CN110533714A - Method and system based on image processing techniques detection target object maximum inscribed circle - Google Patents
Method and system based on image processing techniques detection target object maximum inscribed circle Download PDFInfo
- Publication number
- CN110533714A CN110533714A CN201910777824.2A CN201910777824A CN110533714A CN 110533714 A CN110533714 A CN 110533714A CN 201910777824 A CN201910777824 A CN 201910777824A CN 110533714 A CN110533714 A CN 110533714A
- Authority
- CN
- China
- Prior art keywords
- target object
- reference marker
- origin
- inscribed circle
- maximum inscribed
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- 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
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Geometry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of methods based on image processing techniques detection target object maximum inscribed circle, including step A: building origin label and reference marker;Obtaining includes target object, origin marks, the workbench orthographic view of reference marker obtains the first picture;Step B: the position of origin label and reference marker is determined on the first picture;Step C: pixel coordinate system is constructed labeled as origin with origin, extracts the coordinate of template graphics, the profile line coordinates of target object and four boundary values in reference marker;Step D: the center of circle and the radius of maximum inscribed circle are solved.The advantages of method and system provided by the invention that target object maximum inscribed circle is detected based on image processing techniques, is: can quickly determine the maximum inscribed circle of target object, improve raw material availability, reduce manual labor's cost and time, and the object of arbitrary shape can be detected effectively, it is with a wide range of applications and huge economic benefit.
Description
Technical field
The present invention relates to technical field of image processing, more particularly to detect target object most imperial palace based on image processing techniques
Connect round method and system.
Background technique
When carrying out circular product processing (such as processing of stone, jade are processed), many factories still use hand dipping true
Determine the mode of the inscribed circle of raw material, this operation mode is highly dependent upon personnel's experience, it is easy to there is the problem of waste material,
And working efficiency is low, large labor intensity, and the technological difficulties for carrying out automation processing are how fast and accurately to determine
The maximum inscribed circle of target object;The factors such as shape, texture, the disposing way of target object can all influence detection object most imperial palace
Connect the accuracy of the area of a circle;Digital image processing techniques provide technical support thus.Also there is related algorithm real in the prior art
The detection of existing object maximum inscribed circle, but have the shortcomings that accuracy rate is not high, applicability is not wide.
Summary of the invention
Technical problem to be solved by the present invention lies in provide one kind fast and accurately to examine based on digital image processing techniques
The method and system of the maximum inscribed circle of target object is surveyed, to overcome waste raw material of the existing technology, production efficiency low
The problems such as.
The present invention is to solve above-mentioned technical problem by the following technical programs:
A method of target object maximum inscribed circle is detected based on image processing techniques, including
Step A: building origin label and reference marker;By origin label and reference marker setting on the table, incite somebody to action
Target object is placed on workbench, and obtaining includes target object, origin marks, the workbench orthographic view of reference marker obtains
To the first picture;
Step B: the position of origin label and reference marker is determined on the first picture;
Step C: two-dimensional pixel coordinate system is constructed labeled as origin with origin, extracts the seat of template graphics in reference marker
Four boundary values of mark, the profile line coordinates of target object and profile line coordinates;
Step D: the center of circle for calculating the maximum inscribed circle of target object and radius are solved.
Preferably, the origin label in step A includes a circular colored spot, and reference marker includes two circular colored spots, and
Record the distance L of two circular colored spots in reference markerAB。
Preferably, the method that origin label and reference marker position are determined in step B is that display is all in the first picture
The abnormal color region inconsistent with background alternately point, the position of confirmation origin label and reference marker in alternative point.
Preferably, the origin label in step A includes a template graphics, and reference marker includes three not in same straight line
On template graphics, and the distance between record any two of them template graphics and the line of the two template graphics with
The angle of another template graphics line.
Preferably, the method for the label of determination origin described in step B and reference marker position includes:
Step I: different deformation operations is executed to template graphics and obtains different template pictures;
Step II: being divided into training set and test set for template picture, using training set training graphics processing model and with survey
Examination collection evaluation processing model;
Step III: using four template graphics in processing the first picture of model extraction, three Prototype drawings are then taken respectively
Three interior angles of group joint account of shape center position are constituted with three template graphics on reference marker when three interior angles
When triangle interior angle is equal, using these three template graphics as reference marker, marked using another template graphics as origin.
Preferably, the template graphics have rotational invariance.
Preferably, template graphics described in step A include at least having the filled circles different there are two diameter different colours
Concentric dot.
It preferably, include rotation, translation, scaling, overturning and shearing to the deformation operation that template graphics execute.
Preferably, picture generator keras.preprocessing.image.ImageDataGener is used in step I
Ator carries out deformation operation to template graphics and generates the template picture.
Preferably, training set and test set are generated to the json text for meeting Microsoft's COCO data set format respectively in step B
Part inputs training set to target detection software package Detectron and test set carries out deep learning and obtains the processing model.
Preferably, deep learning is carried out using Detectron in step II obtain processing model.
Preferably, the method for contour line of the target object in the first picture is extracted in step C are as follows: convert the first picture
For grayscale image, cv::Canny () operator is called to detect edge of the target object in the first picture, calls function cv::
FindContours () searches the profile of target object, is denoted as contour, traverses in objects' contour contour
Point, lookup are worth the smallest abscissa Xmin, the smallest ordinate Y of valuemin, the maximum abscissa X of valuemaxWith the maximum ordinate of value
Ymax;
Recording distance in reference marker is LABCoordinate A (x of the two o'clock in pixel coordinate systemA,yA), B (xB,yB)。
Preferably, the method in the center of circle and radius of the maximum inscribed circle of target object is calculated in step D are as follows: for profile
The minimum range of arbitrary point (x, y) distance profile contour in contour is denoted as min Dist (x, y, contour), then asks
The problem of taking maximum inscribed circle inside target object is converted to the problem of seeking min Dist (x, y, contour) maximum value,
I.e.
Wherein, RI indicates that the radius of maximum inscribed circle, CI are the center of circle of maximum inscribed circle;
Call function cv::pointPolygonTest (InputArray contour, Point2f pt, bool
MeasureDist), obtain
Utilize the actual distance L of two o'clock in reference markerABResult under pixel coordinate system is handled, is obtained
Wherein, RA indicates that the true value of radius, CA.x indicate the actual distance of the center of circle and coordinate origin along x-axis, CA.y table
Show the center of circle and coordinate origin along the actual distance of y-axis, CI.x indicates x-axis coordinate of the center of circle under pixel coordinate system, and CI.y is indicated
Y-axis coordinate of the center of circle under pixel coordinate system, eu_dist (A, B) indicate the length of A, B two o'clock under pixel coordinate system.
The present invention also provides a kind of systems based on image processing techniques detection target object maximum inscribed circle, including
Image collection module: building includes origin label and reference marker;Origin label and reference marker are arranged in work
Make on platform, target object be placed on workbench, obtain include target object, origin label, reference marker workbench just
Projected image obtains the first picture;
Identification module: the position of origin label and reference marker is determined on the first picture;
Coordinate system constructs module: constructing two-dimensional pixel coordinate system labeled as origin with origin in the first picture, extracts ginseng
Examine four boundary values of the coordinate of template graphics in label, the profile line coordinates of target object and profile line coordinates;
Processing module: the center of circle for calculating the maximum inscribed circle of target object and radius are solved.
Preferably, the origin label includes a template graphics, and the reference marker includes three not in same straight line
On template graphics, the identification module further include:
Graph transformation unit: different deformation operations is executed to template graphics and obtains different template pictures;
Module training unit: being divided into training set and test set for template picture, uses training set training graphics processing model
And processing model is evaluated with test set;
Identifying processing unit: using four template graphics in processing the first picture of model extraction, three are then taken respectively
Three interior angles of group joint account of template graphics center position, when three interior angles with three template graphics on reference marker
When the triangle interior angle of composition is equal, using these three template graphics as reference marker, using another template graphics as origin mark
Note.
The advantages of method and system provided by the invention based on image processing techniques detection target object maximum inscribed circle
It is: identification positioning is carried out to the point in picture, and center location and radius are calculated based on this, can quickly determine target object
Maximum inscribed circle, improve raw material availability, reduce manual labor's cost and time, and can have to the object of arbitrary shape
Effect detect, and has good robustness and adaptability, is with a wide range of applications and huge economic benefit.
Detailed description of the invention
Fig. 1 is the side for detecting target object maximum inscribed circle provided by the embodiment of the present invention based on image processing techniques
The flow chart of method;
Fig. 2 is to detect target object maximum inscribed circle based on image processing techniques provided by the embodiment of the present invention one
The flow chart of method;
Fig. 3 is the side for detecting target object maximum inscribed circle provided by the embodiment of the present invention based on image processing techniques
The schematic diagram for the template graphics that method uses;
Fig. 4 is the side for detecting target object maximum inscribed circle provided by the embodiment of the present invention based on image processing techniques
The schematic diagram for the reference marker that method is selected.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with specific embodiment, and reference
Attached drawing, the present invention is described in further detail.
Facebook company is proposed the target detection software package Detectron based on depth learning technology;Detectron
It is opening for FAIR (Facebook Artificial Intelligence Research, Facebook artificial intelligence study institute)
Source software packet, it comprises some state-of-the-art algorithm of target detection, such as Mask R-CNN, RetinaNet and Faster R-
CNN.Open-CV be one based on BSD license (open source) distribution cross-platform computer vision library, he realize image procossing with
Many general-purpose algorithms in terms of computer vision;The present invention is based on Detectron and OpenCV open source libraries to propose based on image
The method of processing technique detection target object maximum inscribed circle.
Embodiment one
It present embodiments provides combined with Figure 1 and Figure 2, a kind of based on image processing techniques detection target object maximum inscribed circle
Method, comprising:
Step A: building origin label and reference marker;By origin label and reference marker setting on the table, incite somebody to action
Target object is placed on workbench, and obtaining includes target object, origin marks, the workbench orthographic view of reference marker obtains
To the first picture;
Origin provided in this embodiment label includes a template graphics, reference marker include three not on the same line
Template graphics, i.e., three on reference marker template graphics central point line constitute triangle, record any two of them
The distance between template graphics are defined as LAB;And the corresponding triangle interior angle of its two-end-point is recorded, two-end-point is defined as point A
And B.
Step B: the position of origin label and reference marker is determined on the first picture;Specifically includes the following steps:
Step I: different deformation operations is executed to template graphics and obtains different template pictures;
The template graphics should have rotational invariance, and based on this requirement, the template graphics should be provided in round, and be
Guarantee to distinguish with the shape on target object, multiple not homochromy circular spots with one heart can be set;The present embodiment
The template graphics of middle offer refer to Fig. 3, selected two concentric black circles, wherein dot be black, rgb value be (0,
0,0), radius 0.5mm, large circle point are yellow, and rgb value is (255,255,0), radius 2mm;Use picture generator ker
As.preprocessing.image.ImageDataGenerator template graphics are rotated, are translated, are scaled, are overturn and
The a large amount of template pictures of shearing manipulation;
The present embodiment is as follows using the design parameter of ImageDataGenerator processing picture:
Step II: being divided into training set and test set for template picture, using training set training graphics processing model and with survey
Examination collection evaluation processing model;
Training set and test set are generated to the json file for meeting Microsoft's COCO data set format respectively, it is soft to target detection
Part packet Detectron input training set and test set carry out deep learning and obtain the processing model;Training set and test set pair
The json file content answered is json format, specifically:
Key name | Description | Value Types |
info | Data set basic information | info |
images | Image data list | [image] |
annotations | Annotating data list | [annotation] |
licenses | License list | [license] |
categories | Item name list | [category] |
Table 1: the file format of training set and test set
Value Types therein are json format, specific as follows:
Key name | Description | Value Types |
year | The data set time | int |
version | Data set version | str |
description | The description information of data set | str |
contributor | Contributor | str |
url | The URL of data set | str |
date_created | Creation time | datetime |
Table 2: the file format of Value Types info
Key name | Description | Value Types |
id | Picture number | int |
width | Picture width | int |
height | Picture height | int |
file_name | Filename | str |
license | Licensing number | int |
flickr_url | The URL of picture | str |
coco_url | Address of the picture in COCO data set | str |
date_captured | Obtain the time of picture | datetime |
Table 3: the file format of Value Types image
Key name | Description | Value Types |
id | Licensing number | int |
name | Licensing name | str |
url | Address of the licensing in internet | str |
Table 4: the file format of Value Types license
Key name | Description | Value Types |
id | Example number | int |
image_id | Picture number | int |
category_id | The class number of target in picture | int |
segmentation | The profile information of target | RLE or [polygon] |
area | The area of target | float |
bbox | Coordinate, width and the height of bounding box | [x,y,width,height] |
iscrowd | It whether is multiple targets, value is 0 or 1 | int |
Table 5: the file format of Value Types annotation
Key name | Description | Value Types |
id | Class number | int |
name | Item name | str |
supercategory | Parent class title | str |
Table 6: the file format of Value Types category
RLE (run-length encoding, Run- Length Coding) therein is a kind of pair of bi-level image coding in cybernetics
Mode.
Step III: using four template graphics in processing the first picture of model extraction, three Prototype drawings are then taken respectively
Three interior angles of group joint account of shape center position are constituted with three template graphics on reference marker when three interior angles
When triangle interior angle is equal, using these three template graphics as reference marker, marked using another template graphics as origin;
Based on the method for above-mentioned identification origin label and reference marker, those of ordinary skill in the art should know, this reality
Apply example used a large amount of template graphics to carry out training pattern with ensure handle model can identify template graphics in the first picture,
And the triangle that its center point carries out positioning with three template graphics central points in reference marker are constituted is compared;Due to
Whole process is realized automatically by artificial intelligence, therefore origin is arranged on the table and marks and needs to consider when reference marker to prevent
Only any two template graphics on the template graphics and reference marker on origin label are constituted and triangle on reference marker
Similar triangles.Template graphics on reference marker may be constructed arbitrary triangle, can also be set according to user and increase template
The quantity of figure is confirmed.
With reference to Fig. 4, the triangle interior angle that three template graphics are constituted is respectively set the reference marker of the present embodiment selection
It is 30 °, 60 ° and 90 °, to facilitate user to confirm and system is facilitated to carry out data calculating;And its length of the hypotenuse is defined as LAB。
Step C: two-dimensional pixel coordinate system is constructed labeled as origin with origin, extracts the seat of template graphics in reference marker
Four boundary values of mark, the profile line coordinates of target object and profile line coordinates;
The present embodiment uses NC model construction pixel coordinate system in the prior art, for the origin pair with NC model
It answers, origin marks the upper right corner for being arranged on workbench;Those of ordinary skill in the art can voluntarily determine according to the actual situation
The specific location of origin label and reference marker.
Coordinate A (the x of the point A and B in reference marker are extracted based on pixel coordinate systemA,yA), B (xB,yB)。
The method of four boundary values of the profile line coordinates and profile line coordinates of the extraction target object are as follows:
First picture is converted into grayscale image, calls cv::Canny () operator (Canny () i.e. in OpenCV database
Edge of the target object in the first picture operator, similarly hereinafter) is detected, calls function cv::findContours () search
The profile of target object is denoted as contour, traverses the point in objects' contour contour, and lookup is worth the smallest abscissa
Xmin, the smallest ordinate Y of valuemin, the maximum abscissa X of valuemaxWith the maximum ordinate Y of valuemax。
Step D: the center of circle for calculating the maximum inscribed circle of target object and radius are solved.
Min Dist is denoted as the minimum range of arbitrary point (x, y) the distance profile contour in profile contour
(x, y, contour), then the problem of seeking maximum inscribed circle inside target object be converted to seek min Dist (x, y,
Contour) the problem of maximum value, i.e.,
Wherein, RI indicates that the radius of maximum inscribed circle, CI are the center of circle of maximum inscribed circle;
Call function cv::pointPolygonTest (InputArray contour, Point2f pt, bool
MeasureDist), obtain
The result under pixel coordinate system is converted using the actual distance of bevel edge AB, is obtained
Wherein, RA indicates that the true value of radius, CA.x indicate the actual distance of the center of circle and coordinate origin along x-axis, CA.y table
Show the center of circle and coordinate origin along the actual distance of y-axis, CI.x indicates x-axis coordinate of the center of circle under pixel coordinate system, and CI.y is indicated
Y-axis coordinate of the center of circle under pixel coordinate system, eu_dist (A, B) indicate the length of A, B two o'clock under pixel coordinate system, can be with
It is calculated by the pixel coordinate of A, B two o'clock.
It is evaluated and tested by experiment, the minimal error using the maximum inscribed circle radius detected the present embodiment provides method is
0.0932mm, worst error 0.4623mm, mean error 0.2246mm, the minimal error of the center of circle in the x direction are
0.0123mm, worst error 0.9132mm, mean error 0.5317mm;The minimal error of the center of circle in y-direction is
0.0104mm, worst error 0.8965mm, mean error 0.5158mm;The above results can show that provided in this embodiment
Method based on image processing techniques detection target object maximum inscribed circle can satisfy production requirement.
Embodiment two
The difference of the present embodiment and embodiment one is only that and is shown in the first picture by image processing algorithm in stepb
All alternately points with the inconsistent abnormal color region of background out, then determined in alternative point by user origin label and
Reference marker, subsequent calculation processing step are the same as example 1.
It should be noted that since embodiment one selects artificial intelligence to identify reference marker and origin label, so
The template graphics for needing particular color to combine need Prototype drawing there are three at least tools to be trained to artificial intelligence model
The reference marker of shape carries out identification determination;And the present embodiment passes through user's initiative recognition, so origin label and reference marker
Setting is freer, can be set to any color spot or pattern or fringe area, for the ease of quickly determining the center in the region
Point is set to border circular areas in preferred embodiment, and carries out color filling;Two circles are wherein only set on reference marker
Color spot is L with the central point Distance positioning of the two circular colored spotsAB, to carry out remaining calculating step.
Based on the above embodiment, those of ordinary skill in the art are it is to be understood that can also pass through other existing algorithm moulds
Type handles the first picture to extract the center position of origin label and reference marker and according to method structure provided by the present application
Build the center of circle and radius that coordinate system solves maximum inscribed circle.
Method based on the detection target object maximum inscribed circle that the embodiment of the present application one provides, present invention also provides one
System of the kind based on image processing techniques detection target object maximum inscribed circle, comprising:
Image collection module: building includes origin label and reference marker;Origin label and reference marker are arranged in work
Make on platform, target object be placed on workbench, obtain include target object, origin label, reference marker workbench just
Projected image obtains the first picture;
The origin label includes a template graphics, and the reference marker includes three templates not on the same line
Figure;
Identification module: the position of origin label and reference marker is determined on the first picture;
The identification module further includes graph transformation unit: to template graphics execute different deformation operations obtain it is different
Template picture;
Module training unit: being divided into training set and test set for template picture, uses training set training graphics processing model
And processing model is evaluated with test set;
Identifying processing unit: using four template graphics in processing the first picture of model extraction, three are then taken respectively
Three interior angles of group joint account of template graphics center position, when three interior angles with three template graphics on reference marker
When the triangle interior angle of composition is equal, using these three template graphics as reference marker, using another template graphics as origin mark
Note;
Coordinate system constructs module: constructing two-dimensional pixel coordinate system labeled as origin with origin in the first picture, extracts ginseng
Examine four boundary values of the coordinate of template graphics in label, the profile line coordinates of target object and profile line coordinates;
Processing module: the center of circle for calculating the maximum inscribed circle of target object and radius are solved.
Claims (14)
1. a kind of method based on image processing techniques detection target object maximum inscribed circle, it is characterised in that: including
Step A: building origin label and reference marker;By origin label and reference marker setting on the table, by target
Object is placed on workbench, obtain include target object, origin label, reference marker workbench orthographic view obtain the
One picture;
Step B: the position of origin label and reference marker is determined on the first picture;
Step C: two-dimensional pixel coordinate system is constructed labeled as origin with origin, extracts coordinate, the mesh of template graphics in reference marker
Mark the profile line coordinates of object and four boundary values of profile line coordinates;
Step D: the center of circle for calculating the maximum inscribed circle of target object and radius are solved.
2. a kind of method based on image processing techniques detection target object maximum inscribed circle according to claim 1,
Be characterized in that: the origin label in step A includes a circular colored spot, and reference marker includes two circular colored spots, and records ginseng
Examine the distance L of two circular colored spots in labelAB。
3. a kind of method based on image processing techniques detection target object maximum inscribed circle according to claim 2,
Be characterized in that: the method that origin label and reference marker position are determined in step B is that all and background is shown in the first picture
Inconsistent abnormal color region alternately point, the position of confirmation origin label and reference marker in alternative point.
4. a kind of method based on image processing techniques detection target object maximum inscribed circle according to claim 1,
Be characterized in that: the origin label in step A includes a template graphics, and reference marker includes three moulds not on the same line
Plate figure, and the distance between record any two of them template graphics and the line of the two template graphics and another
The angle of template graphics line.
5. a kind of method based on image processing techniques detection target object maximum inscribed circle according to claim 4,
Be characterized in that: the method for the label of determination origin described in step B and reference marker position includes:
Step I: different deformation operations is executed to template graphics and obtains different template pictures;
Step II: being divided into training set and test set for template picture, using training set training graphics processing model and uses test set
Evaluation processing model;
Step III: it using four template graphics in processing the first picture of model extraction, then takes respectively in three template graphics
Three interior angles of group joint account of heart point position, when the triangle that three interior angles are constituted with three template graphics on reference marker
When shape interior angle is equal, using these three template graphics as reference marker, marked using another template graphics as origin.
6. a kind of method based on image processing techniques detection target object maximum inscribed circle according to claim 5,
Be characterized in that: template graphics described in step A include that at least there are two the concentric circles of the different filled circles of diameter different colours for tool
Point.
7. a kind of method based on image processing techniques detection target object maximum inscribed circle according to claim 6,
It is characterized in that: including rotation, translation, scaling, overturning and shearing to the deformation operation that template graphics execute.
8. a kind of method based on image processing techniques detection target object maximum inscribed circle according to claim 7,
It is characterized in that: using keras.preprocessing.image.ImageDataGenerator pairs of picture generator in step I
Template graphics carry out deformation operation and generate the template picture.
9. a kind of method based on image processing techniques detection target object maximum inscribed circle according to claim 5,
It is characterized in that: training set and test set being generated to the json file for meeting Microsoft's COCO data set format respectively, to mesh in step B
Mark inspection software packet Detectron input training set and test set carry out deep learning and obtain the processing model.
10. a kind of method based on image processing techniques detection target object maximum inscribed circle according to claim 5,
It is characterized in that: carrying out deep learning using Detectron in step II and obtain processing model.
11. detecting target object maximum inscribed circle based on image processing techniques according to claim 2~10 is described in any item
Method, it is characterised in that: the method for contour line of the target object in the first picture is extracted in step C are as follows: turn the first picture
It is changed to grayscale image, cv::Canny () operator is called to detect edge of the target object in the first picture, calls function cv::
FindContours () searches the profile of target object, is denoted as contour, traverses in objects' contour contour
Point, lookup are worth the smallest abscissa Xmin, the smallest ordinate Y of valuemin, the maximum abscissa X of valuemaxWith the maximum ordinate of value
Ymax;
Recording distance in reference marker is LABCoordinate A (x of the two o'clock in pixel coordinate systemA,yA), B (xB,yB)。
12. a kind of method based on image processing techniques detection target object maximum inscribed circle according to claim 11,
It is characterized by: the method for calculating the center of circle and radius of the maximum inscribed circle of target object in step D are as follows: for profile
The minimum range of arbitrary point (x, y) distance profile contour in contour is denoted as min Dist (x, y, contour), then asks
The problem of taking maximum inscribed circle inside target object is converted to the problem of seeking min Dist (x, y, contour) maximum value,
I.e.
Wherein, RI indicates that the radius of maximum inscribed circle, CI are the center of circle of maximum inscribed circle;
Call function cv::pointPolygonTest (InputArray contour, Point2f pt, bool
MeasureDist), obtain
Utilize the actual distance L of two o'clock in reference markerABResult under pixel coordinate system is handled, is obtained
Wherein, RA indicates that the true value of radius, CA.x indicate the center of circle and coordinate origin along the actual distance of x-axis, and CA.y indicates circle
The actual distance of the heart and coordinate origin along y-axis, x-axis coordinate of the CI.x expression center of circle under pixel coordinate system, CI.y indicate the center of circle
Y-axis coordinate under pixel coordinate system, eu_dist (A, B) indicate the length of A, B two o'clock under pixel coordinate system.
13. a kind of system based on image processing techniques detection target object maximum inscribed circle, it is characterised in that: including
Image collection module: building includes origin label and reference marker;Origin label and reference marker are arranged in workbench
On, target object is placed on workbench, the workbench orthographic projection including target object, origin label, reference marker is obtained
Image obtains the first picture;
Identification module: the position of origin label and reference marker is determined on the first picture;
Coordinate system constructs module: constructing two-dimensional pixel coordinate system labeled as origin with origin in the first picture, extracts with reference to mark
Four boundary values of the coordinate of template graphics in note, the profile line coordinates of target object and profile line coordinates;
Processing module: the center of circle for calculating the maximum inscribed circle of target object and radius are solved.
14. a kind of system based on image processing techniques detection target object maximum inscribed circle according to claim 13,
It is characterized by: origin label includes a template graphics, the reference marker include three not on the same line
Template graphics, the identification module further include:
Graph transformation unit: different deformation operations is executed to template graphics and obtains different template pictures;
Module training unit: being divided into training set and test set for template picture, is used in combination using training set training graphics processing model
Test set evaluation processing model;
Identifying processing unit: using four template graphics in processing the first picture of model extraction, three templates are then taken respectively
Three interior angles of group joint account of centre of figure point position, when three interior angles are constituted with three template graphics on reference marker
Triangle interior angle it is equal when, using these three template graphics as reference marker, using another template graphics as origin mark.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910777824.2A CN110533714A (en) | 2019-08-21 | 2019-08-21 | Method and system based on image processing techniques detection target object maximum inscribed circle |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910777824.2A CN110533714A (en) | 2019-08-21 | 2019-08-21 | Method and system based on image processing techniques detection target object maximum inscribed circle |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110533714A true CN110533714A (en) | 2019-12-03 |
Family
ID=68663996
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910777824.2A Pending CN110533714A (en) | 2019-08-21 | 2019-08-21 | Method and system based on image processing techniques detection target object maximum inscribed circle |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110533714A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111242240A (en) * | 2020-02-13 | 2020-06-05 | 深圳市联合视觉创新科技有限公司 | Material detection method and device and terminal equipment |
CN111932491A (en) * | 2020-06-23 | 2020-11-13 | 联宝(合肥)电子科技有限公司 | Component detection method, device and storage medium |
Citations (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101305922A (en) * | 2008-05-29 | 2008-11-19 | 上海交通大学医学院附属第九人民医院 | Calibration and splice method of over-long specification X ray irradiation image |
CN101669144A (en) * | 2007-03-13 | 2010-03-10 | 浦项产业科学研究院 | Landmark for position determination of mobile robot and apparatus and method using it |
CN102376089A (en) * | 2010-12-09 | 2012-03-14 | 深圳大学 | Target correction method and system |
CN103177584A (en) * | 2013-02-05 | 2013-06-26 | 长安大学 | Vehicle speed detection method based on enumeration probe |
CN103714571A (en) * | 2013-09-23 | 2014-04-09 | 西安新拓三维光测科技有限公司 | Single camera three-dimensional reconstruction method based on photogrammetry |
CN105787939A (en) * | 2016-02-26 | 2016-07-20 | 南京理工大学 | Rapid positioning detection method for PCB circular hole |
CN105825193A (en) * | 2016-03-25 | 2016-08-03 | 乐视控股(北京)有限公司 | Method and device for position location of center of palm, gesture recognition device and intelligent terminals |
CN105894544A (en) * | 2015-12-14 | 2016-08-24 | 乐视云计算有限公司 | Method and device for detecting circle-center position in image |
CN106092090A (en) * | 2016-08-06 | 2016-11-09 | 中科院合肥技术创新工程院 | A kind of infrared road sign for indoor mobile robot location and using method thereof |
CN106526467A (en) * | 2016-10-14 | 2017-03-22 | 西安交通大学 | High voltage circuit breaker switch-on and switch-off speed characteristic measurement method based on machine vision |
CN106780623A (en) * | 2016-12-14 | 2017-05-31 | 厦门理工学院 | A kind of robotic vision system quick calibrating method |
CN107154050A (en) * | 2017-05-03 | 2017-09-12 | 魏玉震 | A kind of automatic obtaining method of the stone material geometric parameter based on machine vision |
CN107931012A (en) * | 2017-10-25 | 2018-04-20 | 浙江华睿科技有限公司 | A kind of method and dispenser system for extracting dispensing path |
CN108007388A (en) * | 2017-06-30 | 2018-05-08 | 长沙湘计海盾科技有限公司 | A kind of turntable angle high precision online measuring method based on machine vision |
CN108057946A (en) * | 2016-11-07 | 2018-05-22 | 林肯环球股份有限公司 | For calibrating the system and method for welder training machine |
US10169882B1 (en) * | 2015-09-11 | 2019-01-01 | WinguMD, Inc. | Object size detection with mobile device captured photo |
CN109993790A (en) * | 2017-12-29 | 2019-07-09 | 深圳市优必选科技有限公司 | Marker, marker forming method, positioning method and positioning device |
-
2019
- 2019-08-21 CN CN201910777824.2A patent/CN110533714A/en active Pending
Patent Citations (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101669144A (en) * | 2007-03-13 | 2010-03-10 | 浦项产业科学研究院 | Landmark for position determination of mobile robot and apparatus and method using it |
CN101305922A (en) * | 2008-05-29 | 2008-11-19 | 上海交通大学医学院附属第九人民医院 | Calibration and splice method of over-long specification X ray irradiation image |
CN102376089A (en) * | 2010-12-09 | 2012-03-14 | 深圳大学 | Target correction method and system |
CN103177584A (en) * | 2013-02-05 | 2013-06-26 | 长安大学 | Vehicle speed detection method based on enumeration probe |
CN103714571A (en) * | 2013-09-23 | 2014-04-09 | 西安新拓三维光测科技有限公司 | Single camera three-dimensional reconstruction method based on photogrammetry |
US10169882B1 (en) * | 2015-09-11 | 2019-01-01 | WinguMD, Inc. | Object size detection with mobile device captured photo |
CN105894544A (en) * | 2015-12-14 | 2016-08-24 | 乐视云计算有限公司 | Method and device for detecting circle-center position in image |
CN105787939A (en) * | 2016-02-26 | 2016-07-20 | 南京理工大学 | Rapid positioning detection method for PCB circular hole |
CN105825193A (en) * | 2016-03-25 | 2016-08-03 | 乐视控股(北京)有限公司 | Method and device for position location of center of palm, gesture recognition device and intelligent terminals |
CN106092090A (en) * | 2016-08-06 | 2016-11-09 | 中科院合肥技术创新工程院 | A kind of infrared road sign for indoor mobile robot location and using method thereof |
CN106526467A (en) * | 2016-10-14 | 2017-03-22 | 西安交通大学 | High voltage circuit breaker switch-on and switch-off speed characteristic measurement method based on machine vision |
CN108057946A (en) * | 2016-11-07 | 2018-05-22 | 林肯环球股份有限公司 | For calibrating the system and method for welder training machine |
CN106780623A (en) * | 2016-12-14 | 2017-05-31 | 厦门理工学院 | A kind of robotic vision system quick calibrating method |
CN107154050A (en) * | 2017-05-03 | 2017-09-12 | 魏玉震 | A kind of automatic obtaining method of the stone material geometric parameter based on machine vision |
CN108007388A (en) * | 2017-06-30 | 2018-05-08 | 长沙湘计海盾科技有限公司 | A kind of turntable angle high precision online measuring method based on machine vision |
CN107931012A (en) * | 2017-10-25 | 2018-04-20 | 浙江华睿科技有限公司 | A kind of method and dispenser system for extracting dispensing path |
CN109993790A (en) * | 2017-12-29 | 2019-07-09 | 深圳市优必选科技有限公司 | Marker, marker forming method, positioning method and positioning device |
Non-Patent Citations (2)
Title |
---|
RENBO XIA 等: "Robust Algorithm for Detecting the Maximum Inscribed Circle", 《2007 10TH IEEE INTERNATIONAL CONFERENCE ON COMPUTER-AIDED DESIGN AND COMPUTER GRAPHICS》 * |
刘雨桐 等: "改进卷积神经网络在遥感图像分类中的应用", 《计算机应用》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111242240A (en) * | 2020-02-13 | 2020-06-05 | 深圳市联合视觉创新科技有限公司 | Material detection method and device and terminal equipment |
CN111242240B (en) * | 2020-02-13 | 2023-04-07 | 深圳市联合视觉创新科技有限公司 | Material detection method and device and terminal equipment |
CN111932491A (en) * | 2020-06-23 | 2020-11-13 | 联宝(合肥)电子科技有限公司 | Component detection method, device and storage medium |
CN111932491B (en) * | 2020-06-23 | 2022-02-08 | 联宝(合肥)电子科技有限公司 | Component detection method, device and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Deng et al. | Automatic indoor construction process monitoring for tiles based on BIM and computer vision | |
CN109993827B (en) | Elevation view identification method for converting building drawing into three-dimensional BIM model | |
Zhao et al. | Reconstructing BIM from 2D structural drawings for existing buildings | |
Bosche et al. | Automated retrieval of 3D CAD model objects in construction range images | |
CN108917593B (en) | Intelligent measurement system and method based on element configuration of workpiece to be measured | |
CN103439348B (en) | Remote controller key defect detection method based on difference image method | |
US9367737B2 (en) | Floor plan space detection | |
HUE026478T2 (en) | Method for forming master data for inspecting protruding and recessed figure | |
CN113343976B (en) | Anti-highlight interference engineering measurement mark extraction method based on color-edge fusion feature growth | |
CN110533714A (en) | Method and system based on image processing techniques detection target object maximum inscribed circle | |
Kampel et al. | Rule based system for archaeological pottery classification | |
Li et al. | A deep learning-based indoor acceptance system for assessment on flatness and verticality quality of concrete surfaces | |
CN117474929A (en) | Tray outline dimension detection method and system based on machine vision | |
CN112561989B (en) | Recognition method for hoisting object in construction scene | |
CN117593420A (en) | Plane drawing labeling method, device, medium and equipment based on image processing | |
CN110175952B (en) | Automatic generation method and device of jade processing path based on target detection | |
US11829194B2 (en) | Method and system for deriving a digital representation of an unfolded blank and for cost estimation based upon the same | |
Wang et al. | Edge extraction by merging the 3D point cloud and 2D image data | |
CN110400252B (en) | Material yard contour line digitalization method and system | |
CN109559294B (en) | Method and device for detecting quality of circular hole of drop | |
JPH07121713A (en) | Pattern recognition method | |
CN118135179B (en) | Cascade identification method and device for cutting plate, computer equipment and storage medium | |
Li et al. | Enhancing extraction of two-dimensional engineering drawings from three-dimensional data of existing buildings | |
Nguyen et al. | A vision-based method of reverse engineering for 2D CNC machining | |
JP3679241B2 (en) | Construction drawing recognition method and recognition apparatus |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20191203 |
|
WD01 | Invention patent application deemed withdrawn after publication |