CN108428247A - The detection method and system in bump direction - Google Patents
The detection method and system in bump direction Download PDFInfo
- Publication number
- CN108428247A CN108428247A CN201810164009.4A CN201810164009A CN108428247A CN 108428247 A CN108428247 A CN 108428247A CN 201810164009 A CN201810164009 A CN 201810164009A CN 108428247 A CN108428247 A CN 108428247A
- Authority
- CN
- China
- Prior art keywords
- bump
- subgraph
- scolding tin
- measured
- image
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
- G06T7/75—Determining position or orientation of objects or cameras using feature-based methods involving models
-
- 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
- G06T2207/30152—Solder
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The present invention relates to a kind of detection methods and system in bump direction.The above method includes step:The scolding tin image for obtaining bump, body region and perimeter are divided into according to position of the bump in the scolding tin image by the scolding tin image;Obtain the subgraph to be measured of the scolding tin image at least one direction;Wherein, the subgraph to be measured includes the topography of the body region on a direction and the topography of the perimeter in same direction;It is directional and direction-free probability to obtain the subgraph to be measured using bump angle detecting model trained in advance;If the directive probability of subgraph to be measured is more than direction-free probability, judge the direction of the bump on the corresponding direction of the subgraph to be measured.By the detection method and system in above-mentioned bump direction, the accuracy and precision of bump angle detecting are improved.
Description
Technical field
The present invention relates to automatic optics inspection technical fields, a kind of detection method more particularly to bump direction and are
System.
Background technology
Automatic optics inspection (Automatic Optic Inspection, AOI) is the necessary links of industrial manufacturing process,
The surface state that finished product is obtained using optical mode detects foreign matter or surface blemish with image processing.Scolding tin defects detection is
A kind of common application in automatic optics inspection field, machine obtain image by camera automatic scanning circuit plate, carry automatically
The topography of each solder joint is taken, and by image processing techniques, the bump direction at solder joint is judged, according to bump direction
It judges whether defect, finally the scolding tin of those suspected defects is shown or is marked, conveniently checks and overhauls.
More mature AOI equipment is the defect for detecting patch mostly in the market, also for plug-in unit class tin-soldering device
The technology for comparing weak, traditional detection plug-in unit class bump direction is based on normal image treatment technology, identifies having for each region
Scolding tin accounting is imitated to judge bump direction, accuracy is relatively low, easy tos produce the erroneous judgement in bump direction.
Invention content
Based on this, it is necessary to it is relatively low for detection accuracy, easy to produce the problem of erroneous judgement, provide that a kind of precision is high, sentences
The detection method and system in disconnected accurate bump direction.
A kind of detection method in bump direction, includes the following steps:
The scolding tin image for obtaining bump, according to position of the bump in the scolding tin image by the scolding tin figure
As being divided into body region and perimeter;
Obtain the subgraph to be measured of the scolding tin image at least one direction;Wherein, the subgraph to be measured includes on a direction
The body region topography and the perimeter in same direction topography;
It is directional and non-directional to obtain the subgraph to be measured using bump angle detecting model trained in advance
Probability;
If the directive probability of subgraph to be measured is more than direction-free probability, the direction of the bump is judged
On the corresponding direction of the subgraph to be measured.
In one embodiment, the step of subgraph to be measured for obtaining the scolding tin image at least one direction includes:
Set up detection zone at least one side of scolding tin image, obtain on the detection zone corresponding direction
The subgraph to be measured of scolding tin image;Wherein, the detection zone include the body region on a direction topography and
The topography of the perimeter in same direction.
In one embodiment, described to include the step of at least one side of scolding tin image sets up detection zone:
Choose reference direction of the direction as the scolding tin image;
According to the reference direction, with bump inspection is respectively set on the direction of reference direction interval predetermined angle
Survey region.
In one embodiment, described that bump is being respectively set on the direction of reference direction interval predetermined angle
The step of detection zone includes:
On the direction with 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 ° of the reference direction interval point
It She Zhi not 8 bump detection zones.
In one embodiment, described to be using the bump angle detecting model acquisition subgraph to be measured trained in advance
Further include step before the step of directional and direction-free probability:
Based on depth convolutional network theory and according to preset initial parameter, unbred bump angle detecting is established
Model, and classification based training is carried out to bump angle detecting model using the training subgraph of multiple history scolding tin images, it is instructed
Bump angle detecting model after white silk.
In one embodiment, the described the step of scolding tin image is divided into body region and perimeter, includes:
It determines bump position in scolding tin image, obtains the first area centered on bump, obtain covering and be more than
The second area of first area;
Using first area as the body region of scolding tin image, region after subtracting first area using second area is as weldering
The perimeter of tin image.
In one embodiment, step of the direction of the judgement bump on the corresponding direction of the subgraph to be measured
Further include step after rapid:
If the direction of the bump and the direction of the preset bump of scolding tin standard edition image mismatch, the weldering is judged
There are even tin defects in tin point.
In one embodiment, it is to have obtaining the subgraph to be measured using bump angle detecting model trained in advance
Further include step before the step of directionality and direction-free probability:
According to the direction of the preset bump of scolding tin standard edition image, filters out and examined from the subgraph to be measured
Subgraph to be measured on the direction of survey.
A kind of detecting system in bump direction, including:
Region division module, the scolding tin image for obtaining bump, according to the bump in the scolding tin image
Position the scolding tin image is divided into body region and perimeter;
Subgraph acquisition module, the subgraph to be measured for obtaining the scolding tin image at least one direction;Wherein, described to be measured
Subgraph includes the Local map of the topography of the body region on a direction and the perimeter in same direction
Picture;
Probability evaluation entity is to have for obtaining the subgraph to be measured using bump angle detecting model trained in advance
Directionality and direction-free probability;
Walking direction module, the direction for judging the bump, if the directive probability of subgraph to be measured is big
In direction-free probability, then judge the direction of the bump on the corresponding direction of the subgraph to be measured.
The detection method and system in above-mentioned bump direction, by defeated into bump angle detecting model trained in advance
The subgraph to be measured for entering the scolding tin image on different directions, according to the directional and direction-free probability of subgraph to be measured in all directions
Output, judges the direction of the bump, improves the accuracy and precision of bump angle detecting.
A kind of computer equipment, including memory, processor and be stored on the memory and can be in the processing
The computer program run on device, the processor realize the detection such as above-mentioned bump direction when executing the computer program
Method.
Above computer equipment is realized and is waited for according in all directions by the computer program run on the processor
The directional and direction-free probability output of subgraph is surveyed, the direction of the bump is judged, improves bump angle detecting
Accuracy and precision.
A kind of computer storage media, is stored thereon with computer program, is realized when which is executed by processor as above
State the detection method in bump direction.
Above computer storage medium is realized by the computer program of its storage according to subgraph to be measured in all directions
Directional and direction-free probability output, judges the direction of the bump, improves the accurate of bump angle detecting
Property and precision.
Description of the drawings
Fig. 1 is the flow chart of the detection method in the bump direction of one embodiment;
Fig. 2 is the schematic diagram of the division that scolding tin image is carried out to body region and perimeter of one embodiment;
Fig. 3 is that the bump in one direction of one embodiment has directive schematic diagram;
Fig. 4 is the schematic diagram that the bump detection zone of one embodiment is arranged;
Fig. 5 is the structural schematic diagram of the detecting system in the bump direction of one embodiment.
Specific implementation mode
To facilitate the understanding of the present invention, below with reference to relevant drawings to invention is more fully described.Reference chart
1, Fig. 1 shows the flow chart of the detection method in the bump direction of one embodiment, mainly includes the following steps:
Step S10:The scolding tin image for obtaining bump, according to position of the bump in the scolding tin image by institute
It states scolding tin image and is divided into body region and perimeter.
In this step, above-mentioned scolding tin image is the real-time of the bump to be detected obtained by the camera of detection device
Above-mentioned scolding tin image is carried out body region and perimeter according to the position of bump in above-mentioned scolding tin image by image
It divides;Wherein, in body region be region where normal scolding tin, and if there are scolding tin in perimeter, and at some
It is connected with the scolding tin of body region on direction, then shows that the bump is with directionality.
With reference to figure 2, Fig. 2 shows the divisions that scolding tin image is carried out to body region and perimeter of one embodiment
Schematic diagram.The step of scolding tin image is divided into body region 200 and perimeter 300 include:Determine scolding tin image
100 position of middle bump obtains the first area centered on bump 100, obtains covering and more than the second of first area
Region;Using first area as the body region 200 of scolding tin image, region after subtracting first area using second area is as weldering
The perimeter 300 of tin image.
Specifically, it is determined that position of the bump 100 in scolding tin image, and according to the parameter setting of scolding tin standard format,
Centered on bump 100 obtain meet parameter preset setting size first area, and by the first area determine based on
Body region is extended to the outside the secondth area for obtaining and covering the body region 200 and meeting parameter preset size by body region 200
The second area is removed the epitaxial part obtained after the body region and is set as perimeter 300 by domain.
With reference to figure 3, Fig. 3 shows that the bump in one direction of one embodiment has directive schematic diagram.
As shown in figure 3, most of scolding tin 101 of bump is fallen in body region 200, on 90 ° of directions, there is scolding tin
102 are present in perimeter 300, and in this direction, the scolding tin of scolding tin 101 and perimeter in body region 200
102 are interconnected, then judge the direction of the bump 100 on 90 ° of direction.
Step S20:Obtain the subgraph to be measured of the scolding tin image at least one direction;Wherein, the subgraph to be measured includes
The topography of the body region on one direction and the topography of the perimeter in same direction.
In this step, with the topography of body region in a direction, and the Local map with the body region
The topography of the perimeter of picture in the same direction obtains the son to be measured at least one direction as a subgraph to be measured
Figure.
In one embodiment, the step of subgraph to be measured for obtaining the scolding tin image at least one direction includes:
Detection zone is set up at least one side of scolding tin image, is obtained and the scolding tin image on the detection zone corresponding direction
Subgraph to be measured;Wherein, the detection zone includes topography and the same direction of the body region on a direction
On the perimeter topography.
Specifically, on entire scolding tin image, it can at least select a direction and detection zone is set in this direction,
The size of detection zone is unlimited, can be needed according to detection and required precision is configured;Above-mentioned detection zone includes the party
The topography of upward body region and the topography of perimeter in this direction, according to the setting of detection zone and
It divides, image corresponding with each detection zone is extracted from entire scolding tin image, as subgraph to be measured in this direction.
In one embodiment, include the step of at least one side of scolding tin image sets up detection zone:It chooses
Reference direction of one direction as the scolding tin image;According to the reference direction, default with the reference direction interval
Bump detection zone is respectively set on the direction of angle.
Specifically, because different directions have independence, it is possible to carry out independent directionality for different directions and sentence
It is disconnected, you can multiple detection zones are arranged;A reference direction can be selected, at the angle with the reference direction interval certain predetermined
Multiple bump detection zones are respectively set on the direction of degree;The present embodiment by the way that the angle at reference direction and interval is arranged,
The efficiency of detection can be improved to avoid the overlapping of detection zone.
In one embodiment, be respectively set on the direction of the reference direction interval predetermined angle bump detection
The step of region includes:In the side with 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 ° of the reference direction interval
8 bump detection zones are respectively set upwards.
With reference to figure 4, Fig. 4 shows the schematic diagram of the bump detection zone setting of one embodiment.
Specifically, reference direction is set as 0 ° of direction, with it counterclockwise or clockwise at interval of 45 ° of settings, one detection zone
Domain, until entire scolding tin image is covered comprehensively;Such as Fig. 4, be spaced counterclockwise with reference direction 45 °, 90 °, 135 °, 180 °,
225 °, 270 ° and 315 ° of side sets up detection zone, respectively obtains 0 ° of detection zone, 45 ° of detection zones, 90 ° of detection zones,
135 degree of detection zones, 180 ° of detection zones, 225 degree of detection zones, 270 ° of detection zones and 315 detection zones, the inspection in above-mentioned all directions
It surveys region to cover entire scolding tin image comprehensively, and each detection zone includes the topography of body region 200 and same
The topography of perimeter 300 on one direction.
The present embodiment sets up inspection by 8 sides at 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 °
Region is surveyed, entire scolding tin image is covered comprehensively, the directionality of bump can be judged on each independent direction,
The complete detection to entire scolding tin image is realized, the accuracy and precision of detection are improved.
It in one embodiment, can also be according to the direction of the preset bump of scolding tin standard edition image, from described to be measured
The subgraph to be measured on the direction for needing to be detected is filtered out in subgraph.
Specifically, can be the preset direction of the bump according to scolding tin standard edition image, on the preset direction
The subgraph to be measured of the bump is obtained, and then the subgraph to be measured on the preset direction is detected, judges the bump
Whether direction is on the preset direction.By screening subgraph to be measured, it may be implemented targetedly to detect, reducing work
Detection efficiency is improved while amount.
Step S30:Using bump angle detecting model trained in advance obtain the subgraph to be measured be it is directional and
Direction-free probability.
In this step, the subgraph to be measured of the scolding tin image of acquisition is input to bump angle detecting mould trained in advance
In type, trained bump angle detecting model can carry out classification judgement to the subgraph to be measured in advance, can export this respectively
Subgraph to be measured is the probability value for having direction subgraph and the probability value that the subgraph to be measured is directionless subgraph.
In one embodiment, it is the side of having to obtain the subgraph to be measured using bump angle detecting model trained in advance
Further include step before the step of tropism and direction-free probability:Based on depth convolutional network theory and according to preset initial
Parameter establishes unbred bump angle detecting model, and using the training subgraph of multiple history scolding tin images to scolding tin
Point angle detecting model carries out classification based training, the bump angle detecting model after being trained.
Specifically, it is based on depth convolution theory and according to preset initial parameter, establishes unbred bump direction
Detection model, the bump angle detecting model can need that the correlations such as the convolution number of plies, convolution algorithm are arranged according to accuracy
Model parameter;After model foundation is good, using the training subgraph of a large amount of history scolding tin images gathered in advance to above-mentioned bump
Angle detecting model carries out classification based training, and after the training of training subgraph, bump angle detecting mould can export two points
The probability of class, respectively directive probability and direction-free probability.
In one embodiment, bump angle detecting model may include 3 convolutional layers and pond layer and full connection
Layer, the output result of each convolutional layer is non-linear using ReLU (Rectified Linear Units, activation primitive) progress
After transformation, enters pond layer and carry out pond and by full articulamentum, finally by Softmax function output category probability.
In one embodiment, the subgraph that according to preset standard for manual sampling, can acquire a large amount of history scolding tin image is made
For training subgraph, i.e. training sample;Wherein, training subgraph includes directive subgraph and nondirectional subgraph, nondirectional
Training subgraph may also be referred to as positive sample, and directive trained subgraph may also be referred to as negative sample;The step of acquiring positive and negative samples
For:Multiple history scolding tin images are obtained, the position where bump in history scolding tin image is divided into body region and outer
Portion region;Obtain the training subgraph of history scolding tin image respectively in a plurality of directions, training subgraph includes the master on a direction
The topography of body region and the topography of the perimeter in same direction.Obtaining sufficient amount of trained subgraph
Afterwards, classification based training, the weldering after being trained are carried out to bump angle detecting model using the training subgraph of history scolding tin image
Tin point angle detecting model.
Step S40:If the directive probability of subgraph to be measured is more than direction-free probability, the scolding tin is judged
The direction of point is on the corresponding direction of the subgraph to be measured.
In this step, according to the class probability of bump angle detecting model output trained in advance, judge that this is to be measured
Subgraph belong to directive probability whether be more than belong to direction-free probability;If so, judging that the subgraph to be measured is corresponding
Direction is the direction of the bump.
In one embodiment, the step in the direction for judging the bump on the corresponding direction of the subgraph to be measured
Later, further include step:If the direction of the bump and the direction of the preset bump of scolding tin standard edition image mismatch, sentence
There are even tin defects in the fixed bump.
Above-described embodiment, judgement bump actual direction on the corresponding direction of a certain subgraph to be measured after, by this
The preset direction that actual direction should have with the preset bump of scolding tin standard edition image is compared, if actual direction with
Preset direction is inconsistent, then judges that the bump has even tin defects.By the actual direction and scolding tin mark that compare bump
Quasi- domain as preset bump direction, can the company's of further realizing tin defects detection, improve the precision of defects detection.
It is possible to further detect the subgraph to be measured of the bump in 8 directions respectively, judge the bump at which
There are even tin defects on a or which direction, the complete detection of scolding tin defect may be implemented, and then improve the accuracy of detection.
The detection method in above-mentioned bump direction is different by being inputted into bump angle detecting model trained in advance
The subgraph to be measured of scolding tin image on direction, according to the directional and direction-free probability output of subgraph to be measured in all directions,
The direction for judging the bump improves the accuracy and precision of bump angle detecting.
The specific implementation mode of the detecting system in the bump direction of the present invention is described in detail below in conjunction with the accompanying drawings, is joined
Fig. 5 is examined, Fig. 5 shows the structural schematic diagram of the detecting system in the bump direction of one embodiment.
A kind of detecting system in bump direction, including:Region division module 10, subgraph acquisition module 20, probability calculation
Module 30 and walking direction module 40.
Region division module 10, the scolding tin image for obtaining bump, according to the bump in the scolding tin image
In position the scolding tin image is divided into body region and perimeter;
Subgraph acquisition module 20, the subgraph to be measured for obtaining the scolding tin image at least one direction;Wherein, described to wait for
It includes the topography of the body region on a direction and the part of the perimeter in same direction to survey subgraph
Image;
Probability evaluation entity 30, for being using the bump angle detecting model acquisition subgraph to be measured trained in advance
Directional and direction-free probability;
Walking direction module 40, the direction for judging the bump, if the directive probability of subgraph to be measured
More than direction-free probability, then judge the direction of the bump on the corresponding direction of the subgraph to be measured.
In one embodiment, for region division module 10, can be further used for determining bump in scolding tin image
Position obtains the first area centered on bump, obtains covering and the second area more than first area;With first area
As the body region of scolding tin image, region after subtracting first area using second area is as the perimeter of scolding tin image.
In one embodiment, for subgraph acquisition module 20, can be further used in at least one of scolding tin image
Side sets up detection zone, obtains the subgraph to be measured with the scolding tin image on the detection zone corresponding direction;Wherein, described
Detection zone includes the topography of the body region on a direction and the office of the perimeter in same direction
Portion's image.
In one embodiment, for subgraph acquisition module 20, can be further used for choosing described in a direction conduct
The reference direction of scolding tin image;According to the reference direction, distinguish on the direction of reference direction interval predetermined angle
Bump detection zone is set.The present embodiment, can be to avoid detection zone by setting reference direction and the angle at interval
Overlapping, improves the efficiency of detection.
In one embodiment, for subgraph acquisition module 20, can be further used for the reference direction interval
8 bump detection zones are respectively set on 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 ° of direction.This reality
It applies example and sets up detection zone by 8 sides at 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 °, it will be whole
A scolding tin image covers comprehensively, can judge the directionality of bump on each independent direction, realize to whole
The complete detection of a scolding tin image improves the accuracy and precision of detection.
In one embodiment, it for subgraph acquisition module 20, can be also used for preset according to scolding tin standard edition image
The direction of bump, from the subgraph to be measured filtered out in the subgraph to be measured on the direction for needing to be detected.The present embodiment is logical
It crosses and screens subgraph to be measured, may be implemented targetedly to detect, detection efficiency is improved while reducing workload.
In one embodiment, for probability evaluation entity 30, can be further used for theoretical based on depth convolutional network
And according to preset initial parameter, unbred bump angle detecting model is established, and utilize multiple history scolding tin images
Training subgraph to bump angle detecting model carry out classification based training, the bump angle detecting model after being trained.
In one embodiment, it for walking direction module 40, can be also used for judging whether the bump has even tin
Defect judges the scolding tin if the direction of the bump and the direction of the preset bump of scolding tin standard edition image mismatch
There are even tin defects in point.The present embodiment, by the actual direction and the preset scolding tin of scolding tin standard edition image that compare bump
Point direction, can the company's of further realizing tin defects detection, improve the precision of defects detection.
The detecting system in above-mentioned bump direction is different by being inputted into bump angle detecting model trained in advance
The subgraph to be measured of scolding tin image on direction, according to the directional and direction-free probability output of subgraph to be measured in all directions,
The direction for judging the bump improves the accuracy and precision of bump angle detecting.
In one embodiment, a kind of computer equipment is also provided, which includes memory, processor and deposit
Store up the computer program that can be run on a memory and on a processor, wherein processor is realized as above when executing described program
State the detection method in any one bump direction in each embodiment.
The computer equipment, when processor executes program, by realizing such as any one weldering in the various embodiments described above
The detection method in tin point direction, so as to improve the accuracy and precision of bump angle detecting.
In addition, one of ordinary skill in the art will appreciate that realize above-described embodiment method in all or part of flow,
It is that relevant hardware can be instructed to complete by computer program, the program can be stored in a non-volatile calculating
In machine read/write memory medium, in the embodiment of the present invention, which can be stored in the storage medium of computer system, and by
At least one of computer system processor executes, and includes the reality such as the detection method in above-mentioned each bump direction with realization
Apply the flow of example.
In one embodiment, a kind of storage medium is also provided, computer program is stored thereon with, wherein the program quilt
The detection method such as any one bump direction in the various embodiments described above is realized when processor executes.Wherein, described to deposit
Storage media can be magnetic disc, CD, read-only memory (Read-Only Memory, ROM) or random access memory
(Random Access Memory, RAM) etc..
The computer storage media, the computer program of storage include such as above-mentioned each bump direction by realizing
The flow of the embodiment of detection method, so as to improve the accuracy and precision of bump angle detecting.
The preferred embodiment of the present invention is given in attached drawing.But the present invention can realize in many different forms,
It is not limited to the examples described herein.Make to the disclosure on the contrary, purpose of providing these embodiments is
It is more thorough and comprehensive.
Each technical characteristic of embodiment described above can be combined arbitrarily, to keep description succinct, not to above-mentioned reality
It applies all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, it is all considered to be the range of this specification record.Several implementations of the invention above described embodiment only expresses
Mode, the description thereof is more specific and detailed, but can not therefore be construed as limiting the scope of the patent.It should be understood that
It is that for those of ordinary skill in the art, without departing from the inventive concept of the premise, several deformations can also be made
And improvement, these are all within the scope of protection of the present invention.Therefore, the protection domain of patent of the present invention should be with appended claims
It is accurate.
Claims (10)
1. a kind of detection method in bump direction, which is characterized in that include the following steps:
The scolding tin image for obtaining bump draws the scolding tin image according to position of the bump in the scolding tin image
It is divided into body region and perimeter;
Obtain the subgraph to be measured of the scolding tin image at least one direction;Wherein, the subgraph to be measured includes the institute on a direction
State the topography of the perimeter in the topography and same direction of body region;
It is directional and direction-free general to obtain the subgraph to be measured using bump angle detecting model trained in advance
Rate;
If the directive probability of subgraph to be measured is more than direction-free probability, judge the direction of the bump in institute
It states on the corresponding direction of subgraph to be measured.
2. the detection method in bump direction according to claim 1, which is characterized in that described to obtain at least one direction
On scolding tin image subgraph to be measured the step of include:
Detection zone is set up at least one side of scolding tin image, is obtained and the scolding tin on the detection zone corresponding direction
The subgraph to be measured of image;Wherein, the detection zone includes the topography of the body region on a direction and same
The topography of the perimeter on direction.
3. the detection method in bump direction according to claim 2, which is characterized in that it is described scolding tin image at least
The step of one side sets up detection zone include:
Choose reference direction of the direction as the scolding tin image;
According to the reference direction, bump detection zone is being respectively set on the direction of reference direction interval predetermined angle
Domain.
4. the detection method in bump direction according to claim 3, which is characterized in that it is described with the reference direction
Being spaced the step of bump detection zone is respectively set on the direction of predetermined angle includes:
It is set respectively on the direction with 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 ° of the reference direction interval
Set 8 bump detection zones.
5. the detection method in bump direction according to claim 1, which is characterized in that described to utilize weldering trained in advance
Further include step before tin point angle detecting model obtains the step of subgraph to be measured is directional and direction-free probability
Suddenly:
Based on depth convolutional network theory and according to preset initial parameter, unbred bump angle detecting mould is established
Type, and classification based training is carried out to bump angle detecting model using the training subgraph of multiple history scolding tin images, it is trained
Bump angle detecting model afterwards.
6. the detection method in bump direction according to claim 1, which is characterized in that described to draw the scolding tin image
The step of being divided into body region and perimeter include:
It determines bump position in scolding tin image, obtains the first area centered on bump, obtain covering and be more than first
The second area in region;
Using first area as the body region of scolding tin image, region after subtracting first area using second area is as scolding tin figure
The perimeter of picture.
7. the detection method in bump direction according to claim 1, which is characterized in that the judgement bump
Further include step after step of the direction on the corresponding direction of the subgraph to be measured:
If the direction of the bump and the direction of the preset bump of scolding tin standard edition image mismatch, the bump is judged
In the presence of even tin defects.
8. the detection method in bump direction according to claim 1, which is characterized in that utilizing scolding tin trained in advance
Point angle detecting model further includes step before obtaining the step of subgraph to be measured is directional and direction-free probability:
According to the direction of the preset bump of scolding tin standard edition image, filter out what needs were detected from the subgraph to be measured
Subgraph to be measured on direction.
9. a kind of detecting system in bump direction, which is characterized in that including:
Region division module, the scolding tin image for obtaining bump, according to position of the bump in the scolding tin image
It sets and the scolding tin image is divided into body region and perimeter;
Subgraph acquisition module, the subgraph to be measured for obtaining the scolding tin image at least one direction;Wherein, the subgraph to be measured
The topography of the perimeter in topography and same direction including the body region on a direction;
Probability evaluation entity is to have direction for obtaining the subgraph to be measured using bump angle detecting model trained in advance
Property and direction-free probability;
Walking direction module, the direction for judging the bump, if the directive probability of subgraph to be measured is more than nothing
The probability of directionality then judges the direction of the bump on the corresponding direction of the subgraph to be measured.
10. a kind of computer equipment, including memory, processor and it is stored on the memory and can be in the processor
The computer program of upper operation, which is characterized in that the processor realized when executing the computer program as claim 1 to
The detection method in the bump direction described in 8 any one.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810164009.4A CN108428247B (en) | 2018-02-27 | 2018-02-27 | Method and system for detecting direction of soldering tin point |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810164009.4A CN108428247B (en) | 2018-02-27 | 2018-02-27 | Method and system for detecting direction of soldering tin point |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108428247A true CN108428247A (en) | 2018-08-21 |
CN108428247B CN108428247B (en) | 2020-06-12 |
Family
ID=63157336
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810164009.4A Active CN108428247B (en) | 2018-02-27 | 2018-02-27 | Method and system for detecting direction of soldering tin point |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108428247B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109146873A (en) * | 2018-09-04 | 2019-01-04 | 凌云光技术集团有限责任公司 | A kind of display screen defect intelligent detecting method and device based on study |
TWI745767B (en) * | 2019-10-18 | 2021-11-11 | 汎思數據股份有限公司 | Optical inspection secondary image classification method |
US20220023977A1 (en) * | 2020-07-23 | 2022-01-27 | Pegatron Corporation | Solder joint inspection model training method, solder joint inspection method, and solder joint inspection device |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104764752A (en) * | 2015-03-24 | 2015-07-08 | 广州视源电子科技股份有限公司 | A tin adhesion detecting method and a tin adhesion detecting device for a welding spot |
CN105588840A (en) * | 2015-12-04 | 2016-05-18 | 广州视源电子科技股份有限公司 | Electronic element positioning method and device |
CN105821538A (en) * | 2016-04-20 | 2016-08-03 | 广州视源电子科技股份有限公司 | Detection method and system for breaking of spun yarns |
CN105931226A (en) * | 2016-04-14 | 2016-09-07 | 南京信息工程大学 | Automatic cell detection and segmentation method based on deep learning and using adaptive ellipse fitting |
CN106290388A (en) * | 2016-08-03 | 2017-01-04 | 国网山东省电力公司电力科学研究院 | A kind of insulator breakdown automatic testing method |
CN106372630A (en) * | 2016-11-23 | 2017-02-01 | 华南理工大学 | Face direction detection method based on deep learning |
CN107451997A (en) * | 2017-07-31 | 2017-12-08 | 南昌航空大学 | A kind of automatic identifying method of the welding line ultrasonic TOFD D scanning defect types based on deep learning |
-
2018
- 2018-02-27 CN CN201810164009.4A patent/CN108428247B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104764752A (en) * | 2015-03-24 | 2015-07-08 | 广州视源电子科技股份有限公司 | A tin adhesion detecting method and a tin adhesion detecting device for a welding spot |
CN105588840A (en) * | 2015-12-04 | 2016-05-18 | 广州视源电子科技股份有限公司 | Electronic element positioning method and device |
CN105931226A (en) * | 2016-04-14 | 2016-09-07 | 南京信息工程大学 | Automatic cell detection and segmentation method based on deep learning and using adaptive ellipse fitting |
CN105821538A (en) * | 2016-04-20 | 2016-08-03 | 广州视源电子科技股份有限公司 | Detection method and system for breaking of spun yarns |
CN106290388A (en) * | 2016-08-03 | 2017-01-04 | 国网山东省电力公司电力科学研究院 | A kind of insulator breakdown automatic testing method |
CN106372630A (en) * | 2016-11-23 | 2017-02-01 | 华南理工大学 | Face direction detection method based on deep learning |
CN107451997A (en) * | 2017-07-31 | 2017-12-08 | 南昌航空大学 | A kind of automatic identifying method of the welding line ultrasonic TOFD D scanning defect types based on deep learning |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109146873A (en) * | 2018-09-04 | 2019-01-04 | 凌云光技术集团有限责任公司 | A kind of display screen defect intelligent detecting method and device based on study |
CN109146873B (en) * | 2018-09-04 | 2020-12-29 | 凌云光技术股份有限公司 | Learning-based intelligent detection method and device for defects of display screen |
TWI745767B (en) * | 2019-10-18 | 2021-11-11 | 汎思數據股份有限公司 | Optical inspection secondary image classification method |
US20220023977A1 (en) * | 2020-07-23 | 2022-01-27 | Pegatron Corporation | Solder joint inspection model training method, solder joint inspection method, and solder joint inspection device |
US11833618B2 (en) * | 2020-07-23 | 2023-12-05 | Pegatron Corporation | Solder joint inspection model training method, solder joint inspection method, and solder joint inspection device |
Also Published As
Publication number | Publication date |
---|---|
CN108428247B (en) | 2020-06-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11915407B2 (en) | Automated system and method for clarity measurements and clarity grading | |
WO2022042579A1 (en) | Lcd screen defect detection method and apparatus | |
CN106846344B (en) | A kind of image segmentation optimal identification method based on the complete degree in edge | |
CN105973148B (en) | Written apparent size standardized detection system based on machine vision | |
WO2017067023A1 (en) | Method for detecting body fluid based on special test paper | |
CN105453153B (en) | Traffic lights detects | |
CN108875911A (en) | One kind is parked position detecting method | |
CN110044405B (en) | Automatic automobile instrument detection device and method based on machine vision | |
CN109145830A (en) | A kind of intelligence water gauge recognition methods | |
CN108428247A (en) | The detection method and system in bump direction | |
CN107578021A (en) | Pedestrian detection method, apparatus and system based on deep learning network | |
CN102074031B (en) | Standard establishment method for observational check machine of printed circuit board | |
CN104463240B (en) | A kind of instrument localization method and device | |
CN104053984A (en) | Image examination method and image examination apparatus | |
CN109886960A (en) | The method of glass edge defects detection based on machine vision | |
CN111652883B (en) | Glass surface defect detection method based on deep learning | |
CN113222913B (en) | Circuit board defect detection positioning method, device and storage medium | |
CN105243356A (en) | Method of building pedestrian detection model and device and pedestrian detection method | |
CN108320799B (en) | Image analysis and recognition method for lateral flow paper strip disease diagnosis | |
CN111275705A (en) | Intelligent cloth inspecting method and device, electronic equipment and storage medium | |
CN108154513A (en) | Cell based on two photon imaging data detects automatically and dividing method | |
CN110174063A (en) | A kind of neutral pen ink height detecting system and detection method based on machine vision | |
CN115205288B (en) | Industrial defect detection method and device | |
CN103955929B (en) | Image local edge pattern and non-edge mode judging method and judgment means | |
CN115452844A (en) | Injection molding part detection method and system based on machine vision |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |