WO2018010386A1 - 元件反件检测方法和系统 - Google Patents

元件反件检测方法和系统 Download PDF

Info

Publication number
WO2018010386A1
WO2018010386A1 PCT/CN2016/113129 CN2016113129W WO2018010386A1 WO 2018010386 A1 WO2018010386 A1 WO 2018010386A1 CN 2016113129 W CN2016113129 W CN 2016113129W WO 2018010386 A1 WO2018010386 A1 WO 2018010386A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
polar
color
region
component
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.)
Ceased
Application number
PCT/CN2016/113129
Other languages
English (en)
French (fr)
Inventor
李红匣
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.)
Guangzhou Shiyuan Electronics Thecnology Co Ltd
Original Assignee
Guangzhou Shiyuan Electronics Thecnology Co Ltd
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 Guangzhou Shiyuan Electronics Thecnology Co Ltd filed Critical Guangzhou Shiyuan Electronics Thecnology Co Ltd
Publication of WO2018010386A1 publication Critical patent/WO2018010386A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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]

Definitions

  • the present invention relates to the field of automatic optical detection technology, and in particular to a method and system for detecting component reverse parts.
  • AOI Automatic Optic Inspection
  • common defect detection includes missing parts detection, wrong part detection, reverse part detection, multi-piece detection, and the like.
  • the reverse component detection refers to the detection of polar components such as diodes, capacitors, and sockets, and judges whether there is a reverse phenomenon in the circuit board.
  • the component detection of the component is mainly based on the intelligent method, that is, the deep learning method is used to train a large number of samples to obtain a classification model.
  • Deep learning is a new field of machine learning research. Its purpose is to simulate the mechanism of the human brain to interpret data and discover distributed feature representations of data.
  • the learning models established under different learning frameworks are also different. For example, Convolutional Neural Networks (CNNs) is a deep machine learning model.
  • CNNs Convolutional Neural Networks
  • the component polarity detection classifier trained by the convolutional neural network although achieving a desirable effect in the polarity detection of the component, has its own drawbacks that cannot be solved by itself.
  • the component polarity detection model trained by the convolutional neural network has a high recognition rate for known components and can achieve a good detection effect. However, for unknown components, that is, components that do not exist in the training sample, the recognition rate of the component polarity detection model decreases, and false positives and false negatives often occur.
  • the existing component polarity detection method has a poor detection effect.
  • a method for detecting a component reverse component includes the following steps:
  • polarity area is an area where the electrode of the device to be tested is mounted on the circuit board, and the polarity symmetry area is reversed The area of the electrode on the circuit board;
  • a component reverse component detecting system includes:
  • Obtaining a module configured to obtain an image of a polar region and a region of a polar symmetric region of the device to be tested on the circuit board; wherein the polarity region is an area of the electrode of the device to be tested on the circuit board when the polarity is correctly installed, The region of the electrode on the circuit board when the symmetrical region is the reverse member;
  • a calculating module configured to separately calculate a first color similarity of the polar area image and the pre-stored polar area reference image, and a second color similarity of the polar symmetric area image and the polar area reference image ;
  • a determining module configured to determine, when the first color similarity is less than the second color similarity, the component to be tested.
  • the component reverse component detecting method and system calculate a polar symmetric region image of the device to be tested and the polar region reference image by calculating a first color similarity between the polarity region image of the device to be tested and the polar region reference image a second color similarity, when the first color similarity is smaller than the second color similarity, indicating that a difference between the polar symmetric region image of the device to be tested and the polar region reference image is small, and the The difference between the polar region image of the measuring element and the polar region reference image is large, thereby determining the component counter.
  • the above component reverse component detecting method and system do not require a large number of training samples, and only need to acquire the polar region and the polar symmetric region of the component, and the operation is simple, the recognition rate is high, and the detection effect is good.
  • FIG. 1 is a flow chart of a method for detecting a component reverse member of an embodiment
  • FIG. 2 is a schematic view of a polar region and a polar symmetric region
  • FIG. 3 is a schematic diagram of a template matching method
  • Figure 4 is a schematic diagram of the HSV color space
  • Fig. 5 is a schematic structural view of an element reverse detecting system of an embodiment.
  • the component reverse component detecting method may include the following steps:
  • FIG. 1 A schematic diagram of the polar region and the polar symmetric region of the present invention is shown in FIG.
  • the image of the circuit board may be acquired first, and then the polar region and the polar symmetric region are located from the image of the circuit board, and the polar region and the polar symmetric region are respectively intercepted from the image of the circuit board.
  • the corresponding image is set as the polar area image and the polar symmetrical area image.
  • the polar region image may be acquired by a template matching method.
  • the template matching method is shown in Figure 3. Specifically, a first image region matching the polar region reference image may be selected from the image of the circuit board; and pixels of each pixel in the reference image according to the first image region and the polar region may be selected a value, a first pixel similarity of the image region to the polar region reference image is calculated; and a first image region having a first pixel similarity greater than or equal to a preset first pixel similarity threshold is set as Polar area image.
  • the polar area reference image may be pre-stored in a storage area of the system and called from the storage area when the polar area image is acquired.
  • an area of the image of the circuit board adjacent to the first image area may be set as the first image area, and repeated A step of calculating a first pixel similarity of the first image region and the polar region reference image.
  • the area adjacent to the first image area is an area obtained by moving the first image area to the x-axis and the y-axis by a plurality of pixel points in an image of the circuit board.
  • each moving pixel point may be one pixel point or multiple pixel points, and the moving distance may be set according to actual needs.
  • the polar symmetric region image can also be acquired by a template matching method.
  • a second image region matching the polar symmetric region reference image may be selected from the image of the circuit board; Calculating a second pixel similarity of the image region and the polar symmetric region reference image by using a pixel value of each pixel point in the second image region and the polar symmetric region reference image; and comparing the second pixel similarity to or A second image region equal to a preset second pixel similarity threshold is set as the polar symmetric region image.
  • the polar symmetric region reference image may be pre-stored in a storage area of the system and called from the storage region when the polar region image is acquired.
  • an area adjacent to the second image area in the image of the circuit board may be set as the second image area, and repeated And calculating a second pixel similarity of the second image region and the polar region reference image.
  • the area adjacent to the second image area is an area obtained by moving the second image area to the x-axis and the y-axis by a plurality of pixel points in an image of the circuit board.
  • each moving pixel point may be one pixel point or multiple pixel points, and the moving distance may be set according to actual needs.
  • the polar area image and the polar symmetric area image may also be acquired according to other methods.
  • the first pixel similarity and the second pixel similarity may be calculated according to the following formula:
  • R(x, y) is the pixel similarity of the pixel of the image region and the coordinate region (x, y) in the polar region reference image
  • T(x, y) is the polar region
  • the pixel value of the pixel at coordinates (x, y) in the reference image, I(x, y) is the pixel value of the pixel at coordinates (x, y) in the image region.
  • the first pixel similarity threshold may be set according to actual needs. For example, it can be set to 0.8, or set to 0.9, or set to other values. The larger the first pixel similarity threshold, the higher the accuracy of image acquisition.
  • the polar region and the polar symmetric region of the element to be tested have sharp color differences, it is possible to detect whether the component is reversed according to the color similarity.
  • the color histogram of the image of the component and the reference image can be compared.
  • the image of the component to be tested can also be converted to the HSV color space.
  • HSV color space indication The figure is shown in Figure 4.
  • the most commonly used color space is the RGB model, which is often used for color display and image processing.
  • the HSV model is a color model for the user's perception, focusing on the representation of color.
  • R, G, B refer to red, green, and blue colors respectively
  • H refers to hue, which ranges from 0 to 360 degrees, and is used to indicate the color category, such as red is 0 degrees green is 120 degrees, blue It is 240 degrees
  • S refers to saturation, the value range is 0% to 100%, used to indicate the vividness of the color, such as the saturation of gray is 0%, the saturation of red (255,0,0) is 100%
  • V refers to the brightness, the value range is 0% to 100%, used to indicate the degree of light and darkness of the color, such as black brightness is 0%, white brightness is 100%.
  • RGB space HSV space can express the brightness, color and vividness of colors very intuitively.
  • the pixel values of the polar region image can be converted to the HSV color space according to the following formula:
  • Min min(R, G, B);
  • R, G, and B are the color components of the RGB space.
  • a first color histogram of the polar area image may be acquired, according to the first color histogram
  • the first color similarity is calculated by the number of pixel points of each component amount and the number of pixel points of the corresponding component in the second color histogram of the polar region reference image.
  • a third color histogram of the polar symmetric region image may be acquired, according to the third color
  • the second color similarity is calculated by the number of pixel points of each component amount in the histogram and the number of pixel points of the corresponding component in the second color histogram.
  • the color histogram may be a color histogram of the H-S channel.
  • the first color similarity can be calculated according to the following formula:
  • H 1 (I) is the number of pixel points of the first component amount in the first color histogram, The average number of pixels of each component amount in the first color histogram
  • H 1 ' (I) is the number of pixel points of the first component amount in the second color histogram, The average number of pixels of each component amount in the second color histogram
  • the second color similarity can be calculated according to the following formula:
  • H 2 (I) is the number of pixel points of the first component amount in the third color histogram, The average number of pixels of each component amount in the third color histogram.
  • color similarity may also be calculated in other ways.
  • Chi-Square can be used.
  • the first color similarity can be calculated according to the following formula:
  • Color similarity can also be calculated according to the Intersection algorithm. Specifically, the first color similarity can be calculated according to the following formula:
  • Color similarity can also be calculated from Bhattacharyya.
  • the first color similarity can be calculated according to the following formula:
  • N is the number of histogram bins.
  • the component may be dirty for some reason, ie a non-polar region of the component may appear
  • the color of the polar region is similar, and in order to ensure the correct rate, the third color similarity between the polar region image and the pre-stored polar symmetric region reference image, and the polar symmetric region image and the pre-stored pole may be separately calculated.
  • a fourth color similarity of the symmetrical region reference image if the first color similarity is less than the second color similarity, and the third color similarity is greater than the fourth color similarity, determining the waiting Test component reverse.
  • the manner of calculating the third color similarity and the fourth color similarity may be similar to the manner of calculating the first color similarity and the second color similarity, and details are not described herein again.
  • the present invention further provides a component reverse component detecting system.
  • the component reverse component detecting system may include:
  • the obtaining module 10 is configured to obtain an image of a polar region and a region of a polar symmetry region of the device to be tested on the circuit board; wherein, the polar region is an area of the electrode of the device to be tested on the circuit board when the device is correctly installed.
  • the region of the electrode on the circuit board when the polar symmetric region is the reverse member;
  • FIG. 1 A schematic diagram of the polar region and the polar symmetric region of the present invention is shown in FIG.
  • the image of the circuit board may be acquired first, and then the polar region and the polar symmetric region are located from the image of the circuit board, and the polar region and the polar symmetric region are respectively intercepted from the image of the circuit board.
  • the corresponding image is set as the polar area image and the polar symmetrical area image.
  • the polar region image may be acquired by a template matching method.
  • the template matching method is shown in Figure 3. Specifically, a first image region matching the polar region reference image may be selected from the image of the circuit board; and pixels of each pixel in the reference image according to the first image region and the polar region may be selected a value, a first pixel similarity of the image region to the polar region reference image is calculated; and a first image region having a first pixel similarity greater than or equal to a preset first pixel similarity threshold is set as Polar area image.
  • the polar region The test image may be pre-stored in a storage area of the system and called from the storage area when the polar area image is acquired.
  • an area of the image of the circuit board adjacent to the first image area may be set as the first image area, and repeated A step of calculating a first pixel similarity of the first image region and the polar region reference image.
  • the area adjacent to the first image area is an area obtained by moving the first image area to the x-axis and the y-axis by a plurality of pixel points in an image of the circuit board.
  • the pixel point of each movement may be one pixel point or multiple pixel points, and the moving distance may be set according to actual needs.
  • the polar symmetric region image can also be acquired by a template matching method.
  • a second image region matching the polar symmetric region reference image may be selected from the image of the circuit board; and each pixel in the reference image is referenced according to the second image region and the polar symmetric region a pixel value, calculating a second pixel similarity of the image region and the polar symmetric region reference image; and setting a second image region having a second pixel similarity greater than or equal to a preset second pixel similarity threshold Is the image of the polar symmetric region.
  • the polar symmetric region reference image may be pre-stored in a storage area of the system and called from the storage region when the polar region image is acquired.
  • an area adjacent to the second image area in the image of the circuit board may be set as the second image area, and repeated And calculating a second pixel similarity of the second image region and the polar region reference image.
  • the area adjacent to the second image area is an area obtained by moving the second image area to the x-axis and the y-axis by a plurality of pixel points in an image of the circuit board.
  • the pixel point of each movement may be one pixel point or multiple pixel points, and the moving distance may be set according to actual needs.
  • the polar area image and the polar symmetric area image may also be acquired according to other methods.
  • the first pixel similarity and the second pixel similarity may be calculated according to the following formula:
  • R(x, y) is the pixel of the image region and the polar region reference image with coordinates (x, y) Pixel similarity
  • T(x, y) is a pixel value of a pixel having a coordinate of (x, y) in the reference image of the polar region
  • I(x, y) is a coordinate of the image region (x, y) The pixel value of the pixel.
  • the first pixel similarity threshold may be set according to actual needs. For example, it can be set to 0.8, or set to 0.9, or set to other values. The larger the first pixel similarity threshold, the higher the accuracy of image acquisition.
  • a calculating module 20 configured to separately calculate a first color similarity of the polar area image and the pre-stored polar area reference image, and the polar symmetric area image is similar to the second color of the polar area reference image degree;
  • the polar region and the polar symmetric region of the element to be tested have sharp color differences, it is possible to detect whether the component is reversed according to the color similarity.
  • the color histogram of the image of the component and the reference image can be compared.
  • the image of the component to be tested can also be converted to the HSV color space.
  • a schematic diagram of the HSV color space is shown in Figure 4.
  • the most commonly used color space is the RGB model, which is often used for color display and image processing.
  • the HSV model is a color model for the user's perception, focusing on the representation of color.
  • R, G, B refer to red, green, and blue colors respectively
  • H refers to hue, which ranges from 0 to 360 degrees, and is used to indicate the color category, such as red is 0 degrees green is 120 degrees, blue It is 240 degrees
  • S refers to saturation, the value range is 0% to 100%, used to indicate the vividness of the color, such as the saturation of gray is 0%, the saturation of red (255,0,0) is 100%
  • V refers to the brightness, the value range is 0% to 100%, used to indicate the degree of light and darkness of the color, such as black brightness is 0%, white brightness is 100%.
  • RGB space HSV space can express the brightness, color and vividness of colors very intuitively.
  • the pixel values of the polar region image can be converted to the HSV color space according to the following formula:
  • Min min(R, G, B);
  • R, G, and B are the color components of the RGB space.
  • a first color histogram of the polar area image may be acquired, according to the first color histogram
  • the first color similarity is calculated by the number of pixel points of each component amount and the number of pixel points of the corresponding component in the second color histogram of the polar region reference image.
  • a third color histogram of the polar symmetric region image may be acquired, according to the third color
  • the second color similarity is calculated by the number of pixel points of each component amount in the histogram and the number of pixel points of the corresponding component in the second color histogram.
  • the color histogram may be a color histogram of the H-S channel.
  • the first color similarity can be calculated according to the following formula:
  • H 1 (I) is the number of pixel points of the first component amount in the first color histogram, The average number of pixels of each component amount in the first color histogram
  • H 1 ' (I) is the number of pixel points of the first component amount in the second color histogram, The average number of pixels of each component amount in the second color histogram
  • the second color similarity can be calculated according to the following formula:
  • H 2 (I) is the number of pixel points of the first component amount in the third color histogram, The average number of pixels of each component amount in the third color histogram.
  • color similarity may also be calculated in other ways.
  • Chi-Square can be used.
  • the first color similarity can be calculated according to the following formula:
  • Color similarity can also be calculated according to the Intersection algorithm. Specifically, the first color similarity can be calculated according to the following formula:
  • Color similarity can also be calculated from Bhattacharyya.
  • the first color similarity can be calculated according to the following formula:
  • N is the number of histogram bins.
  • the determining module 30 is configured to determine the component to be tested as the first color similarity is less than the second color similarity.
  • the component may be dirty for some reason, that is, a non-polar region of the component may have a color similar to that of the polar region.
  • the image of the polar region may be separately calculated. a third color similarity of the pre-stored polar symmetric region reference image, and a fourth color similarity of the polar symmetric region image and the pre-stored polar symmetric region reference image; if the first color similarity is less than the a second color similarity, and the third color similarity is greater than the fourth color similarity, and determining the component to be tested.
  • the component reverse component detecting system of the present invention has a one-to-one correspondence with the component reverse component detecting method of the present invention, and the technical features and the beneficial effects thereof described in the embodiment of the component reverse component detecting method are applicable to the embodiment of the component reverse 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)
  • Quality & Reliability (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Image Analysis (AREA)

Abstract

一种元件反件检测方法和系统，其中方法包括以下步骤：获取电路板上待测元件的极性区域图像和极性对称区域图像（S1）；其中，极性区域为安装正确时所述待测元件的电极在所述电路板上的区域，极性对称区域为反件时所述电极在电路板上的区域；分别计算所述极性区域图像与预存的极性区域参考图像的第一颜色相似度，以及所述极性对称区域图像与所述极性区域参考图像的第二颜色相似度（S2）；若所述第一颜色相似度小于所述第二颜色相似度，判定所述待测元件反件（S3）。上述元件反件检测方法和系统无需大量的训练样本，只需要获取元件的极性区域和极性对称区域，操作简单，识别率高，检测效果较好。

Description

元件反件检测方法和系统 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反件检测方法和系统。
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，反件检测是指对二极管、电容、插座等有极性的元件进行检测，判断其在电路板中是否存在反向的现象。
目前，元件的反件检测主要采用智能方法，即利用深度学习的方法对大量样本进行训练，得到分类模型。深度学习是机器学习研究的一个新领域，其目的是模拟人脑的机制来解释数据、发现数据的分布式特征表示。不同的学习框架下建立的学习模型也是不同的。例如，卷积神经网络(Convolutional neural networks，简称CNNs)就是一种深度机器学习模型。
利用卷积神经网络训练的元件极性检测分类器，虽然在元件的极性检测方面达到了比较理想的效果，但是也有其自身无法解决的缺点。首先，利用卷积神经网络训练模型时，为了提高模型的准确率、增强模型的鲁棒性，需要大量的训练样本。但是在实际过程中需要耗费大量的人力、时间采集样本；当采集较多的训练样本后，也需要耗费大量的人力和时间进行数据标注。即使如此，也很难采集到足够多的负样本。另外，利用卷积神经网络训练的元件极性检测模型，对于已知的元件拥有很高的识别率，能够达到很好的检测效果。但是对于未知的元件，即训练样本中不存在的元件，元件极性检测模型的识别率下降，经常会发生误报和漏报。
综上所述，现有的元件极性检测方式检测效果较差。
发明内容
基于此，有必要针对现有的元件极性检测方式检测效果较差的问题，提供一种元件反件检测方法和系统。
一种元件反件检测方法，包括以下步骤：
获取电路板上待测元件的极性区域图像和极性对称区域图像；其中，极性区域为安装正确时所述待测元件的电极在所述电路板上的区域，极性对称区域为反件时所述电极在电路板上的区域；
分别计算所述极性区域图像与预存的极性区域参考图像的第一颜色相似度，以及所述极性对称区域图像与所述极性区域参考图像的第二颜色相似度；
若所述第一颜色相似度小于所述第二颜色相似度，判定所述待测元件反件。
一种元件反件检测系统，包括：
获取模块，用于获取电路板上待测元件的极性区域图像和极性对称区域图像；其中，极性区域为安装正确时所述待测元件的电极在所述电路板上的区域，极性对称区域为反件时所述电极在电路板上的区域；
计算模块，用于分别计算所述极性区域图像与预存的极性区域参考图像的第一颜色相似度，以及所述极性对称区域图像与所述极性区域参考图像的第二颜色相似度；
判断模块，用于若所述第一颜色相似度小于所述第二颜色相似度，判定所述待测元件反件。
上述元件反件检测方法和系统，通过计算待测元件的极性区域图像与极性区域参考图像的第一颜色相似度，计算待测元件的极性对称区域图像与所述极性区域参考图像的第二颜色相似度，当所述第一颜色相似度小于所述第二颜色相似度时，表明待测元件的极性对称区域图像与极性区域参考图像之间的差异较小，而待测元件的极性区域图像与极性区域参考图像之间的差异较大，从而判定元件反件。上述元件反件检测方法和系统无需大量的训练样本，只需要获取元件的极性区域和极性对称区域，操作简单，识别率高，检测效果较好。
附图说明
图1为一个实施例的元件反件检测方法的流程图；
图2为极性区域与极性对称区域的示意图；
图3为模板匹配方法的示意图；
图4为HSV颜色空间示意图；
图5为一个实施例的元件反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元件反件检测方法和系统的实施例进行说明。
图1为一个实施例的元件反件检测方法的流程图。如图1所示，所述元件反件检测方法可包括以下步骤：
S1，获取电路板上待测元件的极性区域图像和极性对称区域图像；其中，极性区域为安装正确时所述待测元件的电极在所述电路板上的区域，极性对称区域为反件时所述电极在电路板上的区域；
本发明所述的极性区域和极性对称区域的示意图如图2所示。
可以先获取所述电路板的图像，再从所述电路板的图像中定位极性区域和极性对称区域，并从所述电路板的图像中分别截取所述极性区域和极性对称区域对应的图像，设为极性区域图像和极性对称区域图像。
在一个实施例中，可以通过模板匹配的方法获取所述极性区域图像。模板匹配方法如图3所示。具体地，可以从所述电路板的图像中选取与所述极性区域参考图像相匹配的第一图像区域；根据所述第一图像区域与所述极性区域参考图像中各个像素点的像素值，计算所述图像区域与所述极性区域参考图像的第一像素相似度；并将第一像素相似度大于或等于预设的第一像素相似度阈值的第一图像区域设为所述极性区域图像。所述极性区域参考图像可以预先存储在系统的存储区域中，并在获取所述极性区域图像时从所述存储区域中调用。
若所述第一像素相似度小于预设的第一像素相似度阈值，可以将所述电路板的图像中与所述第一图像区域相邻的区域设为所述第一图像区域，并重复计算所述第一图像区域与所述极性区域参考图像的第一像素相似度的步骤。其中，与所述第一图像区域相邻的区域是在所述电路板的图像中将所述第一图像区域向x轴和y轴分别移动若干个像素点所得的区域。在上述步骤中，每次移动的像素点可以是一个像素点，也可以是多个像素点，移动的距离可以根据实际需要设定。
类似地，也可以通过模板匹配的方法获取所述极性对称区域图像。具体地，可以从所述电路板的图像中选取与所述极性对称区域参考图像相匹配的第二图像区域；根据所述第 二图像区域与所述极性对称区域参考图像中各个像素点的像素值，计算所述图像区域与所述极性对称区域参考图像的第二像素相似度；并将第二像素相似度大于或等于预设的第二像素相似度阈值的第二图像区域设为所述极性对称区域图像。所述极性对称区域参考图像可以预先存储在系统的存储区域中，并在获取所述极性区域图像时从所述存储区域中调用。
若所述第二像素相似度小于预设的第二像素相似度阈值，可以将所述电路板的图像中与所述第二图像区域相邻的区域设为所述第二图像区域，并重复计算所述第二图像区域与所述极性区域参考图像的第二像素相似度的步骤。其中，与所述第二图像区域相邻的区域是在所述电路板的图像中将所述第二图像区域向x轴和y轴分别移动若干个像素点所得的区域。在上述步骤中，每次移动的像素点可以是一个像素点，也可以是多个像素点，移动的距离可以根据实际需要设定。
还可以根据其他方式获取所述极性区域图像与所述极性对称区域图像。
在上述获取所述极性区域图像与所述极性对称区域图像的实施例中，可以根据如下公式计算所述第一像素相似度和第二像素相似度：
Figure PCTCN2016113129-appb-000001
式中，R(x,y)是所述图像区域与所述极性区域参考图像中坐标为(x,y)的像素点的像素相似度，T(x,y)为所述极性区域参考图像中坐标为(x,y)的像素点的像素值，I(x,y)为所述图像区域中坐标为(x,y)的像素点的像素值。
所述第一像素相似度阈值可以根据实际需要来设定。例如，可以设为0.8，或者设为0.9，或设为其他数值。所述第一像素相似度阈值越大，图像获取的精确度越高。
S2，分别计算所述极性区域图像与预存的极性区域参考图像的第一颜色相似度，以及所述极性对称区域图像与所述极性区域参考图像的第二颜色相似度；
当待测元件的极性区域和极性对称区域具有鲜明的颜色差别时，可以根据颜色相似度来检测元件是否反件。为了更直观地比较待测元件的图像与参考图像，可以比较测元件的图像与参考图像的颜色直方图。
在比较之前，还可以先将待测元件的图像转换到HSV颜色空间。HSV颜色空间示意 图如图4所示。在图像处理中，最常用的颜色空间是RGB模型，常用于颜色显示和图像处理等。而HSV模型，是一种针对用户观感的颜色模型，侧重于色彩的表示。其中，R、G、B分别指红、绿、蓝三种颜色；H指色相，取值范围为0～360度，用来表示颜色的类别，如红色是0度绿色是120度、蓝色是240度；S指饱和度，取值范围为0％～100％，用来表示颜色的鲜艳程度，如灰色的饱和度为0％，大红(255，0，0)的饱和度为100％；V指亮度，取值范围是0％～100％，用来表示颜色的明暗程度，如黑色的亮度为0％，白色的亮度为100％。相对于RGB空间，HSV空间能够非常直观地表达色彩的明暗、色调和鲜艳程度。
可根据如下公式将所述极性区域图像的像素值转换到HSV颜色空间：
Figure PCTCN2016113129-appb-000002
其中，max＝max(R,G,B)；
min＝min(R,G,B)；
V＝max；
S＝(max-min)/max；
H＝H+360,ifH<0
0≤V≤1,0≤S≤1,0≤H≤360；
R、G和B为RGB空间的颜色分量。
例如，在计算所述极性区域图像与预存的极性区域参考图像的第一颜色相似度时，可以获取所述极性区域图像的第一颜色直方图，根据所述第一颜色直方图中各组分量的像素点的数量与所述极性区域参考图像的第二颜色直方图中相应分量的像素点的数量，计算所述第一颜色相似度。
类似地，在计算所述极性对称区域图像与所述极性区域参考图像的第二颜色相似度时，可以获取所述极性对称区域图像的第三颜色直方图，根据所述第三颜色直方图中各组分量的像素点的数量与所述第二颜色直方图中相应分量的像素点的数量，计算所述第二颜色相似度。
其中，所述颜色直方图可以是H-S通道的颜色直方图。
在一个实施例中，可根据如下公式计算所述第一颜色相似度：
Figure PCTCN2016113129-appb-000003
式中，d1为第一颜色相似度，H1(I)为第一颜色直方图中第I组分量的像素点的数量，
Figure PCTCN2016113129-appb-000004
为第一颜色直方图中各组分量的像素点的平均数量，H1'(I)为第二颜色直方图中第I组分量的像素点的数量，
Figure PCTCN2016113129-appb-000005
为第二颜色直方图中各组分量的像素点的平均数量；
类似地，可根据如下公式计算所述第二颜色相似度：
Figure PCTCN2016113129-appb-000006
式中，d2为第二颜色相似度，H2(I)为第三颜色直方图中第I组分量的像素点的数量，
Figure PCTCN2016113129-appb-000007
为第三颜色直方图中各组分量的像素点的平均数量。
在其他实施例中，还可以根据其他方式计算颜色相似度。例如，可采用Chi-Square(卡方检验)方式。以计算第一颜色相似度为例，可根据如下公式计算第一颜色相似度：
Figure PCTCN2016113129-appb-000008
还可以根据Intersection算法计算颜色相似度。具体地，可以根据如下公式计算第一颜色相似度：
Figure PCTCN2016113129-appb-000009
还可以根据Bhattacharyya(巴氏距离)计算颜色相似度。具体地，可以根据如下公式计算第一颜色相似度：
Figure PCTCN2016113129-appb-000010
其中，N为直方图bin的数目。
计算第二颜色相似度的方式类似，此处不再赘述。
S3，若所述第一颜色相似度小于所述第二颜色相似度，判定所述待测元件反件。
在一个实施例中，元件由于某种原因会出现脏污，即元件的非极性区域可能会出现与 极性区域相似的颜色，为了保证正确率，还可以分别计算所述极性区域图像与预存的极性对称区域参考图像的第三颜色相似度，以及所述极性对称区域图像与预存的极性对称区域参考图像的第四颜色相似度；若所述第一颜色相似度小于所述第二颜色相似度，且所述第三颜色相似度大于所述第四颜色相似度，判定所述待测元件反件。
计算第三颜色相似度和第四颜色相似度的方式可与上述计算第一颜色相似度和第二颜色相似度的方式类似，此处不再赘述。
本发明的元件反件检测方法具有以下优点：
(1)无需大量的训练样本，也无需耗费大量的人力和时间进行数据标注，只需要获取元件的极性区域和极性对称区域，简单有效，降低了人力成本，检测效率高；
(2)采用颜色直方图，能够直观地检测出元件是否反件，识别率高，检测效果较好。
(3)可以实现元件极性的自动检测，进一步降低了人力成本，提高了检测效率。
(4)通过将待测元件的极性区域图像和极性对称区域与极性区域的参考图像进行比较，同时将待测元件的极性区域图像和极性对称区域与极性对称区域的参考图像进行比较，进一步提高了识别率，降低了误报和漏报的概率。
与所述元件反件检测方法相对应地，本发明还提供一种元件反件检测系统，如图2所示，所述元件反件检测系统可包括：
获取模块10，用于获取电路板上待测元件的极性区域图像和极性对称区域图像；其中，极性区域为安装正确时所述待测元件的电极在所述电路板上的区域，极性对称区域为反件时所述电极在电路板上的区域；
本发明所述的极性区域和极性对称区域的示意图如图2所示。
可以先获取所述电路板的图像，再从所述电路板的图像中定位极性区域和极性对称区域，并从所述电路板的图像中分别截取所述极性区域和极性对称区域对应的图像，设为极性区域图像和极性对称区域图像。
在一个实施例中，可以通过模板匹配的方法获取所述极性区域图像。模板匹配方法如图3所示。具体地，可以从所述电路板的图像中选取与所述极性区域参考图像相匹配的第一图像区域；根据所述第一图像区域与所述极性区域参考图像中各个像素点的像素值，计算所述图像区域与所述极性区域参考图像的第一像素相似度；并将第一像素相似度大于或等于预设的第一像素相似度阈值的第一图像区域设为所述极性区域图像。所述极性区域参 考图像可以预先存储在系统的存储区域中，并在获取所述极性区域图像时从所述存储区域中调用。
若所述第一像素相似度小于预设的第一像素相似度阈值，可以将所述电路板的图像中与所述第一图像区域相邻的区域设为所述第一图像区域，并重复计算所述第一图像区域与所述极性区域参考图像的第一像素相似度的步骤。其中，与所述第一图像区域相邻的区域是在所述电路板的图像中将所述第一图像区域向x轴和y轴分别移动若干个像素点所得的区域。其中，每次移动的像素点可以是一个像素点，也可以是多个像素点，移动的距离可以根据实际需要设定。
类似地，也可以通过模板匹配的方法获取所述极性对称区域图像。具体地，可以从所述电路板的图像中选取与所述极性对称区域参考图像相匹配的第二图像区域；根据所述第二图像区域与所述极性对称区域参考图像中各个像素点的像素值，计算所述图像区域与所述极性对称区域参考图像的第二像素相似度；并将第二像素相似度大于或等于预设的第二像素相似度阈值的第二图像区域设为所述极性对称区域图像。所述极性对称区域参考图像可以预先存储在系统的存储区域中，并在获取所述极性区域图像时从所述存储区域中调用。
若所述第二像素相似度小于预设的第二像素相似度阈值，可以将所述电路板的图像中与所述第二图像区域相邻的区域设为所述第二图像区域，并重复计算所述第二图像区域与所述极性区域参考图像的第二像素相似度的步骤。其中，与所述第二图像区域相邻的区域是在所述电路板的图像中将所述第二图像区域向x轴和y轴分别移动若干个像素点所得的区域。其中，每次移动的像素点可以是一个像素点，也可以是多个像素点，移动的距离可以根据实际需要设定。
还可以根据其他方式获取所述极性区域图像与所述极性对称区域图像。
在上述获取所述极性区域图像与所述极性对称区域图像的实施例中，可以根据如下公式计算所述第一像素相似度和第二像素相似度：
Figure PCTCN2016113129-appb-000011
式中，R(x,y)是所述图像区域与所述极性区域参考图像中坐标为(x,y)的像素点的 像素相似度，T(x,y)为所述极性区域参考图像中坐标为(x,y)的像素点的像素值，I(x,y)为所述图像区域中坐标为(x,y)的像素点的像素值。
所述第一像素相似度阈值可以根据实际需要来设定。例如，可以设为0.8，或者设为0.9，或设为其他数值。所述第一像素相似度阈值越大，图像获取的精确度越高。
计算模块20，用于分别计算所述极性区域图像与预存的极性区域参考图像的第一颜色相似度，以及所述极性对称区域图像与所述极性区域参考图像的第二颜色相似度；
当待测元件的极性区域和极性对称区域具有鲜明的颜色差别时，可以根据颜色相似度来检测元件是否反件。为了更直观地比较待测元件的图像与参考图像，可以比较测元件的图像与参考图像的颜色直方图。
在比较之前，还可以先将待测元件的图像转换到HSV颜色空间。HSV颜色空间示意图如图4所示。在图像处理中，最常用的颜色空间是RGB模型，常用于颜色显示和图像处理等。而HSV模型，是一种针对用户观感的颜色模型，侧重于色彩的表示。其中，R、G、B分别指红、绿、蓝三种颜色；H指色相，取值范围为0～360度，用来表示颜色的类别，如红色是0度绿色是120度、蓝色是240度；S指饱和度，取值范围为0％～100％，用来表示颜色的鲜艳程度，如灰色的饱和度为0％，大红(255，0，0)的饱和度为100％；V指亮度，取值范围是0％～100％，用来表示颜色的明暗程度，如黑色的亮度为0％，白色的亮度为100％。相对于RGB空间，HSV空间能够非常直观地表达色彩的明暗、色调和鲜艳程度。
可根据如下公式将所述极性区域图像的像素值转换到HSV颜色空间：
Figure PCTCN2016113129-appb-000012
其中，max＝max(R,G,B)；
min＝min(R,G,B)；
V＝max；
S＝(max-min)/max；
H＝H+360,ifH<0
0≤V≤1,0≤S≤1,0≤H≤360；
R、G和B为RGB空间的颜色分量。
例如，在计算所述极性区域图像与预存的极性区域参考图像的第一颜色相似度时，可以获取所述极性区域图像的第一颜色直方图，根据所述第一颜色直方图中各组分量的像素点的数量与所述极性区域参考图像的第二颜色直方图中相应分量的像素点的数量，计算所述第一颜色相似度。
类似地，在计算所述极性对称区域图像与所述极性区域参考图像的第二颜色相似度时，可以获取所述极性对称区域图像的第三颜色直方图，根据所述第三颜色直方图中各组分量的像素点的数量与所述第二颜色直方图中相应分量的像素点的数量，计算所述第二颜色相似度。
其中，所述颜色直方图可以是H-S通道的颜色直方图。
在一个实施例中，可根据如下公式计算所述第一颜色相似度：
Figure PCTCN2016113129-appb-000013
式中，d1为第一颜色相似度，H1(I)为第一颜色直方图中第I组分量的像素点的数量，
Figure PCTCN2016113129-appb-000014
为第一颜色直方图中各组分量的像素点的平均数量，H1'(I)为第二颜色直方图中第I组分量的像素点的数量，
Figure PCTCN2016113129-appb-000015
为第二颜色直方图中各组分量的像素点的平均数量；
类似地，可根据如下公式计算所述第二颜色相似度：
Figure PCTCN2016113129-appb-000016
式中，d2为第二颜色相似度，H2(I)为第三颜色直方图中第I组分量的像素点的数量，
Figure PCTCN2016113129-appb-000017
为第三颜色直方图中各组分量的像素点的平均数量。
在其他实施例中，还可以根据其他方式计算颜色相似度。例如，可采用Chi-Square(卡方检验)方式。以计算第一颜色相似度为例，可根据如下公式计算第一颜色相似度：
Figure PCTCN2016113129-appb-000018
还可以根据Intersection算法计算颜色相似度。具体地，可以根据如下公式计算第一颜色相似度：
Figure PCTCN2016113129-appb-000019
还可以根据Bhattacharyya(巴氏距离)计算颜色相似度。具体地，可以根据如下公式计算第一颜色相似度：
Figure PCTCN2016113129-appb-000020
式中，N为直方图bin的数目。
计算第二颜色相似度的方式类似，此处不再赘述。
判断模块30，用于若所述第一颜色相似度小于所述第二颜色相似度，判定所述待测元件反件。
在一个实施例中，元件由于某种原因会出现脏污，即元件的非极性区域可能会出现与极性区域相似的颜色，为了保证正确率，还可以分别计算所述极性区域图像与预存的极性对称区域参考图像的第三颜色相似度，以及所述极性对称区域图像与预存的极性对称区域参考图像的第四颜色相似度；若所述第一颜色相似度小于所述第二颜色相似度，且所述第三颜色相似度大于所述第四颜色相似度，判定所述待测元件反件。
本发明的元件反件检测系统具有以下优点：
(1)无需大量的训练样本，也无需耗费大量的人力和时间进行数据标注，只需要获取元件的极性区域和极性对称区域，简单有效，降低了人力成本，检测效率高；
(2)采用颜色直方图，能够直观地检测出元件是否反件，识别率高，检测效果较好。
(3)可以实现元件极性的自动检测，进一步降低了人力成本，提高了检测效率。
(4)通过将待测元件的极性区域图像和极性对称区域与极性区域的参考图像进行比较，同时将待测元件的极性区域图像和极性对称区域与极性对称区域的参考图像进行比较，进一步提高了识别率，降低了误报和漏报的概率。
本发明的元件反件检测系统与本发明的元件反件检测方法一一对应，在上述元件反件检测方法的实施例阐述的技术特征及其有益效果均适用于元件反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特征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特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛 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. 一种元件反件检测方法，其特征在于，包括以下步骤：
    获取电路板上待测元件的极性区域图像和极性对称区域图像；其中，极性区域为安装正确时所述待测元件的电极在所述电路板上的区域，极性对称区域为反件时所述电极在电路板上的区域；
    分别计算所述极性区域图像与预存的极性区域参考图像的第一颜色相似度，以及所述极性对称区域图像与所述极性区域参考图像的第二颜色相似度；
    若所述第一颜色相似度小于所述第二颜色相似度，判定所述待测元件反件。
  2. 根据权利要求1所述的元件反件检测方法，其特征在于，获取元件的极性区域图像的步骤包括：
    从所述电路板的图像中选取与所述极性区域参考图像相匹配的图像区域；
    根据所述图像区域与所述极性区域参考图像中各个像素点的像素值，计算所述图像区域与所述极性区域参考图像的像素相似度；
    将像素相似度大于或等于预设的像素相似度阈值的图像区域设为所述极性区域图像。
  3. 根据权利要求2所述的元件反件检测方法，其特征在于，在计算所述图像区域与所述极性区域参考图像的像素相似度之后，还包括以下步骤：
    若所述像素相似度小于预设的像素相似度阈值，将所述电路板的图像中与所述图像区域相邻的区域设为所述图像区域；
    重复计算所述图像区域与所述极性区域参考图像的像素相似度的步骤；
    其中，与所述图像区域相邻的区域是在所述电路板的图像中将所述图像区域向x轴和y轴分别移动若干个像素点所得的区域。
  4. 根据权利要求2所述的元件反件检测方法，其特征在于，计算所述图像区域与所述极性区域参考图像的像素相似度的步骤包括：
    根据如下公式计算所述像素相似度：
    Figure PCTCN2016113129-appb-100001
    式中，R(x,y)是所述图像区域与所述极性区域参考图像中坐标为(x,y)的像素点的 像素相似度，T(x,y)为所述极性区域参考图像中坐标为(x,y)的像素点的像素值，I(x,y)为所述图像区域中坐标为(x,y)的像素点的像素值。
  5. 根据权利要求1所述的元件反件检测方法，其特征在于，计算所述极性区域图像与预存的极性区域参考图像的第一颜色相似度的步骤包括：
    获取所述极性区域图像的第一颜色直方图；
    根据所述第一颜色直方图中各组分量的像素点的数量与所述极性区域参考图像的第二颜色直方图中相应分量的像素点的数量，计算所述第一颜色相似度；
    计算所述极性对称区域图像与所述极性区域参考图像的第二颜色相似度的步骤包括：
    获取所述极性对称区域图像的第三颜色直方图；
    根据所述第三颜色直方图中各组分量的像素点的数量与所述第二颜色直方图中相应分量的像素点的数量，计算所述第二颜色相似度。
  6. 根据权利要求5所述的元件反件检测方法，其特征在于，根据所述第一颜色参数和第二颜色参数计算所述第一颜色相似度的步骤包括：
    根据如下公式计算所述第一颜色相似度：
    Figure PCTCN2016113129-appb-100002
    式中，d1为第一颜色相似度，H1(I)为第一颜色直方图中第I组分量的像素点的数量，
    Figure PCTCN2016113129-appb-100003
    为第一颜色直方图中各组分量的像素点的平均数量，H1'(I)为第二颜色直方图中第I组分量的像素点的数量，
    Figure PCTCN2016113129-appb-100004
    为第二颜色直方图中各组分量的像素点的平均数量；
    根据所述第三颜色参数和第二颜色参数计算所述第二颜色相似度的步骤包括：
    根据如下公式计算所述第二颜色相似度：
    Figure PCTCN2016113129-appb-100005
    式中，d2为第二颜色相似度，H2(I)为第三颜色直方图中第I组分量的像素点的数量，
    Figure PCTCN2016113129-appb-100006
    为第三颜色直方图中各组分量的像素点的平均数量。
  7. 根据权利要求1所述的元件反件检测方法，其特征在于，在计算所述极性区域图 像与所述极性区域参考图像的第一颜色相似度之前，还包括以下步骤：
    将所述极性区域图像的像素值转换到HSV颜色空间。
  8. 根据权利要求7所述的元件反件检测方法，其特征在于，将所述极性区域图像的像素值转换到HSV颜色空间的步骤包括：
    根据如下公式将所述极性区域图像的像素值转换到HSV颜色空间：
    Figure PCTCN2016113129-appb-100007
    其中，max＝max(R,G,B)；
    min＝min(R,G,B)；
    V＝max；
    S＝(max-min)/max；
    H＝H+360,ifH<0
    0≤V≤1,0≤S≤1,0≤H≤360；
    R、G和B为RGB空间的颜色分量。
  9. 根据权利要求1所述的元件反件检测方法，其特征在于，判定所述待测元件反件前，还包括以下步骤：
    分别计算所述极性区域图像与预存的极性对称区域参考图像的第三颜色相似度，以及所述极性对称区域图像与预存的极性对称区域参考图像的第四颜色相似度；
    若所述第一颜色相似度小于所述第二颜色相似度，且所述第三颜色相似度大于所述第四颜色相似度，判定所述待测元件反件。
  10. 一种元件反件检测系统，其特征在于，包括：
    获取模块，用于获取电路板上待测元件的极性区域图像和极性对称区域图像；其中，极性区域为安装正确时所述待测元件的电极在所述电路板上的区域，极性对称区域为反件时所述电极在电路板上的区域；
    计算模块，用于分别计算所述极性区域图像与预存的极性区域参考图像的第一颜色相似度，以及所述极性对称区域图像与所述极性区域参考图像的第二颜色相似度；
    判断模块，用于若所述第一颜色相似度小于所述第二颜色相似度，判定所述待测元件 反件。
PCT/CN2016/113129 2016-07-13 2016-12-29 元件反件检测方法和系统 Ceased WO2018010386A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610556307.9 2016-07-13
CN201610556307.9A CN106204602B (zh) 2016-07-13 2016-07-13 元件反件检测方法和系统

Publications (1)

Publication Number Publication Date
WO2018010386A1 true WO2018010386A1 (zh) 2018-01-18

Family

ID=57475884

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/113129 Ceased WO2018010386A1 (zh) 2016-07-13 2016-12-29 元件反件检测方法和系统

Country Status (2)

Country Link
CN (1) CN106204602B (zh)
WO (1) WO2018010386A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113012088A (zh) * 2019-12-03 2021-06-22 浙江大搜车软件技术有限公司 一种电路板故障检测及孪生网络的训练方法、装置和设备

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106204602B (zh) * 2016-07-13 2019-02-01 广州视源电子科技股份有限公司 元件反件检测方法和系统
CN107180424B (zh) * 2017-04-24 2020-04-28 睿视智觉（厦门）科技有限公司 一种电容计数设备及方法
CN107895064B (zh) 2017-10-19 2020-01-10 上海望友信息科技有限公司 元器件极性检测方法、系统、计算机可读存储介质及设备
CN109858441A (zh) * 2019-01-30 2019-06-07 广州轨道交通建设监理有限公司 一种用于施工现场的异常状态监测方法与装置
CN109903275B (zh) * 2019-02-13 2021-05-18 湖北工业大学 一种基于自适应多尺度滤波和直方图对比的酒醅发霉区域检测方法
CN112767396B (zh) * 2021-04-07 2021-07-20 深圳中科飞测科技股份有限公司 缺陷的检测方法、装置和计算机可读存储介质
CN120235225B (zh) * 2025-03-07 2026-01-06 南京禄口国际机场空港科技有限公司 一种文本-图像增强的多模态知识图谱嵌入方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050084148A1 (en) * 2003-10-20 2005-04-21 Miranda Jose J. M. Inspection systems and methods
CN102479321A (zh) * 2010-11-26 2012-05-30 鸿富锦精密工业（深圳）有限公司 差异影像自动识别系统及方法
CN104459421A (zh) * 2014-12-30 2015-03-25 广州视源电子科技股份有限公司 一种二极管极性检测方法与系统
CN105139399A (zh) * 2015-08-25 2015-12-09 广州视源电子科技股份有限公司 一种二极管极性检测方法及装置
CN106204602A (zh) * 2016-07-13 2016-12-07 广州视源电子科技股份有限公司 元件反件检测方法和系统

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS587164A (ja) * 1981-07-03 1983-01-14 Minolta Camera Co Ltd トナ−濃度制御方法
CN1318839C (zh) * 2002-11-28 2007-05-30 威光机械工程股份有限公司 印刷电路板上瑕疵组件的自动光学检测方法
CN101915769B (zh) * 2010-06-29 2012-03-28 华南理工大学 一种印刷电路板中带电阻元件的自动光学检测方法
CN102831381B (zh) * 2011-06-15 2016-05-04 罗普特（厦门）科技集团有限公司 影像差异比较系统及方法
CN103675588B (zh) * 2013-11-20 2016-01-20 中国矿业大学 印刷电路元件极性的机器视觉检测方法及设备
CN105184778B (zh) * 2015-08-25 2018-04-24 广州视源电子科技股份有限公司 一种检测方法及装置
CN105260740B (zh) * 2015-09-23 2019-03-29 广州视源电子科技股份有限公司 一种元件识别方法及装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050084148A1 (en) * 2003-10-20 2005-04-21 Miranda Jose J. M. Inspection systems and methods
CN102479321A (zh) * 2010-11-26 2012-05-30 鸿富锦精密工业（深圳）有限公司 差异影像自动识别系统及方法
CN104459421A (zh) * 2014-12-30 2015-03-25 广州视源电子科技股份有限公司 一种二极管极性检测方法与系统
CN105139399A (zh) * 2015-08-25 2015-12-09 广州视源电子科技股份有限公司 一种二极管极性检测方法及装置
CN106204602A (zh) * 2016-07-13 2016-12-07 广州视源电子科技股份有限公司 元件反件检测方法和系统

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113012088A (zh) * 2019-12-03 2021-06-22 浙江大搜车软件技术有限公司 一种电路板故障检测及孪生网络的训练方法、装置和设备

Also Published As

Publication number Publication date
CN106204602A (zh) 2016-12-07
CN106204602B (zh) 2019-02-01

Similar Documents

Publication Publication Date Title
WO2018010386A1 (zh) 元件反件检测方法和系统
CN106251333B (zh) 元件反件检测方法和系统
CN109520706B (zh) 一种汽车保险丝盒的螺孔坐标提取方法
CN114627089B (zh) 缺陷识别方法、装置、计算机设备及计算机可读存储介质
US8700498B2 (en) Feature analyzing apparatus for a surface of an object
CN107392890B (zh) 一种fpc铜线表面氧化缺陷检测方法及其检测系统
CN106370671A (zh) 基于机器视觉的pcb上元器件检测系统及方法
TW201702586A (zh) 光學薄膜缺陷辨識方法及其系統
CN109946304A (zh) 基于特征匹配的零件表面缺陷在线检测系统及检测方法
CN105913093A (zh) 一种用于文字识别处理的模板匹配方法
CN108520514A (zh) 基于计算机视觉的印刷电路板电子元器一致性检测方法
CN113096060B (zh) 一种颜色异常灯珠的定位方法、装置及存储介质
CN104899871A (zh) 一种ic元件焊点空焊检测方法
CN109166111B (zh) 一种电子墨水屏缺陷检测方法及系统
CN105303573B (zh) 金针类元件的引脚检测方法和系统
CN110910372A (zh) 基于深度卷积神经网络的匀光板缺陷检测方法
CN110307903B (zh) 一种家禽特定部位无接触温度动态测量的方法
WO2017181722A1 (zh) 元件漏检方法和系统
CN117351499A (zh) 一种分合指示状态识别方法、系统、计算机设备和介质
CN112750113A (zh) 基于深度学习和直线检测的玻璃瓶缺陷检测方法及装置
CN106558044A (zh) 影像模组的解像力测量方法
CN108875749A (zh) 确定颜色的方法、装置和电子设备
CN113866182A (zh) 一种用于检测显示模组缺陷的检测方法及系统
CN119274001A (zh) 基于改进msrcr与随机森林的试纸变色定量识别方法
CN108563997A (zh) 一种建立人脸检测模型、人脸识别的方法和装置

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: 16908708

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: 16908708

Country of ref document: EP

Kind code of ref document: A1