CN110490887A - It is a kind of that the quick recognition positioning method in edge is wrapped up to rectangle based on 3D vision - Google Patents
It is a kind of that the quick recognition positioning method in edge is wrapped up to rectangle based on 3D vision Download PDFInfo
- Publication number
- CN110490887A CN110490887A CN201910561835.7A CN201910561835A CN110490887A CN 110490887 A CN110490887 A CN 110490887A CN 201910561835 A CN201910561835 A CN 201910561835A CN 110490887 A CN110490887 A CN 110490887A
- Authority
- CN
- China
- Prior art keywords
- point
- rectangle
- boundary
- data
- vision
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 230000004438 eyesight Effects 0.000 title claims abstract description 30
- 238000001914 filtration Methods 0.000 claims abstract description 20
- 238000001514 detection method Methods 0.000 claims abstract description 8
- 238000013480 data collection Methods 0.000 claims description 10
- 239000013598 vector Substances 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 6
- 239000012141 concentrate Substances 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 3
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 abstract description 3
- 230000000694 effects Effects 0.000 description 6
- 238000009499 grossing Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 2
- 230000004807 localization Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000001154 acute effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010924 continuous production Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 230000009885 systemic effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- 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
- G06T2207/10012—Stereo images
-
- 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
- G06T2207/20048—Transform domain processing
- G06T2207/20061—Hough transform
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of to wrap up the quick recognition positioning method in edge to rectangle based on 3D vision.The method comprising the steps of acquisition image data, noise spot filtering, boundary point sequence, smothing filtering, angle point coarse positioning, angle point fine positioning and rectangle identification.The boundary of rectangle package can be effectively extracted from the image data of scanning by these steps, accuracy of identification is high, and is not influenced by rectangle package dimensions size, promotes precision compared to by way of the hardware resolution for improving camera, effectively reduces cost.In addition, the sort method and Corner Detection of two-way arest neighbors each other, improve vision positioning under complicated logistics environment, to stability, the accuracy of rectangle package positioning, operator's workload is reduced, the maintainability of system is improved.
Description
Technical field
The present invention relates to logistics technology more particularly to a kind of quickly identifying to rectangle package edge based on 3D vision
Localization method.
Background technique
With the rapid development of logistlcs technology, the crawl of package in entire logistics links, the de-stacking of soft hard packaging, stacking with
And the sort process of real time kinematics scene, the pinpoint demand of polymorphic type package is more and more obvious, traditional measurement side
Method is consequently also increasingly difficult to meet multi items, high-precision measurement demand.
In the prior art, rectangle package quantity is numerous, but lacks practicability and effectiveness quick and precisely identifying and positioning aspect
Solution, therefore, be unfavorable for improve rectangle package sorting efficiency.
Summary of the invention
What the invention mainly solves the technical problem of providing a kind of based on 3D vision quickly identifying rectangle package edge
Localization method solves, shortage systemic solution low to rectangle package automated sorting accuracy in the prior art, Yi Jixiao
The low problem of rate.
In order to solve the above technical problems, the technical solution adopted by the present invention is that provide it is a kind of based on 3D vision to rectangle
Wrap up the quick recognition positioning method in edge, comprising the following steps: image data is obtained, by image scanning apparatus to the square of stacking
Shape package carries out image scanning, obtains the 3-dimensional image data set of rectangle package；Noise spot filtering, extracts 3-dimensional image data set
Original boundaries point data collection, noise spot is filtered, obtain rectangle package endpoint data collection；Boundary point sequence, choosing
Any for taking the endpoint data to concentrate is the first starting point, and it is neighbouring for calculating to the nearest point of the first start distance
Point, then by the neighbor point be the second starting point calculates other to second starting point distance most other than the first starting point
Close point is as third starting point, and so on, sequence is numbered in all boundary points concentrated to the endpoint data；It is flat
Sliding filtering carries out Neighborhood Filtering to all boundary points that the endpoint data is concentrated according to number order；Angle point coarse positioning,
According to sequence and smoothed out boundary dot sequency given threshold interval, clockwise and counterclockwise 2 are found out respectively
Boundary point forms vector angle by judging vector angle size and filters out angle point coarse positioning position；Angle point fine positioning utilizes
Endpoint data collection after sequence is divided into N cross-talk data set by the coordinate of N number of inflection point in the angle point coarse positioning, for
Each cross-talk data set converts fitting space line using hough, then seeks the intersection point of two adjacent space straight lines respectively, do
For pinpoint inflection point；Rectangle identification, after obtaining all inflection points of endpoint data, the constraint condition identified using rectangle,
Identify that rectangle wraps up corresponding square boundary from the endpoint data.
In wrapping up in another embodiment of the quick recognition positioning method in edge rectangle the present invention is based on 3D vision, described
In noise spot filtration step, comprising: firstly, being concentrated in original boundaries point data, seek and current original boundaries point P (x, y, z)
Distance less than threshold value T point number N, as the energy value of current original boundaries point P；Then, according to above-mentioned processing mode,
Corresponding energy value is calculated to each original boundaries point, obtains the Energy distribution of all original boundaries points；Then, all originals are sought
The average energy value of beginning boundary point, 2/3 point using energy value lower than described the average energy value is as noise spot from original boundaries
Point data, which is concentrated, to be removed.
In wrapping up in another embodiment of the quick recognition positioning method in edge rectangle the present invention is based on 3D vision, described
In boundary point sequence step, when the neighbor point for adjusting the distance nearest is chosen, also set distance threshold value, neighbor distance are necessarily less than
The distance threshold value, when the point that neighbor distance is greater than or equal to the distance threshold value cannot function as neighbor point.
In wrapping up in another embodiment of the quick recognition positioning method in edge rectangle the present invention is based on 3D vision, described
Further include that backward searches neighbor point in boundary point sequence step, is not able to satisfy point when boundary point occurs in endpoint data concentration
The distance between be less than distance threshold value when, starting backward search, by above-mentioned boundary point as a new starting point, from
Orderly point after sequence, which concentrates detection range recently and meets the point that above-mentioned distance threshold value requires, is added orderly point set, then after
The continuous process for executing positive sequence and searching.
In wrapping up in another embodiment of the quick recognition positioning method in edge rectangle the present invention is based on 3D vision, described
In angle point fine positioning step, using the distance and straight line of coordinate origin to straight line and the angle of x-axis can uniquely indicate one it is straight
Line is expressed as follows:
ρ=xcos θ+ysin θ
In above formula, (x, y) is the coordinate of any on straight line, and ρ is distance of the coordinate origin to straight line, and θ is straight line and x-axis
Angle.
In wrapping up in another embodiment of the quick recognition positioning method in edge rectangle the present invention is based on 3D vision, described
In angle point fine positioning step, using the data boundary point set after boundary point sequence as the input of hough transformation data, divide
Not Jie Qu N cross-talk data be centrally located at the data of intermediate 2/3 length and inputted as the data of fitting space line.
In wrapping up in another embodiment of the quick recognition positioning method in edge rectangle the present invention is based on 3D vision, rectangle is known
Other constraint condition includes: (1) continuous four right angles of appearance and corner direction counterclockwise is consistent, is confirmed as rectangle；(2) occur
Continuous three right angles and corner direction counterclockwise is consistent, are confirmed as rectangle；(3) there is continuous two right angles and turning counterclockwise
Direction is consistent, is confirmed as potential rectangle；(4) there is a continuous right angle, rectangle cannot be confirmed as.
The beneficial effects of the present invention are: wrapping up the quick recognition positioning method packet in edge to rectangle the present invention is based on 3D vision
Including step has acquisition image data, noise spot filtering, boundary point sequence, smothing filtering, angle point coarse positioning, angle point fine positioning and square
Shape identification.The boundary of rectangle package can be effectively extracted from the image data of scanning by these steps, accuracy of identification is high,
And it is not influenced by rectangle package dimensions size, promotes precision compared to by way of the hardware resolution for improving camera, have
Effect reduces costs.In addition, the sort method and Corner Detection of two-way arest neighbors each other, improve vision positioning in complicated logistics
Under environment, to stability, the accuracy of rectangle package positioning, operator's workload is reduced, the maintainability of system is improved.
Detailed description of the invention
Fig. 1 be according to the present invention based on 3D vision to rectangle package quick one embodiment of recognition positioning method in edge stream
Cheng Tu；
Fig. 2 be according to the present invention based on 3D vision to rectangle package another embodiment of the quick recognition positioning method in edge in
Noise spot filter out after effect picture；
Fig. 3 be according to the present invention based on 3D vision to rectangle package another embodiment of the quick recognition positioning method in edge in
Boundary point sequence after effect picture；
Fig. 4 is the partial enlarged view of Fig. 3；
Fig. 5 be according to the present invention based on 3D vision to rectangle package another embodiment of the quick recognition positioning method in edge in
The straight effect picture of fitting；
Fig. 6 be according to the present invention based on 3D vision to rectangle package another embodiment of the quick recognition positioning method in edge in
The effect picture for identifying rectangle.
Specific embodiment
To facilitate the understanding of the present invention, in the following with reference to the drawings and specific embodiments, the present invention will be described in more detail.
A better embodiment of the invention is given in the attached drawing.But the invention can be realized in many different forms, and unlimited
In this specification described embodiment.On the contrary, purpose of providing these embodiments is makes to the disclosure
Understand more thorough and comprehensive.
It should be noted that unless otherwise defined, all technical and scientific terms used in this specification with belong to
The normally understood meaning of those skilled in the art of the invention is identical.Used term in the description of the invention
It is the purpose in order to describe specific embodiment, is not intended to the limitation present invention.Term "and/or" packet used in this specification
Include any and all combinations of one or more related listed items.
Fig. 1 shows the stream to rectangle package quick one embodiment of recognition positioning method in edge the present invention is based on 3D vision
Cheng Tu.In Fig. 1, comprising:
Step S101: obtaining image data, carries out image scanning by rectangle package of the image scanning apparatus to stacking, obtains
Obtain the 3-dimensional image data set of rectangle package；
Step S102: noise spot filtering extracts the original boundaries point data collection of the 3-dimensional image data set, to noise spot
It is filtered, obtains the endpoint data collection of rectangle package；
Step S103: boundary point sequence chooses any that the endpoint data is concentrated as the first starting point, calculates and arrive institute
State the nearest point of the first start distance be neighbor point, then by the neighbor point be the second starting point, other than the first starting point,
It calculates other and arrives second starting point apart from nearest point as third starting point, and so on, to the endpoint data
Sequence is numbered in all boundary points concentrated；
Step S104: smothing filtering carries out all boundary points that the endpoint data is concentrated adjacent according to number order
Domain filtering；
Step S105: angle point coarse positioning is found out respectively according to sequence and smoothed out boundary dot sequency given threshold interval
2 clockwise and counterclockwise boundary points form vector angle, by judging vector angle size, screen angle of departure
Point coarse positioning position；
Step S106: angle point fine positioning, using the coordinate of N number of inflection point in the angle point coarse positioning, by the side after sequence
Boundary's point data collection is divided into N cross-talk data set, for each cross-talk data set, converts fitting space line using hough, then
The intersection point for seeking two adjacent space straight lines respectively, as pinpoint inflection point；
Step S107: rectangle identification, after obtaining all inflection points of endpoint data, the constraint condition identified using rectangle,
Identify that rectangle wraps up corresponding square boundary from the endpoint data.
It preferably, in step s 102, include: to include: firstly, in original boundaries point to the method that noise spot is filtered
In data set, the number N of the point with the distance of current original boundaries point P (x, y, z) less than threshold value T is sought, as current original
The energy value of boundary point P；Then, according to above-mentioned processing mode, corresponding energy value is calculated to each original boundaries point, obtains institute
There is the Energy distribution of original boundaries point；Then, the average energy value for seeking all original boundaries points is flat lower than described by energy value
2/3 point of equal energy value, which is concentrated as noise spot from original boundaries point data, to be removed.As shown in Figure 2, it is shown that by interference
After point filtering, the endpoint data collection JS1 of obtained multiple rectangles package.
In step s 103, first starting point is A (x1, y1, z1), with the first starting point A (x1, y1, z1) away from
It is neighbor point B (x from nearest point2, y2, z2), neighbor distance between the two are as follows:
Preferably, when the neighbor point for adjusting the distance nearest is chosen, also set distance threshold value, neighbor distance are necessarily less than the range gate
Limit value, when the point that neighbor distance is greater than or equal to the distance threshold value cannot function as neighbor point.May occur some sides as a result,
Boundary's point is unable to number sorting, thus in step s 103 further preferably include backward search consecutive points, be exactly to work as endpoint data
There is boundary point and is not able to satisfy the distance between point less than after distance threshold value in collection, and starting backward is searched, by above-mentioned boundary point
As a new starting point, the orderly point from after sorted, which concentrates detection range recently and meets above-mentioned distance threshold value, to be wanted
Orderly point set is added in the point asked, and then proceedes to execute the process that positive sequence is searched.This mode can guarantee all points all in accordance with
It is ranked up apart from least way.As shown in Fig. 3, including the number sorting to each boundary point.Fig. 4 is shown further
The partial enlarged view of Fig. 3.
Preferably, in step S104, for a boundary point (x, y, z), N number of boundary adjacent thereto after sequence is utilized
Point carries out Gaussian smoothing filter processing to P.
It can be with further progress angle point coarse positioning after above-mentioned steps, it is therefore an objective to for finding single or multiple rectangle packets
The corner position of cabinet in complex situations is wrapped up in, but need first to filter the miscellaneous point of original boundaries point data progress of acquisition before,
Sequence, smoothing processing.The meaning of miscellaneous point filtering is the influence that exclusive PCR point sorts to boundary point, since smoothing process is base
It is carried out in the endpoint data in local neighborhood, thus endpoint data needs first before carrying out Gaussian smoothing to boundary
Point data is ranked up.Gaussian smoothing be in order to reduce endpoint data in angle point coarse positioning inflection point, because of adjacent edge
Biggish data fluctuations between boundary's point cause the angle point (inflection point) of mistake to identify.
The effect of the angle point coarse positioning of step S105 straight line line as shown in Figure 5.This rough positioning method, precision
It is too poor, the standard that scene uses is not achieved.Therefore we need to be accurately positioned corner location, are converted in this method by hough
Straight line fitting is accurately positioned corner location.
For this purpose, in step s 106, it is the duality using point with line using the basic principle of Hough transform, it will be former
The given curve negotiating curve representation form of beginning image space becomes a point of parameter space.Thus in original image
The test problems of given curve are converted into the spike problem found in parameter space, namely detection overall permanence is converted into detection
Local characteristics.
Straight line can uniquely be indicated using the distance and straight line of coordinate origin to straight line and the angle of x-axis, formula is such as
Under:
ρ=xcos θ+ysin θ
In above formula, (x, y) is the coordinate of any on straight line, and ρ is distance of the coordinate origin to straight line, and θ is straight line and x-axis
Angle.
Preferably, in step s 106, further include two aspect: first is that using sequence after data boundary point set as
Hough converts the input of data, the reason is that the data after one side smoothing processing lose data precision, it is on the other hand miscellaneous point
Data before filtering are easy to be interfered；Second is that interception N cross-talk data are centrally located at the data of intermediate 2/3 length as quasi- respectively
The data input for closing space line, because the precision of coarse positioning is not enough to accurately navigate to corner position, close to inflection point
The data at place cannot be inputted as the data of fitting a straight line.Fig. 6 shows the space line figure using hough transformation fitting.
Preferably, in step s 107, the constraint condition of rectangle identification includes:
(1) there are continuous four right angles and corner direction counterclockwise is consistent, be confirmed as rectangle；
(2) there are continuous three right angles and corner direction counterclockwise is consistent, be confirmed as rectangle；
(3) there are continuous two right angles and corner direction counterclockwise is consistent, be confirmed as potential rectangle；
(4) there is a continuous right angle, rectangle can not be confirmed as.
Preferably, under aforementioned four constraint condition, the rectangle that data boundary point is concentrated, potential rectangle can successively be identified
And exclude non-rectangle corner point.Wherein judge at inflection point whether to be right angle, the angle by solving two space vectors solves；
Counterclockwise judge whether the steering between adjacent comers is consistent, the multiplication cross vector of two vectors of inflection point constituted by solving,
Whether judge with whether the direction vector of data boundary is the mode of acute angle.Fig. 6 shows the package boundary finally identified
Corresponding rectangle.
It can be seen that it to include step obtains to the rectangle package quick recognition positioning method in edge that the present invention is based on 3D visions
Take image data, noise spot filtering, boundary point sequence, smothing filtering, angle point coarse positioning, angle point fine positioning and rectangle identification.It is logical
The boundary of rectangle package can effectively be extracted from the image data of scanning by crossing these steps, and accuracy of identification is high, and not by
The influence of rectangle package dimensions size promotes precision compared to by way of the hardware resolution for improving camera, effectively reduces
Cost.In addition, the sort method and Corner Detection of two-way arest neighbors each other, improve vision positioning under complicated logistics environment,
To stability, the accuracy of rectangle package positioning, operator's workload is reduced, the maintainability of system is improved.
The above description is only an embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure transformation made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant technical fields,
Similarly it is included within the scope of the present invention.
Claims (7)
1. a kind of wrap up the quick recognition positioning method in edge to rectangle based on 3D vision, which comprises the following steps:
Image data is obtained, image scanning is carried out by rectangle package of the image scanning apparatus to stacking, obtains rectangle package
3-dimensional image data set；
Noise spot filtering, extracts the original boundaries point data collection of the 3-dimensional image data set, is filtered, obtains to noise spot
The endpoint data collection of rectangle package；
Boundary point sequence chooses any that the endpoint data is concentrated as the first starting point, calculates and arrive first starting point
It is neighbor point apart from nearest point, then the neighbor point is calculated other to institute other than the first starting point for the second starting point
The second starting point is stated apart from nearest point as third starting point, and so on, all sides that the endpoint data is concentrated
Sequence is numbered in boundary's point；
Smothing filtering carries out Neighborhood Filtering to all boundary points that the endpoint data is concentrated according to number order；
Angle point coarse positioning, according to sequence and smoothed out boundary dot sequency given threshold interval, find out respectively clockwise and
Anticlockwise 2 boundary points form vector angle by judging vector angle size and filter out angle point coarse positioning position；
Angle point fine positioning is divided the endpoint data collection after sequence using the coordinate of N number of inflection point in the angle point coarse positioning
At N cross-talk data set, for each cross-talk data set, fitting space line is converted using hough, it is adjacent then to seek two respectively
The intersection point of space line, as pinpoint inflection point；
Rectangle identification, after obtaining all inflection points of endpoint data, the constraint condition identified using rectangle is counted from the boundary
Identify that rectangle wraps up corresponding square boundary in.
2. according to claim 1 wrap up the quick recognition positioning method in edge to rectangle based on 3D vision, feature exists
In in the noise spot filtration step, comprising: firstly, being concentrated in original boundaries point data, seek and current original boundaries point
The number N of point of the distance of P (x, y, z) less than threshold value T, as the energy value of current original boundaries point P；Then, according to above-mentioned
Processing mode calculates corresponding energy value to each original boundaries point, obtains the Energy distribution of all original boundaries points；Then,
The average energy value for seeking all original boundaries points, 2/3 point using energy value lower than described the average energy value is as noise spot
It concentrates and removes from original boundaries point data.
3. according to claim 1 wrap up the quick recognition positioning method in edge to rectangle based on 3D vision, feature exists
In, in the boundary point sequence step, adjust the distance nearest neighbor point choose when, also set distance threshold value, it is adjacent away from
From being necessarily less than the distance threshold value, when the point that neighbor distance is greater than or equal to the distance threshold value cannot function as it is neighbouring
Point.
4. according to claim 3 wrap up the quick recognition positioning method in edge to rectangle based on 3D vision, feature exists
In, in the boundary point sequence step, further include backward search neighbor point, when endpoint data concentration there is boundary point cannot
When meeting the distance between point less than distance threshold value, starting backward is searched, by above-mentioned boundary point as a new starting
Point, the orderly point from after sorted, which concentrates detection range recently and meets the point that above-mentioned distance threshold value requires, is added orderly point
Collection then proceedes to execute the process that positive sequence is searched.
5. according to claim 1 wrap up the quick recognition positioning method in edge to rectangle based on 3D vision, feature exists
In using coordinate origin can be with only table to the distance and straight line of straight line and the angle of x-axis in the angle point fine positioning step
Show straight line, be expressed as follows:
ρ=xcos θ+ysin θ
In above formula, (x, y) is the coordinate of any on straight line, and ρ is distance of the coordinate origin to straight line, and θ is the folder of straight line and x-axis
Angle.
6. according to claim 5 wrap up the quick recognition positioning method in edge to rectangle based on 3D vision, feature exists
In in the angle point fine positioning step, using the data boundary point set after boundary point sequence as hough transformation data
Input, intercept N cross-talk data respectively and be centrally located at the data of intermediate 2/3 length and inputted as the data of fitting space line.
7. according to claim 1 wrap up the quick recognition positioning method in edge to rectangle based on 3D vision, feature exists
In the constraint condition of rectangle identification includes:
(1) there are continuous four right angles and corner direction counterclockwise is consistent, be confirmed as rectangle；
(2) there are continuous three right angles and corner direction counterclockwise is consistent, be confirmed as rectangle；
(3) there are continuous two right angles and corner direction counterclockwise is consistent, be confirmed as potential rectangle；
(4) there is a continuous right angle, rectangle cannot be confirmed as.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910561835.7A CN110490887B (en) | 2019-06-26 | 2019-06-26 | 3D vision-based method for quickly identifying and positioning edges of rectangular packages |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910561835.7A CN110490887B (en) | 2019-06-26 | 2019-06-26 | 3D vision-based method for quickly identifying and positioning edges of rectangular packages |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110490887A true CN110490887A (en) | 2019-11-22 |
| CN110490887B CN110490887B (en) | 2023-02-28 |
Family
ID=68546347
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910561835.7A Active CN110490887B (en) | 2019-06-26 | 2019-06-26 | 3D vision-based method for quickly identifying and positioning edges of rectangular packages |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110490887B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113267143A (en) * | 2021-06-30 | 2021-08-17 | 三一建筑机器人(西安)研究院有限公司 | Side die identification method |
| WO2022161172A1 (en) * | 2021-01-29 | 2022-08-04 | 深圳光峰科技股份有限公司 | Method and apparatus for identifying corner points of pattern in image, and medium and electronic device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5616905A (en) * | 1994-02-24 | 1997-04-01 | Kabushiki Kaisha Tec | Two-dimensional code recognition method |
| JPH10275213A (en) * | 1997-03-31 | 1998-10-13 | Fuji Photo Film Co Ltd | Radiograph picture irradiating field recognizing method, device therefor, blackening processing method, and device therefor |
| CN104952072A (en) * | 2015-06-16 | 2015-09-30 | 华中科技大学 | Rectangle detection method based on genetic algorithm |
| CN107516098A (en) * | 2017-07-30 | 2017-12-26 | 华南理工大学 | A kind of objective contour 3-D information fetching method based on edge angle |
| WO2018107939A1 (en) * | 2016-12-14 | 2018-06-21 | 国家海洋局第二海洋研究所 | Edge completeness-based optimal identification method for image segmentation |
| CN109359652A (en) * | 2018-11-23 | 2019-02-19 | 浙江理工大学 | A method of the fast automatic extraction rectangular scanning part from digital photograph |
-
2019
- 2019-06-26 CN CN201910561835.7A patent/CN110490887B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5616905A (en) * | 1994-02-24 | 1997-04-01 | Kabushiki Kaisha Tec | Two-dimensional code recognition method |
| JPH10275213A (en) * | 1997-03-31 | 1998-10-13 | Fuji Photo Film Co Ltd | Radiograph picture irradiating field recognizing method, device therefor, blackening processing method, and device therefor |
| CN104952072A (en) * | 2015-06-16 | 2015-09-30 | 华中科技大学 | Rectangle detection method based on genetic algorithm |
| WO2018107939A1 (en) * | 2016-12-14 | 2018-06-21 | 国家海洋局第二海洋研究所 | Edge completeness-based optimal identification method for image segmentation |
| CN107516098A (en) * | 2017-07-30 | 2017-12-26 | 华南理工大学 | A kind of objective contour 3-D information fetching method based on edge angle |
| CN109359652A (en) * | 2018-11-23 | 2019-02-19 | 浙江理工大学 | A method of the fast automatic extraction rectangular scanning part from digital photograph |
Non-Patent Citations (5)
| Title |
|---|
| K. RAO等: "《A computer vision system to detect 3-D rectangular solids》", 《PROCEEDINGS THIRD IEEE WORKSHOP ON APPLICATIONS OF COMPUTER VISION. WACV"96》 * |
| 彭泽林: "《机器人堆叠目标识别与定位抓取系统研究》", 《中国优秀硕士学位论文全文数据库》 * |
| 范鹤鹤: "一种基于谱聚类和遗传算法的矩形检测方法", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 赵小阳等: "基于LiDAR点云的建筑物边界提取及规则化", 《地理空间信息》 * |
| 赵建强: "《适用于表面污染检测的机械臂轨迹生成研究》", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022161172A1 (en) * | 2021-01-29 | 2022-08-04 | 深圳光峰科技股份有限公司 | Method and apparatus for identifying corner points of pattern in image, and medium and electronic device |
| CN113267143A (en) * | 2021-06-30 | 2021-08-17 | 三一建筑机器人(西安)研究院有限公司 | Side die identification method |
| CN113267143B (en) * | 2021-06-30 | 2023-08-29 | 三一建筑机器人(西安)研究院有限公司 | Side mold identification method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110490887B (en) | 2023-02-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106485275B (en) | A method of realizing that cover-plate glass is bonded with liquid crystal display positioning | |
| CN103426186B (en) | A kind of SURF fast matching method of improvement | |
| CN102722731A (en) | Efficient image matching method based on improved scale invariant feature transform (SIFT) algorithm | |
| CA2535828C (en) | Computer-vision system for classification and spatial localization of bounded 3d-objects | |
| CN104680516B (en) | A kind of acquisition methods of image quality features set of matches | |
| CN104933434A (en) | Image matching method combining length between perpendiculars (LBP) feature extraction method and surf feature extraction method | |
| CN103310453A (en) | Rapid image registration method based on sub-image corner features | |
| CN110490887A (en) | It is a kind of that the quick recognition positioning method in edge is wrapped up to rectangle based on 3D vision | |
| CN105809651A (en) | Image saliency detection method based on edge non-similarity comparison | |
| Ückermann et al. | Realtime 3D segmentation for human-robot interaction | |
| CN103768792B (en) | Partitioning and recognition method applied to wide-range video space object positioning system | |
| CN102446356A (en) | Parallel and adaptive matching method for acquiring remote sensing images with homogeneously-distributed matched points | |
| US7006694B1 (en) | System and method for pattern identification | |
| Wang et al. | Grasping objects mixed with towels | |
| CN107423738A (en) | A kind of paper topic localization method and device based on template matches | |
| CN106934832B (en) | A kind of simple straight line automatic positioning method towards vision line walking | |
| CN104036494B (en) | A kind of rapid matching computation method for fruit image | |
| CN106023096B (en) | Eliminate the image matching method that edge sawtoothization influences | |
| CN103886314A (en) | Two-level matching method based on SIFT feature scale component in object recognition | |
| CN108961285B (en) | Method and device for extracting welding seam edge of container hinge | |
| CN103700119A (en) | Local texture description method based on local grouping comparison mode column diagram | |
| CN106204889A (en) | The localization method of a kind of bank note fluorescence area and device | |
| CN103729850B (en) | Method for linear extraction in panorama | |
| CN102663395A (en) | A straight line detection method based on self-adaptation multi-scale fast discrete Beamlet transform | |
| CN105677669A (en) | Sorting method and device for panoramic splicing images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |