CN108830840A - A kind of active intelligent detecting method of circuit board defect and its application - Google Patents
A kind of active intelligent detecting method of circuit board defect and its application Download PDFInfo
- Publication number
- CN108830840A CN108830840A CN201810549271.0A CN201810549271A CN108830840A CN 108830840 A CN108830840 A CN 108830840A CN 201810549271 A CN201810549271 A CN 201810549271A CN 108830840 A CN108830840 A CN 108830840A
- Authority
- CN
- China
- Prior art keywords
- defect
- circuit board
- image
- detector
- detection
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/251—Fusion techniques of input or preprocessed data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30148—Semiconductor; IC; Wafer
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Investigating Or Analyzing Materials Using Thermal Means (AREA)
Abstract
The present invention discloses a kind of active intelligent detecting method of circuit board defect, includes the following steps:(1) complex probe module is constructed;(2) circuit board defect detection is carried out according to board type and testing requirements, forms defect image;(3) whole defect images of acquisition are carried out compound;(4) surface and internal defect in autonomous discriminatory analysis circuit board or chip are carried out to each defect image after fusion positioning, forms unified spatial defects image;(5) the spatial defects image that output is formed.Circuit board defect detection method of the invention has a variety of different types of detection sensors, any one group of open defect detector and internal flaw detector can be independently selected according to the type difference of circuit board, different defect images is merged again, calculate defect characteristic, and the output of spatial defects image is formed, reach the purpose of circuit board comprehensive physical examination online.
Description
Technical field
The invention belongs to industrial nondestructive testing fields, and in particular to a kind of defect inspection method towards circuit board.
Background technique
With the demand for development that the raising increasingly of the highly reliable demand of electronic equipment and circuit board detecting automate, tradition
Detection means increasingly south meet the demand, there is an urgent need to the online defects detection machine people of high-precision architecture to reach modern
The testing requirements of change.
At present in terms of circuit board defect context of detection remains in appearance detection (AOI) mostly both at home and abroad.And circuit board
Defect type includes visually observed open defect (scratch of such as circuit board, flash, bending, short circuit, open circuit), equally
There is also internal flaw (such as circuit board air entrapment, via hole, rosin joint, missing solder, chip interior defect), height defects (as drawn
Point) etc., these defects by the appearance of circuit board be difficult to observe Lai.
Although in internal context of detection there is also having the methods of flying probe, X-ray and active thermal imaging detection equipment, so
And cannot detect with appearance and effectively combine, and that there are instant performances is poor, is unable to the disadvantages of intelligent decision, it relies on manually, is difficult
Industrial scale application, there has been no effective means at present for online physical examination comprehensively for circuit board.
Summary of the invention
Goal of the invention:Present invention aims in view of the deficiencies of the prior art, provide a kind of active intelligence of circuit board defect
Energy detection method, the inside and outside defect of autonomous detection circuit board, and it is fused to visualization defect image output, reach circuit
The purpose of plate comprehensive physical examination online;
Another object of the present invention is to provide a kind of industrial complex probe cameras using above-mentioned detection method.
Technical solution:The active intelligent detecting method of circuit board defect of the present invention, includes the following steps:
(1) the complex probe module being made of several groups open defect detector and internal flaw detector is constructed;
(2) at least one set of open defect detector and one group of internal flaw is selected to visit according to board type and testing requirements
It surveys device and carries out circuit board defect detection, be respectively formed defect image;
(3) compound to whole defect images progress of step (2) acquisition, more visions are merged and are positioned;
(4) surface and inside in autonomous discriminatory analysis circuit board or chip are carried out to each defect image after fusion positioning
Defect, and carry out spatial defects data fusion modeling, form unified spatial defects image;
(5) the spatial defects image that output is formed, and visualization display.
Further preferably technical solution is the present invention, and the specific method that space defective data merges in step (4) is:
A, training, defect characteristic signal to be sorted, which pre-process, to be waited for total k class characteristics of image, extracts m kind difference defect
Feature;
B, the m kind different characteristic to training objective is inputted into m classifiers respectively, wherein every classifiers include 1
PSVM, and be trained according to tree;When the input of PSVM is greater than two classes, these class labels are uniformly divided as far as possible
It is made decisions at two major classes, until the PSVM input of the bottom is two class labels, end tree training;
C, the m kind different images feature of target to be tested is inputted into m classifiers respectively, the probability for obtaining all PSVM is defeated
Out;Every classifiers have k probability output, and m classifiers share m probability output, calculate the corresponding power of each classifier
Value;
D, the probability output of every kind of label in every classifiers is weighted processing, the maximum label conduct of weighted results
Fusion results output, the blending algorithm are expressed as follows:
F (x)=argmax [∑ wjPij (class=i | input)] 1.;
E, operation is then carried out in a manner of Matrix Multiplication with feature weight vector W, obtain comprehensive weight matrix, i.e.,:
F, matrix w is multiplied with matrix P, establishes WT weight matrix;
G, PF is k × k matrix, diagonal entry:
H, the synthesis weight probability output of jth class label is indicated, classifier final judging result is:
F (x)=argmax (PFi, j) ④。
Preferably, appearance defect detector includes visible detection sensor in step (1);Internal flaw detector includes
Thermal sensation vision-based detection sensor, laser scanning inspection sensor and ultrasonic detection sensor are one such or several.
Preferably, the type and testing requirements of circuit board correspond in step (2), and open defect detector and inside lack
The type for falling into detector independently selects to determine according to board type.
Preferably, blending algorithm is positioned using spatial isomerism visual sensor in step (3), to photopic vision and inside
Vision carries out fusion positioning.
A kind of application of above-mentioned detection method, for building active intelligent industrial complex probe camera.
Preferably, the camera includes detecting module, active Intelligent Fusion processing module and peripheral interface module, institute
Stating detecting module includes at least one set of open defect detector and one group of internal flaw detector, the open defect detector packet
Include visible detection sensor;Internal flaw detector include thermal sensation vision-based detection sensor, laser scanning inspection sensor and
Ultrasonic detection sensor is one such or several;
The active Intelligent Fusion processing module carries out defective data automation synthesis, and carries out spatial defects data and build
Mould forms unified 3D defect image, and is input to defects detection master control system by the peripheral interface module.
Beneficial effect:(1) circuit board defect detection method of the invention has a variety of different types of detection sensors, root
Any one group of open defect detector and internal flaw detector can be independently selected according to the type difference of circuit board, then will be different
Defect image merged, calculate defect characteristic, and formed spatial defects image output, reach circuit board comprehensive body online
The purpose of inspection;Detection process is intelligently completed, it is only necessary to be inputted certain parameter in advance, can be obtained comprehensive defect map of circuit board
Picture is participated in without artificial, therefore industrial scale may be implemented and use;
(2) defect image obtained in the present invention according to detector, carries out autonomous fusion operation, generates unified space and lacks
Image is fallen into, operation is rapid, and time delay≤500ms of data fusion output is detected by data, it is comprehensively online to can be realized circuit board
Formula physical examination;
(3) the open defect detector of complex probe module includes visible detection sensor in the present invention;Internal flaw
Detector include thermal sensation vision-based detection sensor, laser scanning inspection sensor and ultrasonic detection sensor it is one such or
It is several, can open defect to circuit board, internal flaw and height defect carry out complete detection, and effective integration.
Detailed description of the invention
Fig. 1 is the flow chart of the active intelligent detecting method of circuit board defect of the present invention.
Specific embodiment
Technical solution of the present invention is described in detail below by attached drawing, but protection scope of the present invention is not limited to
The embodiment.
Embodiment:A kind of active intelligent detecting method of circuit board defect, includes the following steps:
(1) the complex probe module being made of several groups open defect detector and internal flaw detector is constructed;
(2) at least one set of open defect detector and one group of internal flaw is selected to visit according to board type and testing requirements
It surveys device and carries out circuit board defect detection, be respectively formed defect image;The type and testing requirements of circuit board correspond, and appearance lacks
The type for falling into detector and internal flaw detector independently selects to determine according to board type;
(3) blending algorithm is positioned using spatial isomerism visual sensor, whole defect images that step (2) obtain is carried out
It is compound, it will be seen that light vision and internal vision carry out fusion positioning;
(4) surface and inside in autonomous discriminatory analysis circuit board or chip are carried out to each defect image after fusion positioning
Defect, and carry out spatial defects data fusion modeling, form unified spatial defects image, specific method is:
A, training, defect characteristic signal to be sorted, which pre-process, to be waited for total k class characteristics of image, extracts m kind difference defect
Feature;
B, the m kind different characteristic to training objective is inputted into m classifiers respectively, wherein every classifiers include 1
PSVM, and be trained according to tree;When the input of PSVM is greater than two classes, these class labels are uniformly divided as far as possible
It is made decisions at two major classes, until the PSVM input of the bottom is two class labels, end tree training;
C, the m kind different images feature of target to be tested is inputted into m classifiers respectively, the probability for obtaining all PSVM is defeated
Out;Every classifiers have k probability output, and m classifiers share m probability output, calculate the corresponding power of each classifier
Value;
D, the probability output of every kind of label in every classifiers is weighted processing, the maximum label conduct of weighted results
Fusion results output, the blending algorithm are expressed as follows:
F (x)=argmax [∑ wjPij (class=i | input)] 1.;
E, operation is then carried out in a manner of Matrix Multiplication with feature weight vector W, obtain comprehensive weight matrix, i.e.,:
F, matrix w is multiplied with matrix P, establishes WT weight matrix;
G, PF is k × k matrix, diagonal entry:
H, the synthesis weight probability output of jth class label is indicated, classifier final judging result is:
F (x)=argmax (PFj, j) ④。
(5) the spatial defects image that output is formed, and visualization display.
The active intelligent industrial complex probe camera built using above-mentioned detection method, including detecting module, active intelligence
Energy fusion treatment module and peripheral interface module, the detecting module include inside at least one set of open defect detector and one group
Defect detector, the open defect detector include visible detection sensor;Internal flaw detector includes thermal sensation vision
Detection sensor, laser scanning inspection sensor and ultrasonic detection sensor are one such or several;The active intelligence
Fusion treatment module carries out defective data automation synthesis, and carries out spatial defects data modeling, forms unified 3D defect image,
And defects detection master control system is input to by the peripheral interface module.
As described above, must not be explained although the present invention has been indicated and described referring to specific preferred embodiment
For the limitation to invention itself.It without prejudice to the spirit and scope of the invention as defined in the appended claims, can be right
Various changes can be made in the form and details for it.
Claims (7)
1. a kind of active intelligent detecting method of circuit board defect, which is characterized in that include the following steps:
(1) the complex probe module being made of several groups open defect detector and internal flaw detector is constructed;
(2) at least one set of open defect detector and one group of internal flaw detector are selected according to board type and testing requirements
Circuit board defect detection is carried out, defect image is respectively formed;
(3) compound to whole defect images progress of step (2) acquisition, more visions are merged and are positioned;
(4) surface in autonomous discriminatory analysis circuit board or chip is carried out to each defect image after fusion positioning to lack with internal
It falls into, and carries out spatial defects data fusion modeling, form unified spatial defects image;
(5) the spatial defects image that output is formed, and visualization display.
2. the active intelligent detecting method of circuit board defect according to claim 1, which is characterized in that step (4) is hollow
Between defective data merge specific method be:
A, training, defect characteristic signal to be sorted, which pre-process, to be waited for total k class characteristics of image, extracts m kind difference defect characteristic;
B, the m kind different characteristic to training objective is inputted into m classifiers respectively, wherein every classifiers include 1 PSVM, and
It is trained according to tree;When the input of PSVM is greater than two classes, it is big that these class labels are uniformly divided into two as far as possible
Class makes decisions, until the PSVM input of the bottom is two class labels, end tree training;
C, the m kind different images feature of target to be tested is inputted into m classifiers respectively, obtains the probability output of all PSVM;
Every classifiers have k probability output, and m classifiers share m probability output, calculate the correspondence weight of each classifier;
D, the probability output of every kind of label in every classifiers is weighted processing, the maximum label of weighted results is as fusion
As a result it exports, which is expressed as follows:
F (x)=argmax [∑ wjPij (class=i | input)] 1.;
E, operation is then carried out in a manner of Matrix Multiplication with feature weight vector W, obtain comprehensive weight matrix, i.e.,:
F, matrix w is multiplied with matrix P, establishes WT weight matrix;
G, PF is k × k matrix, diagonal entry:
H, the synthesis weight probability output of jth class label is indicated, classifier final judging result is:
F (x)=argmax (PFj, j) ④。
3. the active intelligent detecting method of circuit board defect according to claim 1, which is characterized in that step (1) China and foreign countries
Seeing defect detector includes visible detection sensor;Internal flaw detector includes that thermal sensation vision-based detection sensor, laser are swept
It retouches detection sensor and ultrasonic detection sensor is one such or several.
4. the active intelligent detecting method of circuit board defect according to claim 3, which is characterized in that electric in step (2)
The type and testing requirements of road plate correspond, and the type of open defect detector and internal flaw detector is according to circuit board class
Type independently selects to determine.
5. the active intelligent detecting method of circuit board defect according to claim 1, which is characterized in that step is adopted in (3)
Blending algorithm is positioned with spatial isomerism visual sensor, fusion positioning is carried out to photopic vision and internal vision.
6. a kind of application of detection method described in Claims 1 to 5 any one, which is characterized in that for building actively intelligence
Industrial complex probe camera.
7. application according to claim 6, which is characterized in that the camera includes detecting module, at active Intelligent Fusion
Module and peripheral interface module are managed, the detecting module includes at least one set of open defect detector and one group of internal flaw detection
Device, the open defect detector include visible detection sensor;Internal flaw detector includes thermal sensation vision-based detection sensing
Device, laser scanning inspection sensor and ultrasonic detection sensor are one such or several;
The active Intelligent Fusion processing module carries out defective data automation synthesis, and carries out spatial defects data modeling, shape
At unified 3D defect image, and defects detection master control system is input to by the peripheral interface module.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810549271.0A CN108830840A (en) | 2018-05-31 | 2018-05-31 | A kind of active intelligent detecting method of circuit board defect and its application |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810549271.0A CN108830840A (en) | 2018-05-31 | 2018-05-31 | A kind of active intelligent detecting method of circuit board defect and its application |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108830840A true CN108830840A (en) | 2018-11-16 |
Family
ID=64145504
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810549271.0A Pending CN108830840A (en) | 2018-05-31 | 2018-05-31 | A kind of active intelligent detecting method of circuit board defect and its application |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108830840A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109829883A (en) * | 2018-12-19 | 2019-05-31 | 歌尔股份有限公司 | Product quality detection method and device |
CN109856167A (en) * | 2018-12-05 | 2019-06-07 | 惠州市骏亚数字技术有限公司 | A kind of non-destructive testing device and its detection method |
WO2020242746A1 (en) * | 2019-05-31 | 2020-12-03 | Advanced Micro Devices, Inc. | Temperature-based adjustments for in-memory matrix multiplication |
CN112345556A (en) * | 2020-11-23 | 2021-02-09 | 兰州大学 | Fault diagnosis system and method for integrated circuit |
CN112836274A (en) * | 2021-02-02 | 2021-05-25 | 南京审计大学 | Data fusion method for tracing and auditing hidden engineering |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1576829A (en) * | 2003-07-09 | 2005-02-09 | 通用电气公司 | System and method for analyzing and identifying flaws in a manufactured part |
CN101661004A (en) * | 2009-07-21 | 2010-03-03 | 湖南大学 | Visible detection method of welding quality of circuit board based on support vector machine |
CN105135591A (en) * | 2015-07-01 | 2015-12-09 | 西安理工大学 | Train air conditioning unit fault diagnosing method based on multi-classification strategy |
CN105184793A (en) * | 2015-09-02 | 2015-12-23 | 广东电网有限责任公司汕尾供电局 | Electric energy meter sample appearance and PCB element detection method |
CN106896780A (en) * | 2017-02-27 | 2017-06-27 | 西安交通大学 | A kind of Cutting Properties of Materials energy integrated evaluating method |
US20170318043A1 (en) * | 2016-04-27 | 2017-11-02 | Korea Advanced Institute Of Science And Technology | Method for detecting network anomaly in distributed software defined networking environment, apparatus therefor, and computer program therefor |
-
2018
- 2018-05-31 CN CN201810549271.0A patent/CN108830840A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1576829A (en) * | 2003-07-09 | 2005-02-09 | 通用电气公司 | System and method for analyzing and identifying flaws in a manufactured part |
CN101661004A (en) * | 2009-07-21 | 2010-03-03 | 湖南大学 | Visible detection method of welding quality of circuit board based on support vector machine |
CN105135591A (en) * | 2015-07-01 | 2015-12-09 | 西安理工大学 | Train air conditioning unit fault diagnosing method based on multi-classification strategy |
CN105184793A (en) * | 2015-09-02 | 2015-12-23 | 广东电网有限责任公司汕尾供电局 | Electric energy meter sample appearance and PCB element detection method |
US20170318043A1 (en) * | 2016-04-27 | 2017-11-02 | Korea Advanced Institute Of Science And Technology | Method for detecting network anomaly in distributed software defined networking environment, apparatus therefor, and computer program therefor |
CN106896780A (en) * | 2017-02-27 | 2017-06-27 | 西安交通大学 | A kind of Cutting Properties of Materials energy integrated evaluating method |
Non-Patent Citations (4)
Title |
---|
T. HANUMANTHA REDDY .ETC: ""Proximal SVM for Face Recognition"", 《JOURNAL OF COMPUTATIONAL LINGUISTICS》 * |
刘佳乐,等: ""基于近似支持向量机的流体识别方法"", 《辽宁化工》 * |
张文博,等: ""一种自适应权值的多特征融合分类方法"", 《系统工程与电子技术》 * |
殷天石,等: ""基于树型结构的SVM多类组合分类器在文本分类中的应用"", 《情报技术》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109856167A (en) * | 2018-12-05 | 2019-06-07 | 惠州市骏亚数字技术有限公司 | A kind of non-destructive testing device and its detection method |
CN109856167B (en) * | 2018-12-05 | 2021-06-18 | 惠州市骏亚数字技术有限公司 | Nondestructive testing device and testing method thereof |
CN109829883A (en) * | 2018-12-19 | 2019-05-31 | 歌尔股份有限公司 | Product quality detection method and device |
CN109829883B (en) * | 2018-12-19 | 2020-11-17 | 歌尔光学科技有限公司 | Product quality detection method and device |
WO2020242746A1 (en) * | 2019-05-31 | 2020-12-03 | Advanced Micro Devices, Inc. | Temperature-based adjustments for in-memory matrix multiplication |
US11507641B2 (en) | 2019-05-31 | 2022-11-22 | Advanced Micro Devices, Inc. | Temperature-based adjustments for in-memory matrix multiplication |
CN112345556A (en) * | 2020-11-23 | 2021-02-09 | 兰州大学 | Fault diagnosis system and method for integrated circuit |
CN112345556B (en) * | 2020-11-23 | 2023-07-21 | 兰州大学 | Fault diagnosis system and method for integrated circuit |
CN112836274A (en) * | 2021-02-02 | 2021-05-25 | 南京审计大学 | Data fusion method for tracing and auditing hidden engineering |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108830840A (en) | A kind of active intelligent detecting method of circuit board defect and its application | |
CN109977813B (en) | Inspection robot target positioning method based on deep learning framework | |
CN112766103B (en) | Machine room inspection method and device | |
CN102253048B (en) | Machine vision detection method and system for detection of various products | |
CN109615611A (en) | A kind of insulator self-destruction defect inspection method based on inspection image | |
CN110142785A (en) | A kind of crusing robot visual servo method based on target detection | |
CN106340009B (en) | A kind of electric power line detecting method and system based on parallel binocular | |
CN116879308A (en) | Industrial machine vision system image processing method | |
CN109544548A (en) | Defect inspection method, device, server, equipment and the storage medium of cutlery box | |
CN103454285A (en) | Transmission chain quality detection system based on machine vision | |
CN109307675A (en) | A kind of product appearance detection method and system | |
CN109840900A (en) | A kind of line detection system for failure and detection method applied to intelligence manufacture workshop | |
CN106918306A (en) | Industrial products three-dimensional appearance real-time detecting system based on light field one camera | |
CN113030102A (en) | Paint surface flaw inspection system based on machine vision | |
CN112304960B (en) | High-resolution image object surface defect detection method based on deep learning | |
CN107238534A (en) | The method and device of on-line monitoring plate stretch performance is calculated based on image | |
CN110097531A (en) | A kind of isomery image co-registration detection method for unmanned plane electric inspection process | |
CN109685038A (en) | A kind of article clean level monitoring method and its device | |
CN113077416A (en) | Welding spot welding defect detection method and system based on image processing | |
CN113822882A (en) | Circuit board surface defect detection method and device based on deep learning | |
CN109557098A (en) | Metal surface detection system based on machine vision | |
CN113706496B (en) | Aircraft structure crack detection method based on deep learning model | |
CN109886936B (en) | Low-contrast defect detection method and device | |
TW202127014A (en) | Intelligent Production Line Monitoring System and Implementation Method Thereof | |
CN108180871B (en) | method for quantitatively evaluating composite insulator surface pulverization roughness |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181116 |