCN2016113146-appb-100001
    式中,M为转移矩阵,(x,y)为坐标变换前经轮廓提取的图像上的像素点的坐标值,(x',y')为坐标变换后经轮廓提取的图像上的像素点的坐标值。
  8. 根据权利要求1所述的元件错件检测方法,其特征在于,获取所述特征区域图像与所述特征区域模板图像的像素相似度的步骤包括:
    根据如下公式计算所述相似度:
    Figure PCTCN2016113146-appb-100002
    式中,(x,y)表示所述特征区域图像中的像素点的坐标值,(x+d,y+d)表示所述特征区域模板图像中与所述特征区域图像中坐标值为(x,y)的像素点相对应的像素点的坐标值,I(x,y)表示所述特征区域图像中坐标值为(x,y)的像素点的像素值,M(x+d,y+d)表示所述特征区域模板图像中坐标值为(x+d,y+d)的像素点的像素值,C表示所述相似度。
  9. 根据权利要求1所述的元件错件检测方法,其特征在于,所述特征信息为字符信息。
  10. 一种元件错件检测系统,其特征在于,包括:
    定位模块,用于获取待测元件在电路板上的原始图像,从所述原始图像中定位所述待测元件的特征区域图像;其中,所述特征区域图像包含所述待测元件的特征信息,所述特征信息用于对所述待测元件与其他元件进行区分;
    比较模块,用于将所述特征区域图像中各个像素点的像素值分别与预存的特征区域模板图像中对应像素点的像素值进行比较,获取所述特征区域图像与所述特征区域模板图像的像素相似度;
    判断模块,用于若所述像素相似度小于预设的相似度阈值,判定所述待测元件错件。
PCT/CN2016/113146 2016-10-11 2016-12-29 元件错件检测方法和系统 WO2018068415A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610887939.3 2016-10-11
CN201610887939.3A CN106530275B (zh) 2016-10-11 2016-10-11 元件错件检测方法和系统

Publications (1)

Publication Number Publication Date
WO2018068415A1 true WO2018068415A1 (zh) 2018-04-19

Family

ID=58331466

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/113146 WO2018068415A1 (zh) 2016-10-11 2016-12-29 元件错件检测方法和系统

Country Status (2)

Country Link
CN (1) CN106530275B (zh)
WO (1) WO2018068415A1 (zh)

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109614879A (zh) * 2018-11-19 2019-04-12 温州大学 基于图像识别的料斗颗粒检测方法
CN109685781A (zh) * 2018-12-17 2019-04-26 江苏蜂奥生物科技有限公司 一种应用于蜂胶软胶囊的基于一定规则的多目标快速识别方法
CN111462035A (zh) * 2020-01-21 2020-07-28 北京明略软件系统有限公司 一种图片检测方法及装置
CN111476782A (zh) * 2020-04-10 2020-07-31 南通大学 一种汽车后摄像头延长线端子图像检测系统
CN111539933A (zh) * 2020-04-22 2020-08-14 大连日佳电子有限公司 一种直插元件检测方法及系统
CN111553376A (zh) * 2019-12-24 2020-08-18 西安元智系统技术有限责任公司 一种文物轮廓监测方法
CN111738247A (zh) * 2020-05-15 2020-10-02 上海望友信息科技有限公司 极性标识的识别方法、识别装置、电子设备及存储介质
CN112308814A (zh) * 2019-07-26 2021-02-02 北京四方继保自动化股份有限公司 一种电力系统刀闸分合位状态自动识别方法和系统
CN113034488A (zh) * 2021-04-13 2021-06-25 荣旗工业科技(苏州)股份有限公司 一种喷墨印刷品的视觉检测方法
CN114332069A (zh) * 2022-01-05 2022-04-12 合肥工业大学 一种基于机器视觉的接插件检测方法及装置
CN114441499A (zh) * 2022-04-11 2022-05-06 天津美腾科技股份有限公司 品位检测方法及装置、识别设备、矿浆品位仪及存储介质
CN115482234A (zh) * 2022-10-08 2022-12-16 浙江花园药业有限公司 一种铝塑泡罩药品用高精度缺陷检测方法及系统
CN115797361A (zh) * 2023-02-13 2023-03-14 山东淼泰建筑科技有限公司 一种铝模板表面缺陷检测方法
CN116205919A (zh) * 2023-05-05 2023-06-02 深圳市智宇精密五金塑胶有限公司 基于人工智能的五金零件生产质量检测方法及系统

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106504231A (zh) * 2016-10-11 2017-03-15 广州视源电子科技股份有限公司 元件缺陷检测方法和系统
CN107330879A (zh) * 2017-06-21 2017-11-07 广州视源电子科技股份有限公司 一种器件的检查方法、装置、设备及存储介质
CN109870117A (zh) * 2017-12-05 2019-06-11 英业达科技有限公司 二值化轮廓检测系统及其方法
CN110544640A (zh) * 2018-05-28 2019-12-06 长鑫存储技术有限公司 半导体基片的检测方法及检测装置
CN111209944B (zh) * 2019-12-31 2023-09-01 上海索广映像有限公司 一种自适应的图像检测系统及图像检测方法
CN112750116B (zh) * 2021-01-15 2023-08-11 北京市商汤科技开发有限公司 一种缺陷检测方法、装置、计算机设备及存储介质
CN112801987B (zh) * 2021-02-01 2022-11-08 上海万物新生环保科技集团有限公司 一种手机零件异常检测方法及设备
CN115131287A (zh) * 2022-05-16 2022-09-30 奥蒂玛光学科技(深圳)有限公司 一种图像处理方法、区域检测方法、检测设备和存储介质
WO2024044913A1 (en) * 2022-08-29 2024-03-07 Siemens Aktiengesellschaft Method, apparatus, electronic device, storage medium and computer program product for detecting circuit board assembly defect
CN115564779B (zh) * 2022-12-08 2023-05-26 东莞先知大数据有限公司 一种零件缺陷检测方法、装置及存储介质
CN116152251B (zh) * 2023-04-20 2023-07-14 成都数之联科技股份有限公司 电视机背板检测方法、模型训练方法、装置、设备和介质
CN117152145B (zh) * 2023-10-31 2024-02-23 威海天拓合创电子工程有限公司 一种基于图像的板卡工艺检测方法及装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101055620A (zh) * 2006-04-12 2007-10-17 富士通株式会社 形状比较装置和方法
CN102831381A (zh) * 2011-06-15 2012-12-19 鸿富锦精密工业(深圳)有限公司 影像差异比较系统及方法
CN104463178A (zh) * 2014-12-29 2015-03-25 广州视源电子科技股份有限公司 电子元件识别方法和系统
CN105957059A (zh) * 2016-04-20 2016-09-21 广州视源电子科技股份有限公司 电子元件漏件检测方法和系统

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102222328B (zh) * 2011-07-01 2012-10-03 杭州电子科技大学 一种边缘保持的自然场景图像自适应加权滤波方法
CN103810478B (zh) * 2014-02-21 2018-01-09 广东小天才科技有限公司 一种坐姿检测方法和装置
CN104363449B (zh) * 2014-10-31 2017-10-10 华为技术有限公司 图像预测方法及相关装置
CN104931907B (zh) * 2015-05-13 2017-12-26 中国矿业大学 基于机器视觉的数显电测量仪表质量群检系统

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101055620A (zh) * 2006-04-12 2007-10-17 富士通株式会社 形状比较装置和方法
CN102831381A (zh) * 2011-06-15 2012-12-19 鸿富锦精密工业(深圳)有限公司 影像差异比较系统及方法
CN104463178A (zh) * 2014-12-29 2015-03-25 广州视源电子科技股份有限公司 电子元件识别方法和系统
CN105957059A (zh) * 2016-04-20 2016-09-21 广州视源电子科技股份有限公司 电子元件漏件检测方法和系统

Cited By (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109614879B (zh) * 2018-11-19 2022-12-02 温州大学 基于图像识别的料斗颗粒检测方法
CN109614879A (zh) * 2018-11-19 2019-04-12 温州大学 基于图像识别的料斗颗粒检测方法
CN109685781A (zh) * 2018-12-17 2019-04-26 江苏蜂奥生物科技有限公司 一种应用于蜂胶软胶囊的基于一定规则的多目标快速识别方法
CN109685781B (zh) * 2018-12-17 2022-11-29 江苏蜂奥生物科技有限公司 一种应用于蜂胶软胶囊的基于一定规则的多目标快速识别方法
CN112308814A (zh) * 2019-07-26 2021-02-02 北京四方继保自动化股份有限公司 一种电力系统刀闸分合位状态自动识别方法和系统
CN111553376A (zh) * 2019-12-24 2020-08-18 西安元智系统技术有限责任公司 一种文物轮廓监测方法
CN111553376B (zh) * 2019-12-24 2023-05-23 西安元智系统技术有限责任公司 一种文物轮廓监测方法
CN111462035A (zh) * 2020-01-21 2020-07-28 北京明略软件系统有限公司 一种图片检测方法及装置
CN111462035B (zh) * 2020-01-21 2024-03-08 北京明略软件系统有限公司 一种图片检测方法及装置
CN111476782A (zh) * 2020-04-10 2020-07-31 南通大学 一种汽车后摄像头延长线端子图像检测系统
CN111539933A (zh) * 2020-04-22 2020-08-14 大连日佳电子有限公司 一种直插元件检测方法及系统
CN111539933B (zh) * 2020-04-22 2023-06-06 大连日佳电子有限公司 一种直插元件检测方法及系统
CN111738247A (zh) * 2020-05-15 2020-10-02 上海望友信息科技有限公司 极性标识的识别方法、识别装置、电子设备及存储介质
CN113034488A (zh) * 2021-04-13 2021-06-25 荣旗工业科技(苏州)股份有限公司 一种喷墨印刷品的视觉检测方法
CN113034488B (zh) * 2021-04-13 2024-04-19 荣旗工业科技(苏州)股份有限公司 一种喷墨印刷品的视觉检测方法
CN114332069A (zh) * 2022-01-05 2022-04-12 合肥工业大学 一种基于机器视觉的接插件检测方法及装置
CN114332069B (zh) * 2022-01-05 2024-02-20 合肥工业大学 一种基于机器视觉的接插件检测方法及装置
CN114441499A (zh) * 2022-04-11 2022-05-06 天津美腾科技股份有限公司 品位检测方法及装置、识别设备、矿浆品位仪及存储介质
CN115482234A (zh) * 2022-10-08 2022-12-16 浙江花园药业有限公司 一种铝塑泡罩药品用高精度缺陷检测方法及系统
CN115482234B (zh) * 2022-10-08 2023-09-15 浙江花园药业有限公司 一种铝塑泡罩药品用高精度缺陷检测方法及系统
CN115797361A (zh) * 2023-02-13 2023-03-14 山东淼泰建筑科技有限公司 一种铝模板表面缺陷检测方法
CN115797361B (zh) * 2023-02-13 2023-06-27 山东淼泰建筑科技有限公司 一种铝模板表面缺陷检测方法
CN116205919A (zh) * 2023-05-05 2023-06-02 深圳市智宇精密五金塑胶有限公司 基于人工智能的五金零件生产质量检测方法及系统
CN116205919B (zh) * 2023-05-05 2023-06-30 深圳市智宇精密五金塑胶有限公司 基于人工智能的五金零件生产质量检测方法及系统

Also Published As

Publication number Publication date
CN106530275B (zh) 2019-06-11
CN106530275A (zh) 2017-03-22

Similar Documents

Publication Publication Date Title
WO2018068415A1 (zh) 元件错件检测方法和系统
CN111474184B (zh) 基于工业机器视觉的aoi字符缺陷检测方法和装置
CN110210409B (zh) 表格单据中表格框线检测方法及系统
CN109472271B (zh) 印刷电路板图像轮廓提取方法及装置
WO2018068417A1 (zh) 元件缺陷检测方法和系统
WO2017181724A1 (zh) 电子元件漏件检测方法和系统
WO2017020559A1 (zh) 一种基于行列直线聚类的多类型bga芯片视觉识别方法
WO2018010391A1 (zh) 一种板卡的检测方法及装置
CN113344931B (zh) 一种插件视觉检测识别方法、可读存储介质及设备
US7406191B2 (en) Inspection data producing method and board inspection apparatus using the method
WO2018068418A1 (zh) 元件安装状态检测方法和系统
US8515156B2 (en) Image analysis device and method
WO2017071406A1 (zh) 金针类元件的引脚检测方法和系统
CN116168218A (zh) 一种基于图像识别技术的电路板故障诊断方法
JP2018096908A (ja) 検査装置及び検査方法
WO2017107529A1 (zh) 一种并排二极管的定位方法及装置
CN112419207A (zh) 一种图像矫正方法及装置、系统
CN113012096A (zh) 显示屏子像素定位及亮度提取方法、设备以及存储介质
CN114677348A (zh) 一种基于视觉的ic芯片缺陷的检测方法、系统和存储介质
Anitha et al. SMT component inspection in PCBA’s using image processing techniques
CN109509165B (zh) 图像定位区域选取方法及装置
CN113744252A (zh) 用于标记和检测缺陷的方法、设备、存储介质和程序产品
US9443165B2 (en) Blob-encoding
CN112419225A (zh) 一种基于引脚分割的sop型芯片检测方法及系统
CN108898584B (zh) 一种基于图像分析的全自动贴面电容装焊极性判别方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16918526

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16918526

Country of ref document: EP

Kind code of ref document: A1