CN108230327A - A kind of packaging location based on MVP platforms and sort research universal method - Google Patents
A kind of packaging location based on MVP platforms and sort research universal method Download PDFInfo
- Publication number
- CN108230327A CN108230327A CN201611155676.3A CN201611155676A CN108230327A CN 108230327 A CN108230327 A CN 108230327A CN 201611155676 A CN201611155676 A CN 201611155676A CN 108230327 A CN108230327 A CN 108230327A
- Authority
- CN
- China
- Prior art keywords
- packaging
- image
- mvp
- platforms
- universal method
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
Abstract
The invention discloses a kind of packaging location based on MVP platforms of vision sorter field of locating technology and sort research universal methods, are somebody's turn to do the packaging location based on MVP platforms and are as follows with sort research universal method：S1：Image is pre-processed with global threshold segmentation and connectivity analysis；S2：Obtain the exact position and direction at packaging edge, drawing template establishment；S3：To being packaged in the accurate acquisition of position on image；S4：Different Package is trained, classification model construction；S5：Extract packaging area；The present invention carries out packaging location and sort research by NI Vision Builder for Automated Inspection, realizes the accurate acquisition to Packaging image position, lays the foundation for manipulator crawl, while Package Classification mistake caused by can avoiding human factor, realizes to packing Accurate classification.
Description
Technical field
The present invention relates to vision sorter field of locating technology, specially a kind of packaging location and classification based on MVP platforms
Study universal method.
Background technology
Demand of the workforce capital to " machine substitute human labor " is increasingly promoted.The aging of China human mortality structure, promotes
Labor cost is further raised, and China's manufacturing industry automatization level is relatively low in addition, and automation equipment is more old, leads to China
Workforce capital has larger equipment purchase and the demand to update, so China's workforce capital is to machine
Device vision system potential of demand is huge.
Intelligent vision system has many advantages, such as that accuracy is high, speed is fast, precision is high relative to conventional mechanical system.It is modern
The rapid development of chemical industry, the arrival of attenuation and the third time industrial revolution of demographic dividend so that science and technology rapid development, intelligence
The combination of energy vision system and industrial robot so that complicated automation control is possibly realized, and intelligent vision system is major
Field extensive use.
MVP can standardize machine vision software, and each function module can reach the durability of height, and can be certainly
By flexibly assembling, product development does not need to the peopleware of profession, it is only necessary to which product group member understands carries out assembly i.e. by handbook
Can, human resources are greatly saved, shorten the research and development of products time, while also ensure that product has the stability of height.
Previous packaging order completion be it is artificial read order, manual picking, packings, it is time-consuming, effort, error-prone, compare intelligence
The way of energy is that the packaging of fixed position is captured by control machinery hand teaching, order stacking is completed, when kind of packing number
When more, need to occupy larger space, be unfavorable for forming unified production line.For this purpose, we have proposed one kind to be based on MVP platforms
Packaging location come into operation with sort research universal method, to solve the above problems.
Invention content
The purpose of the present invention is to provide a kind of packaging locations based on MVP platforms and sort research universal method, have solved
Previous packaging order completion certainly mentioned above in the background art is manually to read order, manual picking, packing, time-consuming, effort,
Error-prone, the way of intelligent is that the packaging of fixed position is captured by control machinery hand teaching, completes order stacking, when
When kind of packing number is more, needs to occupy larger space, be unfavorable for forming unified production line problem.
To achieve the above object, the present invention provides following technical solution：A kind of packaging location and classification based on MVP platforms
Universal method is studied, the packaging location based on MVP platforms is somebody's turn to do and is as follows with sort research universal method：
S1：Image is pre-processed with global threshold segmentation and connectivity analysis, packaging area is divided from image
From；
S2：Step S1 is packed and carries out edge detection and tracking, obtains the exact position and direction at packaging edge, creates mould
Plate；
S3：Template matches based on image border find template using Pyramidal search strategy, reduce search in the picture
Range reaches the accurate acquisition to being packaged in position on image；
S4：Package Classification method based on support vector machines, using gray level co-occurrence matrixes, by calculate energy, correlation,
7 contrast, entropy, mean value, variance and anisotropy texture characteristic amounts, are trained Different Package, classification model construction；
S5：Packaging area is extracted, using the grader in step S4, realizes the classification to packaging.
Preferably, high clear colorful image acquires acquisition by CCD high definitions face battle array color camera by light source in the step S1.
Preferably, in the step S3, when carrying out template matches, all marginal points of template and the image side nearest from it
Mean square distance between edge point is minimum, and minimum distance is found out using range conversion.
Preferably, the classification model construction image in the step S4 uses statistical average method for multiple packaging high clear colorful images
Synthesis.
Preferably, mould grader is created in the step S4 to include the following steps：
S41：Using gray level co-occurrence matrixes, from 7 mean value, variance, energy, correlation, contrast, entropy and anisotropy spies
Sign carrys out training sample, and feature space dimension is set as 7, classifies to 15 classes packaging, sample classification number is set as 15；
S42：Addition, analyzing and packaging sample, extract 7 different characteristics of sample, and training sample forms inhomogeneity.
Compared with prior art, the beneficial effects of the invention are as follows：It is more accurate that the present invention positions Package Classification, is conducive to
Unified production line is formed, saves human and material resources, improves production efficiency, the present invention is packed by NI Vision Builder for Automated Inspection
Positioning and sort research, realize the accurate acquisition to Packaging image position, lay the foundation, while can avoid people for manipulator crawl
For Package Classification mistake caused by factor, realize to packing Accurate classification.
Description of the drawings
Fig. 1 is work flow diagram of the present invention；
Fig. 2 is image pyramid model structure in the embodiment of the present invention one；
Fig. 3 is support vector machines mapping structure figure in the embodiment of the present invention two.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other without making creative work
Embodiment shall fall within the protection scope of the present invention.
Embodiment one
When calculating testing image edge with template image similarity, treated using all marginal points of template and nearest from it
Mean square distance minimum between altimetric image marginal point is as similarity.Range conversion may be used during this and find out most low coverage
From can also reduce search range, finally calculate the local minimum of similarity, equal method, edge distance (SED) is：
If without using stopping criterion, the complexity of above-mentioned template matching algorithm is O (whn), and w is the width of image, and h is
The height of image, n are the number at template midpoint, even if using stopping criterion, can not change algorithm complexity, algorithm speed carries
The ratio of liter is also a constant.In machine vision applications, image processing time is reduced as far as possible, to improve algorithm speed,
W, h, n numerical value must be reduced, image pyramid, which is exactly based on, reduces the method that the wide height of image carrys out boosting algorithm operational efficiency.
Referring to Fig. 2, image pyramid is to carry out double sampling to image or template, image size is repeatedly reduced into
The half of last layer image forms pyramid diagram picture.With the increase of figure layer, wide height halves, and resolution ratio is gradually lowered, in order to
Sawtooth effect of image after sampling is eliminated, uses mean filter smoothed image.After template instances are searched in high level, it will match
As a result lower layer of pyramid is mapped to, i.e., coordinate is multiplied by 2, it is contemplated that matching position is there may be deviation, by matching result week
The region enclosed finally tracks the image pyramid bottom as region of search, due to be carried out in zonule it is similar
Calculating, Threshold segmentation, extraction local extremum are spent, so the speed of service is very fast.
Embodiment two
When sample size is enough, gauss hybrid models have preferable robustness for classification, but small for solving
When sample, non-linear and high dimensional pattern identification problem, gauss hybrid models but show unsatisfactory, and support vector machines is but very big
Advantage, the VC of Statistical Learning Theory dimension theory with Structural risk minization principle is combined, it is super flat to define optimum linearity first
Face is then based on Mercer core expansion theorems, by Nonlinear Mapping ψ, by the way that sample space is mapped to higher-dimension or even nothing
The feature space tieed up thoroughly, so that the method that linear learning machine can be applied in feature space solves the height in sample space
The problems such as tieing up Nonlinear Classification and returning.Since support vector machines applies expansion and computational theory based on core, so being not required to
The display expression formula of Nonlinear Mapping is solved, compared to linear model, using linear learning machine in higher dimensional space, can be kept away
Exempt from " dimension calamity number " and the complexity calculated.
Non-linear svm classifier function cans be compared to a neural network, and each middle layer node is input sample and a support
The inner product of vector, output is the linear combination of several middle layer nodes.
Referring to Fig. 3, gray level co-occurrence matrixes represent the Joint Distribution probability of pixel pair, it is a symmetrical matrix, not only reflects
Gradation of image phase adjacent spaces, adjacent direction, amplitude of variation integrated information, also reflect between identical gray-level pixels
Position distribution feature, be calculate textural characteristics basis.It is calculated after packaging gray level co-occurrence matrixes, often directly should not
With it, but calculate texture characteristic amount on this basis, this system using energy, correlation, contrast, entropy, mean value, variance,
The characteristic quantities such as anisotropy represent.By means of above 7 characteristics, classify to 15 classes packaging.
It although an embodiment of the present invention has been shown and described, for the ordinary skill in the art, can be with
Understanding without departing from the principles and spirit of the present invention can carry out these embodiments a variety of variations, modification, replace
And modification, the scope of the present invention is defined by the appended.
Claims (5)
1. a kind of packaging location based on MVP platforms and sort research universal method, it is characterised in that：This is based on MVP platforms
Packaging location is as follows with sort research universal method：
S1：Image is pre-processed with global threshold segmentation and connectivity analysis, packaging area is detached from image；
S2：Step S1 is packed and carries out edge detection and tracking, obtains the exact position and direction at packaging edge, drawing template establishment；
S3：Template matches based on image border find template in the picture using Pyramidal search strategy, reduce search model
It encloses, reaches the accurate acquisition to being packaged in position on image；
S4：Package Classification method based on support vector machines, using gray level co-occurrence matrixes, by calculating energy, correlation, comparison
7 degree, entropy, mean value, variance and anisotropy texture characteristic amounts, are trained Different Package, classification model construction；
S5：Packaging area is extracted, using the grader in step S4, realizes the classification to packaging.
2. a kind of packaging location based on MVP platforms according to claim 1 and sort research universal method, feature exist
In：High clear colorful image is acquired by light source by CCD high definitions face battle array color camera and obtained in the step S1.
3. a kind of packaging location based on MVP platforms according to claim 1 and sort research universal method, feature exist
In：It is equal between all marginal points of template and the image border nearest from it point when carrying out template matches in the step S3
Side's distance is minimum, and minimum distance is found out using range conversion.
4. a kind of packaging location based on MVP platforms according to claim 1 and sort research universal method, feature exist
In：Classification model construction image in the step S4 is synthesized for multiple packaging high clear colorful images using statistical average method.
5. a kind of packaging location based on MVP platforms according to claim 1 and sort research universal method, feature exist
In：Mould grader is created in the step S4 to include the following steps：
S41：Using gray level co-occurrence matrixes, come from 7 mean value, variance, energy, correlation, contrast, entropy and anisotropy features
Training sample, feature space dimension are set as 7, classify to 15 classes packaging, sample classification number is set as 15；
S42：Addition, analyzing and packaging sample, extract 7 different characteristics of sample, and training sample forms inhomogeneity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611155676.3A CN108230327A (en) | 2016-12-14 | 2016-12-14 | A kind of packaging location based on MVP platforms and sort research universal method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611155676.3A CN108230327A (en) | 2016-12-14 | 2016-12-14 | A kind of packaging location based on MVP platforms and sort research universal method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108230327A true CN108230327A (en) | 2018-06-29 |
Family
ID=62650133
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611155676.3A Pending CN108230327A (en) | 2016-12-14 | 2016-12-14 | A kind of packaging location based on MVP platforms and sort research universal method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108230327A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102654902A (en) * | 2012-01-16 | 2012-09-05 | 江南大学 | Contour vector feature-based embedded real-time image matching method |
| CN103499297A (en) * | 2013-10-25 | 2014-01-08 | 爱科维申科技（天津）有限公司 | CCD (Charge Coupled Device)-based high-accuracy measuring method |
| CN103914707A (en) * | 2014-04-15 | 2014-07-09 | 广西交通投资集团有限公司 | Green channel product auxiliary discriminating method based on support vector machine |
| CN104574408A (en) * | 2015-01-16 | 2015-04-29 | 东华大学 | Industry transparent film package detecting method and device based on shape feature extraction |
| CN105701476A (en) * | 2016-02-04 | 2016-06-22 | 合肥泰禾光电科技股份有限公司 | Machine vision-based automatic identification system and method for production line products |
| CN106204618A (en) * | 2016-07-20 | 2016-12-07 | 南京文采科技有限责任公司 | Product surface of package defects detection based on machine vision and sorting technique |
-
2016
- 2016-12-14 CN CN201611155676.3A patent/CN108230327A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102654902A (en) * | 2012-01-16 | 2012-09-05 | 江南大学 | Contour vector feature-based embedded real-time image matching method |
| CN103499297A (en) * | 2013-10-25 | 2014-01-08 | 爱科维申科技（天津）有限公司 | CCD (Charge Coupled Device)-based high-accuracy measuring method |
| CN103914707A (en) * | 2014-04-15 | 2014-07-09 | 广西交通投资集团有限公司 | Green channel product auxiliary discriminating method based on support vector machine |
| CN104574408A (en) * | 2015-01-16 | 2015-04-29 | 东华大学 | Industry transparent film package detecting method and device based on shape feature extraction |
| CN105701476A (en) * | 2016-02-04 | 2016-06-22 | 合肥泰禾光电科技股份有限公司 | Machine vision-based automatic identification system and method for production line products |
| CN106204618A (en) * | 2016-07-20 | 2016-12-07 | 南京文采科技有限责任公司 | Product surface of package defects detection based on machine vision and sorting technique |
Non-Patent Citations (3)
| Title |
|---|
| 李勇等: ""基于机器视觉的零件表面瑕疵自动分拣系统设计"", 《自动化与仪器仪表》 * |
| 焦亮等: ""基于机器视觉的随机纹理瓷砖的分选系统"", 《计算机系统应用》 * |
| 金灿: ""基于HALCON的印刷图像质量检测技术研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109870983B (en) | Method and device for processing tray stack image and system for warehousing goods picking | |
| CN109800864B (en) | Robot active learning method based on image input | |
| CN113192040A (en) | Fabric flaw detection method based on YOLO v4 improved algorithm | |
| Alidoost et al. | A CNN-based approach for automatic building detection and recognition of roof types using a single aerial image | |
| CN103729885B (en) | Various visual angles projection registers united Freehandhand-drawing scene three-dimensional modeling method with three-dimensional | |
| CN111507976B (en) | Defect detection method and system based on multi-angle imaging | |
| CN109598211A (en) | A kind of real-time dynamic human face recognition methods and system | |
| CN111553949A (en) | Positioning and grabbing method for irregular workpiece based on single-frame RGB-D image deep learning | |
| CN108648194A (en) | Based on the segmentation of CAD model Three-dimensional target recognition and pose measuring method and device | |
| CN104637066B (en) | The quick framework extraction method of bianry image based on sequential refinement | |
| Hussain et al. | A simple and efficient deep learning-based framework for automatic fruit recognition | |
| Roggiolani et al. | Hierarchical approach for joint semantic, plant instance, and leaf instance segmentation in the agricultural domain | |
| CN109584206A (en) | The synthetic method of the training sample of neural network in piece surface Defect Detection | |
| Zheng et al. | Feature enhancement for multi-scale object detection | |
| Yang et al. | An end-to-end deep learning change detection framework for remote sensing images | |
| CN111724354A (en) | Image processing-based method for measuring spike length and small spike number of multiple wheat | |
| CN113628170B (en) | Laser line extraction method and system based on deep learning | |
| CN104778683B (en) | A kind of multi-modality images dividing method based on Functional Mapping | |
| CN117079125A (en) | Kiwi fruit pollination flower identification method based on improved YOLOv5 | |
| CN111368637A (en) | Multi-mask convolution neural network-based object recognition method for transfer robot | |
| CN108230327A (en) | A kind of packaging location based on MVP platforms and sort research universal method | |
| CN106056575A (en) | Image matching method based on object similarity recommended algorithm | |
| CN116051808A (en) | YOLOv 5-based lightweight part identification and positioning method | |
| CN114723757A (en) | High-precision wafer defect detection method and system based on deep learning algorithm | |
| CN103942764B (en) | A kind of two dimensional optical fiber spectrum picture based on module analysis technology repairs algorithm |
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 |