CN103793712B - Image recognition method and system based on edge geometric features - Google Patents
Image recognition method and system based on edge geometric features Download PDFInfo
- Publication number
- CN103793712B CN103793712B CN201410055598.4A CN201410055598A CN103793712B CN 103793712 B CN103793712 B CN 103793712B CN 201410055598 A CN201410055598 A CN 201410055598A CN 103793712 B CN103793712 B CN 103793712B
- Authority
- CN
- China
- Prior art keywords
- profile
- workpiece
- subsequence
- energy value
- 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.)
- Active
Links
Abstract
The invention discloses an image recognition method based on edge geometric features. The image recognition method based on the edge geometric features comprises the steps that filtering processing is carried out on an original image of a workpiece; binarization processing is carried out on the image after the filter processing; the image after the binarization processing is searched to obtain a contour sequence set; a qualified workpiece contour is screened out of the contour sequence set; a minimum enclosing rectangle of the workpiece contour is determined, and the central point of the workpiece contour is determined; four subsequences of the workpiece contour are intercepted in the directions of the four edges of the minimum enclosing rectangle; energy values of all the four subsequences are calculated by using edge geometric feature operators; the subsequence with the maximum energy value in the four subsequences is determined, and direction angles of the workpiece are determined. The invention further provides a corresponding image recognition system. According to the image recognition method and system based on the edge geometric features, a template database of workpieces does not need to be established, storage consumption is lowered, calculation time is reduced, the requirement for high speed real-time performance can be met, and meanwhile the image recognition method and system have high fast recognition capability.
Description
Technical field
The invention belongs to machine vision technical field of image processing, it is based on edge geometric properties more particularly, to a kind of
Image-recognizing method and system.
Background technology
Machine vision is one of important research field of computer science, detects identification skill in conjunction with light, mechanical, electrical integrated application
Art, development is very rapid.Main research category includes Image Feature Detection, profile expression, the segmentation of feature based, range image
Analysis, shape and expression, stereoscopic vision, motion analysiss, color vision, active vision, self-calibration system, object detection,
Two dimension and Three-dimension object recognition and positioning etc..Its range of application also expanding day, is related to robot, industrial detection, object knowledge
Not, the numerous areas such as medical image analysis, military navigation and traffic administration.
Intelligent assembly using machine vision technique can greatly shorten time of product development, in the situation not changing hardware
The various parts of lower process.Machine vision technique can be not only used for completing some seem very simple (as auto parts assembling
Deng) task, more can realize under severe or harmful working environment assemble.At present, in industrial circle, had a lot of with
Mechanical hand, visual system are that the robot system with perception of main body enters the practical stage.Such as transistor robot welding system,
Flange of pipe welding robot, the people that puts together machines having vision, automotive wheel load automatic system taking turns firm operation etc..Machine
People applies the mode of vision to have many kinds, such as provides feedback control information to mechanical hand positioning and tracking target;Determine and distinguish
The location fix of other part is to pick up part;Control the assembling to part;Guiding weld seam robot etc..
In machine vision applications, can workpiece identification ability depend primarily on image processing algorithm and obtain stable workpiece
Feature.Early in 1972, R.O.Duda and P.E.Hart propose visual field found using Hough transformation straight line, circle and its
His simple shape.Hereafter, a lot of characteristics algorithms extracting image information are occurred in that.In 1988, C.Harris and
M.Stephens proposes a kind of point feature extraction operator Harris Corner Detection Algorithm based on signal, carrys out generation using characteristic point
The content of table image.Serge Belongie, Jitendra Malik, Jan Puzicha proposed based on shape in 2000
The form fit of context and target identification method, form feature histogram by counting the distribution probability of different characteristic point.
D.G.Lowe has delivered SIFT algorithm in paper within 2004, using scale invariant feature conversion described local feature, by meter
The energy value calculating testing image with template to determine best match object.
The domestic algorithm that it is also proposed much practicalities in machine vision applications, Gu Hongxun of HUST et al. carries
Go out the flat shape based on sub- centre of form collection Hough distance and identify new method.Chen Dong, Wang Yan of Harbin Institute of Technology proposes to change
The method of the Fourier descriptors entered, extracts target all immovable feature under any affine transformation, and to six kinds of aircrafts
Carry out feature extraction and identify.The Wang Tao of Tsing-Hua University, Liu Wenyin, grandson family wide etc. using continuous based on Optimum Polygonal Approximation of Digital Curves
Fourier transform method calculates Fourier descriptors, and eliminates the side of border starting point phase effect by the principal direction of shape
Method, defines the new rotation translation that has to carry out shape recognition with the normalization Fourier descriptors of Scale invariant.Zhejiang is big
Tang Guoliang learning et al. is identified to five kinds of Aircraft Targets from two-dimensional digital image using not bending moment and standard square.Beijing postal
Wang Botao of TV university et al. have studied relative moment and its application in Geometric Shape Recognition.The Huang of Shanghai Communications University is red
Gorgeous, Yang Huangpu have studied the machine components shape recognition based on higher order neural network, proposes a kind of on-line automatic inspection of machine components
The Shape Recognition System surveyed.This system with the length on each side of part, angle, central angle and representing with 4 features of adjacent side angle
The shape of part.
The above-mentioned image processing algorithm being applied in machine vision referring to, the wherein description of the image information feature of proposition
Son is theoretically to achieve some more to have robustness mostly, and is not subject to space scaling, the feature affecting such as translates, rotate
Method for expressing.But it is difficult to adapting to severe work on the spot environment and adding for Automated assembly application in commercial production
The requirement of work technique, practical ability is restricted.And the improved practical algorithm of other, then need first to mesh to be detected
Mark is analyzed and sets up template database, just must can find Optimum Matching knot by the template matching of traversal formula in detection
Really, space complexity and time complexity all can increase.
Content of the invention
Complex environment for manufacturing industry existing machine vision automatic assembling scene and high real-time requires, the mesh of the present invention
Be provide a kind of quickly, the workpiece pose recognizer of template database need not be set up.This algorithm can be complicated
Extract in two dimensional surface the stable Geometry edge feature of workpiece in industrial noise environment and each class workpiece is automatically determined uniquely
Deflection and position.And result will send to industrial robot in real time, thus completing the sort operation of batch workpiece.
For achieving the above object, the invention provides a kind of image-recognizing method based on edge geometric properties, described side
Method comprises the following steps:
S1:The original image of workpiece is filtered processing, to remove noise;
S2:According to selected threshold value, binary conversion treatment is carried out to the image after above-mentioned Filtering Processing;
S3:Profile lookup algorithm is used to the image after binary conversion treatment, obtains profile sequence sets;
S4:It is used geometrical rule as condition, from above-mentioned profile sequence sets, filter out qualified workpiece profile;
S5:Determine the minimum enclosed rectangle of above-mentioned workpiece profile, and determine the central point of above-mentioned workpiece profile;
S6:Direction along the four edges of above-mentioned minimum enclosed rectangle intercepts four subsequences of above-mentioned workpiece profile;
S7:Calculate the energy value of each subsequence in aforementioned four subsequence using edge geometric properties operator;
S8:Relatively in aforementioned four subsequence, the size of the energy value of each subsequence, determines the maximum subsequence of energy value,
According to the central point of the maximum subsequence of energy value and above-mentioned workpiece profile, determine the deflection of workpiece.
Preferably, described step S3 is specially:
Entire image after traversal binary conversion treatment, finds the separation of simultaneously labelling all 0 and 1 pixel, each group of closure
Separation constitute border sequences;
All border sequences detecting are carried out classification in the form of profile tree integrate, obtain the profile sequence of entire image
Row collection.
Preferably, described step S4 is specially:By arranging profile threshold value, exclude error detection from above-mentioned profile sequence sets
Profile sequence, obtain workpiece profile, wherein said profile threshold value includes size and the length-width ratio of profile.
Preferably, described step S5 specifically includes:
Got stuck algorithm using rotation, calculate workpiece profile along four extreme points of two change in coordinate axis direction;
In the rectangular area that aforementioned four extreme point corresponding coordinate axess parallel lines are limited, circumscribed with any bar profile
Line constructs boundary rectangle for side, and calculates its area;
Rotate outer tangent line clockwise or counterclockwise, constantly construct new boundary rectangle, area minimum is external
Rectangle is defined as the minimum enclosed rectangle of workpiece profile;
Determine cornerwise intersection point of minimum enclosed rectangle, as the central point of workpiece profile.
Preferably, described step S6 specifically includes:
Respectively with the central point of minimum enclosed rectangle four edges as the center of circle, with 0.1-0.4 times of the rectangle length of side that is located for half
Circle is drawn in footpath;
Choose four profile fragment composition workpiece profile son sequence sets that workpiece profile is intercepted by four circles.
Preferably, described step S7 specifically includes:
By workpiece profile son sequence set { CL,CR,CT,CBBring energy value function intoRespectively
Obtain each edge contour energy value F (CL),F(CR),F(CT),F(CB);Wherein:
C represents workpiece profile, CL,CR,CT,CBRepresent in workpiece profile respectively and be intercepted
Left and right, upper and lower part profile subsequence;
N represents the sequence of points total quantity in subsequence C', XJ、YJRepresent subsequence C' midpoint P respectivelyJAbscissa and vertical
Sit target value.
Preferably, described step S8 specifically includes:
Energy value F (the C of two sub-sequences in four subsequences obtaining in comparison step S7L) and F (CR),F(CT) and F
(CB), whereinC represents workpiece profile, CL,CR,CT,CBRepresent respectively and to be intercepted in workpiece profile
Left and right, upper and lower part profile subsequence;
Take and differ that centering energy value the greater of large percentage in two sub-sequences as maximum energy value workpiece wheel
Wide;
By in minimum enclosed rectangle, the midpoint on the side corresponding to maximum energy value workpiece profile is designated as PD;
Obtain the center point P of workpiece profileOArrive point PDThe ray made and the angle of image coordinate system transverse axis positive direction, make
Deflection for workpiece.
It is another aspect of this invention to provide that additionally providing a kind of image identification system based on edge geometric properties, described
System is included with lower module:
Filtration module, for being filtered to original image processing, to remove noise;
Binarization block, for according to the threshold value selected, carrying out binary conversion treatment to the image after above-mentioned Filtering Processing;
Profile sequence sets acquisition module, for using profile lookup algorithm to the image after binary conversion treatment, obtains profile
Sequence sets;
Workpiece profile screening module, for being used geometrical rule as condition, filters out conjunction from above-mentioned profile sequence sets
The workpiece profile of lattice;
Minimum enclosed rectangle determining module, for determining the minimum enclosed rectangle of above-mentioned workpiece profile, and determines above-mentioned work
The central point of part profile;
Profile subsequence acquisition module, the direction for the four edges along above-mentioned minimum enclosed rectangle intercepts above-mentioned workpiece wheel
Four wide subsequences;
Subsequence energy value computing module, for calculating each son in aforementioned four subsequence using edge geometric properties operator
The energy value of sequence;
Workpiece deflection determining module, for comparing the size of the energy value of each subsequence in aforementioned four subsequence, really
The surely maximum subsequence of value, according to the central point of the maximum subsequence of energy value and above-mentioned workpiece profile, determines workpiece
Deflection.
Specifically, described subsequence energy value computing module is used for:
Workpiece profile son sequence set { the C that profile subsequence acquisition module is obtainedL,CR,CT,CBBring energy value function intoRespectively obtain each edge contour energy value F (CL), F (CR), F (CT),F(CB);Wherein:
C represents workpiece profile, CL,CR,CT,CBRepresent in workpiece profile respectively and be intercepted
Left and right, upper and lower part profile subsequence;
N represents the sequence of points total quantity in subsequence C', XJ、YJRepresent subsequence C' midpoint P respectivelyJAbscissa and vertical
Sit target value.
Specifically, described subsequence energy value computing module is used for:
Energy value F (the C of two sub-sequences in four subsequences that relatively subsequence energy value computing module obtainsL) and F
(CR),F(CT) and F (CB), whereinC represents workpiece profile, CL,CR,CT,CBRepresent workpiece profile respectively
In left and right, the upper and lower part profile subsequence that are intercepted;
Take and differ that centering energy value the greater of large percentage in two sub-sequences as maximum energy value workpiece wheel
Wide;
By in minimum enclosed rectangle, the midpoint on the side corresponding to maximum energy value workpiece profile is designated as PD;
Obtain the center point P of workpiece profileOArrive point PDThe ray made and the angle of image coordinate system transverse axis positive direction, make
Deflection for workpiece.
Image processing method for identifying workpiece pose in industrial automation sorting proposed by the present invention, need not use
The a large amount of pictures of front collection, to set up the template database of workpiece, reduce storage consumption, avoid because of template matching process simultaneously
And the calculating time expended is so that processing procedure disclosure satisfy that the high speed requirement of real-time of robot grasping manipulation.On the other hand,
Edge of work geometric properties operator proposed by the present invention has higher stability, can adapt to industry spot greasy dirt, complicated light
According to condition, in the sorting application of multiple types high-volume workpiece, there is higher quick identification ability.
Brief description
The image-recognizing method flow chart based on edge geometric properties that Fig. 1 describes for the present invention;
Fig. 2 is that original image is filtered out with the exemplary plot after denoising;
Fig. 3 is that Fig. 2 is carried out with the binaryzation exemplary plot that binary conversion treatment obtains;
Fig. 4 is that bianry image is carried out with the profile sequence exemplary plot that profile lookup obtains;
Fig. 5 is using the workpiece profile exemplary plot obtaining after geometrical rule screening;
Fig. 6 is to position green minimum enclosed rectangle exemplary plot in workpiece profile in figure;
Fig. 7 is intercepting local configuration subsequence exemplary plot on workpiece profile;
Fig. 8 is the maximum profile subsequence of energy value and unique deflection arrow exemplary plot;
Fig. 9 is the image identification system block diagram in the present invention based on edge geometric properties.
Specific embodiment
In order that the objects, technical solutions and advantages of the present invention become more apparent, below in conjunction with drawings and Examples, right
The present invention is further elaborated.It should be appreciated that specific embodiment described herein is only in order to explain the present invention, and
It is not used in the restriction present invention.As long as additionally, involved technical characteristic in each embodiment of invention described below
The conflict of not constituting each other just can be mutually combined.
Step S1, the original image for the workpiece obtaining is filtered process to remove noise.Clap for site environment
The original image taken the photograph, it is possible to use medium filtering, to eliminate small-sized white noise point, obtains Fig. 2.
Step S2, carries out binary conversion treatment to the image after Filtering Processing.Result such as Fig. 3, can be according to live ring
Suitable threshold value is selected in border, converts gray images into bianry image.If the photoenvironment at workpiece sorting scene is relatively stable,
Then binary conversion treatment can be carried out using fixed threshold, if the intensity of illumination at crawl scene has a certain degree of fluctuation, excellent
Adaptive Thresholding is selected to obtain bianry image.
Step S3, obtains profile sequence sets using profile lookup algorithm from bianry image.Can adopt with row scanning or
Column scan or other scan modes traversal entire image, find the separation of simultaneously labelling all 0 and 1 pixel, each group of closure
Separation constitutes border sequences.Then, all border sequences detecting are carried out classification in the form of profile tree to integrate.
Just the profile sequence sets of entire image have been obtained:{C1, C2, C3,…,CJ,…CN}.Profile lookup result is as shown in figure 4, can see
Go out, in figure, in addition to the larger workpiece profile in middle part, also has many relatively especially little profiles and is located on the left of image and the right side
Upper angle.
Step S4, is used geometrical rule as constraints, excludes the profile sequence of erroneous judgement, filter out qualified work
Part profile.Workpiece has the geometric shape of comparatively rule mostly, and its physical dimension mobility scale is also certain
In the range of.By arranging the size of profile, the isoparametric threshold value of length-width ratio, the profile sequence of error detection can be excluded, final
To workpiece profile sequence:C=CJ.
Can be according to the basic geometric shape information such as size of workpiece, to all profiles finding in previous step
Selected, just obtained qualified workpiece profile, as shown in Figure 5.In figure only exists the workpiece profile positioned at image center, and
It is located at left side vertical direction before and the profile group in the upper right corner has been excluded.
Step S5, the minimum enclosed rectangle of positioning workpiece profile.Got stuck algorithm with rotation, first calculate profile sequence along two
Four extreme points of individual change in coordinate axis direction.Then, in the rectangular area that extreme point corresponding coordinate axess parallel lines are limited, with
The outer tangent line of any bar profile constructs boundary rectangle for side, and calculates its area.Last rotation clockwise or counterclockwise is outer
Tangent line, constantly constructs new boundary rectangle.Then area reckling is the minimum enclosed rectangle of profile sequence.Positioning result is as schemed
Shown in 6, the minimum enclosed rectangle of the workpiece profile that in figure rectangle as searches.Record its diagonal intersection point, for this simultaneously
Embodiment is the white round dot that in figure is located at workpiece centre(464,443), as workpiece profile geometric center anchor point.
Step S6, intercepts contour line subsequence along boundary rectangle sideline direction.With the central point of rectangle four edges it is respectively
The center of circle, with suitable fixed length(It is generally 0.1-0.4 times of the place rectangle length of side)Draw circle for radius, embodiments herein selects radius
For the center of circle place rectangle length of side 0.2 times.Then four profile fragment composition workpiece that workpiece profile is intercepted by four circles are chosen
Profile son sequence setRepresent left and right, upper and lower part profile being intercepted respectively
Sequence)., as shown in fig. 7, each cross-talk sequence is made up of one group of point set, each point is in figure for the four profile subsequences intercepting
Use white circle labelling.Each group of white circular is punctuated the profile subsequence continuously getting up in i.e. pie graph.
Step S7, calculates the energy value of each profile subsequence using edge geometric properties operator.Previous step intercepts
Four subsequences, each sequence is made up of the coordinate of one group of pixel, and every group of point set is to an expression in image
The approximate fits of the curve of workpiece portion feature.Define the geometric properties that energy value function F (C') characterizes curve, size is antithetical phrase
Sequence point set carries out the residual sum of squares (RSS) of first-order linear matching.C' represents profile sequence to be calculated, each of sequence point
PJBy(XJ、YJ)Coordinate is constituted.Then energy value function calculating process is as follows.
First, ask for intermediate variable SXX,SYY,SXY:
Wherein, N represents the sequence of points total quantity in subsequence C', XJ、YJRepresent subsequence C' midpoint P respectivelyJAbscissa
Value with vertical coordinate.
Then calculate correlation coefficient:
Finally obtain ability value function formula:
In this embodiment, by workpiece profile son sequence set { CL,CR,CT,CBBring energy value function into, respectively obtain each
Edge contour energy value F (CL)=34.73,F(CR)=28.08,F(CT)=3.88,F(CB)=3.78.
Step S8, relatively each edge energy value simultaneously determine workpiece deflection.The pose of workpiece includes the position coordinateses of workpiece
And attitude angle, and position coordinateses have been obtained by minimum enclosed rectangle in step s 5, are designated as PO.It is respectively compared in step S7
The two couples of edge energy value F (C obtainingL)(34.73)With F (CR)(28.08),F(CT)(3.88)With F (CB)(3.78), it is two right to take
That a pair " F (C of middle difference large percentageL) and F (CR) " in energy value the greater F (CL) maximum as workpiece profile energy value
Person, then by the maximum profile subsequence of energy value, the corresponding center of circle is designated as P in step s 6D.As shown in figure 8, then workpiece
Deflection size is defined as by workpiece centre point POTo centre point PDThe ray made(In figure white arrow)Horizontal with image coordinate system
166.64 ° of the angle of axle positive direction.
Further, present invention also offers a kind of image identification system based on edge geometric properties, as shown in figure 9,
Described system is included with lower module:
Filtration module, for being filtered to original image processing, to remove noise;
Binarization block, for according to the threshold value selected, carrying out binary conversion treatment to the image after above-mentioned Filtering Processing;
Profile sequence sets acquisition module, for using profile lookup algorithm to the image after binary conversion treatment, obtains profile
Sequence sets;
Workpiece profile screening module, for being used geometrical rule as condition, filters out conjunction from above-mentioned profile sequence sets
The workpiece profile of lattice;
Minimum enclosed rectangle determining module, for determining the minimum enclosed rectangle of above-mentioned workpiece profile, and determines above-mentioned work
The central point of part profile;
Profile subsequence acquisition module, the direction for the four edges along above-mentioned minimum enclosed rectangle intercepts above-mentioned workpiece wheel
Four wide subsequences;
Subsequence energy value computing module, for calculating each son in aforementioned four subsequence using edge geometric properties operator
The energy value of sequence;
Workpiece deflection determining module, for comparing the size of the energy value of each subsequence in aforementioned four subsequence, really
The surely maximum subsequence of value, according to the central point of the maximum subsequence of energy value and above-mentioned workpiece profile, determines workpiece
Deflection.
Specifically, described subsequence energy value computing module is used for:
Workpiece profile son sequence set { the C that profile subsequence acquisition module is obtainedL,CR,CT,CBBring energy value function intoRespectively obtain each edge contour energy value F (CL),F(CR),F(CT),F(CB);Wherein:
C represents workpiece profile, CL,CR,CT,CBRepresent in workpiece profile respectively and be intercepted
Left and right, upper and lower part profile subsequence;
N represents the sequence of points total quantity in subsequence C', XJ、YJRepresent subsequence C' midpoint P respectivelyJAbscissa and vertical
Sit target value.
Specifically, described subsequence energy value computing module is used for:
Energy value F (the C of two sub-sequences in four subsequences that relatively subsequence energy value computing module obtainsL) and F
(CR),F(CT) and F (CB), whereinC represents workpiece profile, CL,CR,CT,CBRepresent workpiece profile respectively
In left and right, the upper and lower part profile subsequence that are intercepted;
Take and differ that centering energy value the greater of large percentage in two sub-sequences as maximum energy value workpiece wheel
Wide;
By in minimum enclosed rectangle, the midpoint on the side corresponding to maximum energy value workpiece profile is designated as PD;
Obtain the center point P of workpiece profileOArrive point PDThe ray made and the angle of image coordinate system transverse axis positive direction, make
Deflection for workpiece.
As it will be easily appreciated by one skilled in the art that the foregoing is only presently preferred embodiments of the present invention, not in order to
Limit the present invention, all any modification, equivalent and improvement made within the spirit and principles in the present invention etc., all should comprise
Within protection scope of the present invention.
Claims (5)
1. a kind of image-recognizing method based on edge geometric properties is it is characterised in that the method comprising the steps of:
S1:The original image of workpiece is filtered processing, to remove noise;
S2:According to selected threshold value, binary conversion treatment is carried out to the image after above-mentioned Filtering Processing;
S3:Profile lookup algorithm is used to the image after binary conversion treatment, obtains profile sequence sets;
S4:It is used geometrical rule as condition, from above-mentioned profile sequence sets, filter out qualified workpiece profile;
S5:Determine the minimum enclosed rectangle of above-mentioned workpiece profile, and determine the central point of above-mentioned workpiece profile;
S6:Direction along the four edges of above-mentioned minimum enclosed rectangle intercepts four subsequences of above-mentioned workpiece profile;Specifically include:
Respectively with the central point of minimum enclosed rectangle four edges as the center of circle, draw with 0.1-0.4 times of the rectangle length of side that is located for radius
Circle;
Choose four profile fragment composition workpiece profile son sequence sets that workpiece profile is intercepted by four circles;
S7:Calculate the energy value of each subsequence in aforementioned four subsequence using edge geometric properties operator, specifically include:
By workpiece profile son sequence set { CL,CR,CT,CBBring energy value function intoRespectively obtain each
Edge contour energy value F (CL),F(CR),F(CT),F(CB);Wherein:
C represents workpiece profile, CL,CR,CT,CBRepresent respectively the left side being intercepted in workpiece profile,
Right, upper and lower part profile subsequence;
N represents the sequence of points total quantity in subsequence C', XJ、YJRepresent subsequence C' midpoint P respectivelyJAbscissa and vertical coordinate
Value;
S8:Relatively in aforementioned four subsequence, the size of the energy value of each subsequence, determines the maximum subsequence of energy value, according to
The maximum subsequence of energy value and the central point of above-mentioned workpiece profile, determine the deflection of workpiece;
Step S8 specifically includes:
Energy value F (the C of two sub-sequences in four subsequences obtaining in comparison step S7L) and F (CR),F(CT) and F (CB),
WhereinC represents workpiece profile, CL,CR,CT,CBRepresent respectively be intercepted in workpiece profile left and right,
Upper and lower part profile subsequence;
Take and differ that centering energy value the greater of large percentage in two sub-sequences as maximum energy value workpiece profile;
By in minimum enclosed rectangle, the midpoint on the side corresponding to maximum energy value workpiece profile is designated as PD;
Obtain the center point P of workpiece profileOArrive point PDThe ray made and the angle of image coordinate system transverse axis positive direction, as work
The deflection of part.
2. image-recognizing method as claimed in claim 1 is it is characterised in that described step S3 is specially:
Entire image after traversal binary conversion treatment, find and labelling all 0 and 1 pixel separation, each group of closure point
Boundary's point constitutes border sequences;
All border sequences detecting are carried out classification in the form of profile tree integrate, obtain the profile sequence of entire image
Collection.
3. image-recognizing method as claimed in claim 1 is it is characterised in that described step S4 is specially:By arranging profile
Threshold value, excludes the profile sequence of error detection from above-mentioned profile sequence sets, obtains workpiece profile, wherein said profile threshold value includes
The size of profile and length-width ratio.
4. image-recognizing method as claimed in claim 1 is it is characterised in that described step S5 specifically includes:
Got stuck algorithm using rotation, calculate workpiece profile along four extreme points of two change in coordinate axis direction;
In the rectangular area that aforementioned four extreme point corresponding coordinate axess parallel lines are limited, with the outer tangent line of any bar profile it is
Side constructs boundary rectangle, and calculates its area;
Rotate outer tangent line clockwise or counterclockwise, constantly construct new boundary rectangle, by the boundary rectangle that area is minimum
It is defined as the minimum enclosed rectangle of workpiece profile;
Determine cornerwise intersection point of minimum enclosed rectangle, as the central point of workpiece profile.
5. a kind of image identification system based on edge geometric properties is it is characterised in that described system is included with lower module:
Filtration module, for being filtered to original image processing, to remove noise;
Binarization block, for according to the threshold value selected, carrying out binary conversion treatment to the image after above-mentioned Filtering Processing;
Profile sequence sets acquisition module, for using profile lookup algorithm to the image after binary conversion treatment, obtains profile sequence
Collection;
Workpiece profile screening module, for being used geometrical rule as condition, it is qualified to filter out from above-mentioned profile sequence sets
Workpiece profile;
Minimum enclosed rectangle determining module, for determining the minimum enclosed rectangle of above-mentioned workpiece profile, and determines above-mentioned workpiece wheel
Wide central point;
Profile subsequence acquisition module, the direction for the four edges along above-mentioned minimum enclosed rectangle intercepts above-mentioned workpiece profile
Four subsequences;Specifically respectively with the central point of minimum enclosed rectangle four edges as the center of circle, with the 0.1- of the rectangle length of side that is located
Draw circle for radius for 0.4 times;Choose four profile fragment composition workpiece profile son sequence sets that workpiece profile is intercepted by four circles;
Subsequence energy value computing module, for calculating each subsequence in aforementioned four subsequence using edge geometric properties operator
Energy value;Specifically for:
Workpiece profile son sequence set { the C that profile subsequence acquisition module is obtainedL,CR,CT,CBBring energy value function intoRespectively obtain each edge contour energy value F (CL),F(CR),F(CT),F(CB);Wherein:
C represents workpiece profile, CL,CR,CT,CBRepresent respectively the left side being intercepted in workpiece profile,
Right, upper and lower part profile subsequence;
N represents the sequence of points total quantity in subsequence C', XJ、YJRepresent subsequence C' midpoint P respectivelyJAbscissa and vertical coordinate
Value;
Workpiece deflection determining module, for comparing the size of the energy value of each subsequence in aforementioned four subsequence, determines energy
The maximum subsequence of value, according to the central point of the maximum subsequence of energy value and above-mentioned workpiece profile, determines the direction of workpiece
Angle;
Described subsequence energy value computing module specifically for:
Energy value F (the C of two sub-sequences in four subsequences that relatively subsequence energy value computing module obtainsL) and F (CR),F
(CT) and F (CB), whereinC represents workpiece profile, CL,CR,CT,CBRepresent respectively in workpiece profile and cut
Left and right, the upper and lower part profile subsequence taking;
Take and differ that centering energy value the greater of large percentage in two sub-sequences as maximum energy value workpiece profile;
By in minimum enclosed rectangle, the midpoint on the side corresponding to maximum energy value workpiece profile is designated as PD;
Obtain the center point P of workpiece profileOArrive point PDThe ray made and the angle of image coordinate system transverse axis positive direction, as work
The deflection of part.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410055598.4A CN103793712B (en) | 2014-02-19 | 2014-02-19 | Image recognition method and system based on edge geometric features |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410055598.4A CN103793712B (en) | 2014-02-19 | 2014-02-19 | Image recognition method and system based on edge geometric features |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103793712A CN103793712A (en) | 2014-05-14 |
CN103793712B true CN103793712B (en) | 2017-02-08 |
Family
ID=50669355
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410055598.4A Active CN103793712B (en) | 2014-02-19 | 2014-02-19 | Image recognition method and system based on edge geometric features |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103793712B (en) |
Families Citing this family (31)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106651825A (en) * | 2015-11-03 | 2017-05-10 | 中国科学院沈阳计算技术研究所有限公司 | Workpiece positioning and identification method based on image segmentation |
DE102016120131B4 (en) | 2016-10-21 | 2020-08-06 | Trumpf Werkzeugmaschinen Gmbh + Co. Kg | Sorting support method and flatbed machine tool |
DE102016120132A1 (en) | 2016-10-21 | 2018-04-26 | Trumpf Werkzeugmaschinen Gmbh + Co. Kg | Workpiece collection unit and method for supporting the machining of workpieces |
CN106393376B (en) * | 2016-10-21 | 2019-03-05 | 北京源著智能科技有限公司 | The processing method and system of plate |
CN106493086B (en) * | 2016-10-21 | 2019-03-05 | 北京源著智能科技有限公司 | The sorting processing method and system of plate |
CN106650608B (en) * | 2016-10-31 | 2019-11-26 | 广东工业大学 | The recognition methods of rectangle posting in a kind of no anchor point paper |
CN108073932A (en) * | 2016-11-16 | 2018-05-25 | 中国科学院沈阳计算技术研究所有限公司 | A kind of workpiece image Feature extraction and recognition method based on Gabor filtering |
CN106846404A (en) * | 2017-01-16 | 2017-06-13 | 河海大学常州校区 | A kind of quick position and posture detection method based on workpiece inside and outside contour geometric properties |
CN106952307B (en) * | 2017-03-22 | 2020-01-07 | 广东工业大学 | Method and system for positioning sphere in image |
CN107084666B (en) * | 2017-05-10 | 2020-08-28 | 中国计量大学 | Comprehensive detection method for size of brake pad based on machine vision |
CN107423770B (en) * | 2017-08-03 | 2021-03-26 | 广东顺德西安交通大学研究院 | Robot vision positioning method for high-speed production line |
CN109325949A (en) * | 2018-10-17 | 2019-02-12 | 合肥常青机械股份有限公司 | A kind of auto parts processing detection information extracting method |
CN109741394B (en) * | 2018-12-10 | 2021-02-26 | 北京拓尔思信息技术股份有限公司 | Image processing method, image processing device, electronic equipment and storage medium |
CN109658402B (en) * | 2018-12-17 | 2023-04-18 | 中山大学 | Automatic detection method for geometric dimension of industrial profile based on computer vision imaging |
CN109741314A (en) * | 2018-12-29 | 2019-05-10 | 广州博通信息技术有限公司 | A kind of visible detection method and system of part |
CN109506569B (en) * | 2019-01-08 | 2020-04-07 | 大连理工大学 | Method for monitoring three-dimensional sizes of cubic and columnar crystals in crystallization process based on binocular vision |
CN110110760A (en) * | 2019-04-17 | 2019-08-09 | 浙江工业大学 | A kind of workpiece positioning and recognition methods based on machine vision |
CN110102906A (en) * | 2019-05-10 | 2019-08-09 | 定颖电子(黄石)有限公司 | A kind of radium carves stamp system and radium carves stamp technique |
CN110298292A (en) * | 2019-06-25 | 2019-10-01 | 东北大学 | Detection method is grabbed when the high-precision real of rule-based object polygon Corner Detection |
CN110956181A (en) * | 2019-07-04 | 2020-04-03 | 中联重科股份有限公司 | Crane counterweight block detection method and equipment and crane |
CN110712202B (en) * | 2019-09-24 | 2021-07-16 | 鲁班嫡系机器人(深圳)有限公司 | Special-shaped component grabbing method, device and system, control device and storage medium |
CN111290582B (en) * | 2020-02-29 | 2021-09-21 | 华南理工大学 | Projection interaction area positioning method based on improved linear detection |
CN111633470B (en) * | 2020-05-08 | 2021-12-14 | 广东长盈精密技术有限公司 | Mechanical coordinate detection method and device for product coordinate origin and computer equipment |
WO2021227286A1 (en) * | 2020-05-09 | 2021-11-18 | 中南大学 | Mobile testing method and apparatus for regularity of honeycomb product with ultra-large area |
CN111583239B (en) * | 2020-05-09 | 2021-03-30 | 中南大学 | Honeycomb structure geometric regularity image recognition method and system |
CN112215893B (en) * | 2020-10-28 | 2022-10-28 | 安徽农业大学 | Method, device and equipment for determining target two-dimensional center coordinate point and ranging system |
CN112727704B (en) * | 2020-12-15 | 2021-11-30 | 北京天泽智云科技有限公司 | Method and system for monitoring corrosion of leading edge of blade |
CN113358058B (en) * | 2021-06-07 | 2022-10-14 | 四川航天长征装备制造有限公司 | Computer vision detection method for weld contour features based on discrete sequence points |
CN113269835A (en) * | 2021-07-19 | 2021-08-17 | 季华实验室 | Industrial part pose identification method and device based on contour features and electronic equipment |
CN115063421B (en) * | 2022-08-16 | 2022-10-28 | 成都数联云算科技有限公司 | Pole piece region detection method, system and device, medium and defect detection method |
CN116188761B (en) * | 2023-02-02 | 2024-02-02 | 杭州图谱光电科技有限公司 | Automatic identification and positioning method and device for regular-shape array diagram |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2002024804A (en) * | 2000-07-07 | 2002-01-25 | Matsushita Electric Ind Co Ltd | Part recognition data forming method, forming apparatus electronic component mounting apparatus, and recording medium |
CN101059836A (en) * | 2007-06-01 | 2007-10-24 | 华南理工大学 | Human eye positioning and human eye state recognition method |
CN101211341A (en) * | 2006-12-29 | 2008-07-02 | 上海芯盛电子科技有限公司 | Image intelligent mode recognition and searching method |
CN101398899A (en) * | 2007-09-26 | 2009-04-01 | 鸿富锦精密工业(深圳)有限公司 | System and method for auto- recognizing metal plate part edge type |
-
2014
- 2014-02-19 CN CN201410055598.4A patent/CN103793712B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2002024804A (en) * | 2000-07-07 | 2002-01-25 | Matsushita Electric Ind Co Ltd | Part recognition data forming method, forming apparatus electronic component mounting apparatus, and recording medium |
CN101211341A (en) * | 2006-12-29 | 2008-07-02 | 上海芯盛电子科技有限公司 | Image intelligent mode recognition and searching method |
CN101059836A (en) * | 2007-06-01 | 2007-10-24 | 华南理工大学 | Human eye positioning and human eye state recognition method |
CN101398899A (en) * | 2007-09-26 | 2009-04-01 | 鸿富锦精密工业(深圳)有限公司 | System and method for auto- recognizing metal plate part edge type |
Non-Patent Citations (3)
Title |
---|
一种提取目标图像最小外接矩形的快速方法;卢蓉;《计算机工程》;20101105;全文 * |
基于几何特征的目标识别及跟踪技术的研究;张强;《中国优秀硕士学位论文全文数据库》;20090215;全文 * |
基于零件轮廓形心与最小外接矩形的图形配准方法;陈栋梁;《制造业自动化》;20120425;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN103793712A (en) | 2014-05-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103793712B (en) | Image recognition method and system based on edge geometric features | |
CN111299815B (en) | Visual detection and laser cutting trajectory planning method for low-gray rubber pad | |
CN106845408B (en) | Street garbage identification method under complex environment | |
CN105740899B (en) | A kind of detection of machine vision image characteristic point and match compound optimization method | |
CN103093191B (en) | A kind of three dimensional point cloud is in conjunction with the object identification method of digital image data | |
CN111553409B (en) | Point cloud identification method based on voxel shape descriptor | |
CN105335973B (en) | Apply to the visual processing method of strip machining production line | |
CN111582123B (en) | AGV positioning method based on beacon identification and visual SLAM | |
CN107392929B (en) | Intelligent target detection and size measurement method based on human eye vision model | |
CN106355577A (en) | Method and system for quickly matching images on basis of feature states and global consistency | |
CN104933434A (en) | Image matching method combining length between perpendiculars (LBP) feature extraction method and surf feature extraction method | |
CN104090972A (en) | Image feature extraction and similarity measurement method used for three-dimensional city model retrieval | |
CN107818598B (en) | Three-dimensional point cloud map fusion method based on visual correction | |
CN101639947A (en) | Image-based plant three-dimensional shape measurement and reconstruction method and system | |
CN104123554B (en) | SIFT image characteristic extracting methods based on MMTD | |
CN106408597A (en) | Neighborhood entropy and consistency detection-based SAR (synthetic aperture radar) image registration method | |
CN106682700A (en) | Block quick matching algorithm based on key point description operator | |
CN108491786A (en) | A kind of method for detecting human face based on hierarchical network and Cluster merging | |
CN110222661B (en) | Feature extraction method for moving target identification and tracking | |
CN109920050B (en) | Single-view three-dimensional flame reconstruction method based on deep learning and thin plate spline | |
CN101833763A (en) | Method for detecting reflection image on water surface | |
CN107895166B (en) | Method for realizing target robust recognition based on feature descriptor by geometric hash method | |
CN114092478B (en) | Anomaly detection method | |
CN114898041A (en) | Improved ICP method based on luminosity error | |
CN1216343C (en) | Infrared target identification method based on unchanged rotary morphology neural net |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |