CN103604388B - The quick accurate positioning method of special marking thing in large-scale steel plate three-dimensional measurement - Google Patents
The quick accurate positioning method of special marking thing in large-scale steel plate three-dimensional measurement Download PDFInfo
- Publication number
- CN103604388B CN103604388B CN201310551868.6A CN201310551868A CN103604388B CN 103604388 B CN103604388 B CN 103604388B CN 201310551868 A CN201310551868 A CN 201310551868A CN 103604388 B CN103604388 B CN 103604388B
- Authority
- CN
- China
- Prior art keywords
- laser tag
- steel plate
- laser
- tag
- 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.)
- Expired - Fee Related
Links
Landscapes
- Length Measuring Devices By Optical Means (AREA)
Abstract
The quick accurate positioning method of special marking thing in a kind of large-scale steel plate three-dimensional measurement of the present invention, first in heavy steel plate, laser tag is marked, and laser tag and non-laser tag object are repeatedly taken pictures, choose two groups of photograph collections-laser tag image and imperfect laser tag image, positive and negative training sample set is trained; Then the heavy steel plate containing laser tag taken pictures and reach server, on the server laser tag being detected, realizing the coarse positioning of laser tag; Finally further precise positioning is carried out to laser tag, obtain locator data to be supplied to the Parametric registration of camera measuring system and laser measurement system.Method of the present invention can carry out fast accurately location to the laser tag on steel plate automatically, in time, conveniently, accurately.
Description
Technical field
The invention belongs to shipbuilding field of measuring technique, relate to the quick accurate positioning method of special marking thing in a kind of large-scale steel plate three-dimensional measurement.
Background technology
In the measuring system in conjunction with laser and vision, need to place specific markers on testee, first measure the key signature point of object with camera; By laser, same gauge point is measured again, and then set up the registration relation of two kinds of measuring systems, solve registration parameter; In follow-up measuring process, directly utilize camera to measure, and revise camera measurement result with registration parameter, eliminate the problems such as the distortion occurred in measuring, thus improve precision and the speed of measuring system simultaneously, fully merge the advantage of two kinds of measuring techniques.But method spininess conventional at present, for the accurate location of specific objective, stably cannot work, and not be suitable for the large scale steel plate measuring system in shipbuilding process under the external environment condition of complexity.In order to not by the impact of the extraneous factors such as natural lighting, camera angles and time shutter, a kind of laser tag placement technology that is accurate, that can be applicable to complex environment and industrial requirements is fast proposed most important.
Summary of the invention
The object of this invention is to provide the quick accurate positioning method of special marking thing in a kind of large-scale steel plate three-dimensional measurement.
For achieving the above object, technical scheme of the present invention is:
The quick accurate positioning method of special marking thing in a kind of large-scale steel plate three-dimensional measurement, the realization of method based on condition comprise: industrial camera, special marking thing and GPU server, wherein special marking thing is the laser tag of 2cm*2cm, the resolution of industrial camera is more than 1440*1080, frame per second is more than 10fps, links the GPU server received and have storage and analytic function via 1394 lines and 1394.
The quick accurate positioning method of special marking thing in a kind of large-scale steel plate three-dimensional measurement of the present invention, comprises following concrete steps:
A. in heavy steel plate, mark special marking thing (such as laser tag);
B. repeatedly taken pictures by the object of industrial camera to laser tag and unmarked laser tag;
C. the method by manually choosing chooses two groups of photograph collection I and II respectively, wherein be laser tag image in photograph collection I, the imperfect laser label image image of the unmarked laser tag (or be only laser tag subregion image) is, using photograph collection I and II as the positive and negative training sample set in this method machine learning stage in photograph collection II;
D. the training of negative sample collection is aligned with the HOG algorithm based on GPU computing;
E. industrial camera is adopted to take pictures to the heavy steel plate containing laser tag and reach GPU server;
F. on GPU server, adopt the HOG algorithm based on GPU to detect the laser tag in steel plate photo, testing result adopts rectangle frame to identify in the picture, realizes the coarse positioning of laser tag;
G. adopt the binarization method based on DOG to carry out further precise positioning to laser tag, obtain locator data to be supplied to the Parametric registration of camera measuring system and laser measurement system.
Accompanying drawing explanation
Fig. 1: the quick accurate positioning method implementation procedure figure of special marking thing of the present invention;
Fig. 2: each hardware device connection layout in the present invention;
Fig. 3: HOG algorithm flow chart in the present invention;
Fig. 4: the laser tag binarization figure that the present invention is based on DOG, wherein: (a) is laser tag, (b) laser tag image I grey level histogram, c DOG extreme point that () is image I, d () is preliminary binary image Ibin, e image Ibin2 that () obtains for carrying out expansive working to Ibin, f () is the centralized positioning image to laser tag, (g) is the partial enlarged drawing to (f) figure.
Embodiment
As shown in Figure 2, be hardware device connection layout in the present invention, be made up of High performance industrial camera and a high-performance GPU server and special marking thing.Wherein special marking thing is the laser tag of 2cm*2cm, and the resolution of industrial camera is more than 1440*1080, frame per second is more than 10fps, links the GPU server received and have storage and analytic function via 1394 lines and 1394.
Shown in Fig. 1, be the quick accurate positioning method implementation procedure figure of special marking thing in large-scale steel plate three-dimensional measurement of the present invention;
In the inventive method, on large scale steel plate, the detection (coarse positioning) of laser tag adopts the quick HOG algorithm based on GPU, and the process flow diagram of this algorithm as shown in Figure 3.
In this algorithm, gradient calculation method is: for image I, if certain pixel position is (x, y), its gray scale size is designated as I (x, y), then the gradient magnitude of this pixel and direction respectively as shown in formula (1) and (2).
After calculating the gradient of each pixel, build the histogram of gradients based on pixel at cell (connected region less in image), and by these set of histograms morphogenesis characters descriptor altogether.Finally svm classifier identification is carried out to the HOG descriptor of each block, to detect required characteristic information.
The flow process of the accurate location (carefully locating) of the laser tag in the inventive method on large scale steel plate as shown in Figure 4; Concrete steps are:
Step1. DOG computing is carried out to the gray level image I of such as Fig. 4 (a), obtain matrix I
dog.The template of twice Gaussian smoothing is n1 × n1 and n2 × n2.In this method, the value of n1 and n2 is respectively 3 and 15;
Step2. to I
dogin 5 × 5 scopes, find extreme point, if the pixel value of this extreme point is more than or equal to a certain threshold value Th
ex(Th in this method
exget 4), then put into set E.Extreme's distribution in image I is as shown in Fig. 4 (c);
Step3. to each pixel P in set E, calculate its in image I around the average of all pixels of 5 × 5 windows.If the difference of the pixel value of this average and P exceedes certain threshold value, then think that P is a key point, average is put into set A vg.Finally add up the average Th of set A vg, think that Th is exactly the threshold value required for binary image;
Step4. with Th, binaryzation is carried out to image, obtain image I
bin, as shown in Fig. 4 (d);
Step5. in order to the impact of the horizontal and vertical black line of label in rejection image I and other character symbols, to I
bincarry out expansive working and obtain image I
bin2, as shown in Fig. 4 (e).Because final utilization barycenter carrys out calculating central position, so the impact that the edge that expansive working brings expands can be ignored;
Step6. to I
bin2calculate the center point coordinate of each connected region, after the connected region that removal area is too small (noise information), average to all central points, then this HCCI combustion is the accurate locator value of laser tag, as shown in the red point in Fig. 4 (f).Fig. 4 (g) is the partial enlarged drawing of (f) figure.
The accurate locator data of laser tag can be obtained by above method, thus be supplied to the Parametric registration of laser measurement system and camera measuring system.
Claims (3)
1. the quick accurate positioning method of special marking thing in large-scale steel plate three-dimensional measurement, the realization of the method needs industrial camera, laser tag and a high-performance GPU server that can carry out storing and analyzing, and it is characterized in that the method comprises the following steps:
A. in heavy steel plate, laser tag is marked;
B. repeatedly taken pictures by the object of industrial camera to mark laser tag and unmarked laser tag;
C. the method by manually choosing chooses two groups of photograph collection I and II respectively, wherein be laser tag image in photograph collection I, imperfect laser tag image is, using photograph collection I and II as the positive and negative training sample set in this method machine learning stage in photograph collection II;
D. with the HOG algorithm based on GPU computing, positive and negative training sample set is trained;
E. in formal measurement, adopt industrial camera take pictures to the heavy steel plate containing laser tag and reach GPU server;
F. on GPU server, adopt the HOG algorithm based on GPU to detect the laser tag in steel plate photo, testing result adopts rectangle frame to identify in the picture, realizes the coarse positioning of laser tag;
G. adopt the binarization method based on DOG to carry out further precise positioning to laser tag, obtain locator data to be supplied to the Parametric registration of camera measuring system and laser measurement system.
2. the quick accurate positioning method of special marking thing in large-scale steel plate three-dimensional measurement according to claim 1, is characterized in that the size of described laser tag is 2cm*2cm.
3. the quick accurate positioning method of special marking thing in large-scale steel plate three-dimensional measurement according to claim 1, is characterized in that described imperfect laser tag image is unmarked laser tag image or is only laser tag subregion image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310551868.6A CN103604388B (en) | 2013-11-08 | 2013-11-08 | The quick accurate positioning method of special marking thing in large-scale steel plate three-dimensional measurement |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310551868.6A CN103604388B (en) | 2013-11-08 | 2013-11-08 | The quick accurate positioning method of special marking thing in large-scale steel plate three-dimensional measurement |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103604388A CN103604388A (en) | 2014-02-26 |
CN103604388B true CN103604388B (en) | 2016-03-02 |
Family
ID=50122634
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310551868.6A Expired - Fee Related CN103604388B (en) | 2013-11-08 | 2013-11-08 | The quick accurate positioning method of special marking thing in large-scale steel plate three-dimensional measurement |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103604388B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104390591B (en) * | 2014-11-27 | 2017-04-12 | 上海江南长兴造船有限责任公司 | Accurate positioning method for circular marker in large-sized curved plate measurement |
CN105444692B (en) * | 2015-11-18 | 2018-03-09 | 广东工业大学 | A kind of flame forming plate curved surface extracting method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN201572793U (en) * | 2009-12-01 | 2010-09-08 | 武汉钢铁(集团)公司 | Overproof cold-rolled sheet thickness automatic tracking and labeling apparatus |
CN101894324A (en) * | 2010-07-13 | 2010-11-24 | 大连斯博瑞科技有限公司 | System for controlling in and out of storage of steel plates |
CN202046471U (en) * | 2011-02-28 | 2011-11-23 | 宝山钢铁股份有限公司 | RFID (Radio Frequency Identification) labeling device for steel plate product |
CN102446355A (en) * | 2011-09-05 | 2012-05-09 | 西安理工大学 | Method for detecting target protruding from plane based on double viewing fields without calibration |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7388488B2 (en) * | 2003-10-30 | 2008-06-17 | Peter Lupoli | Method and system for storing, retrieving, and managing data for tags |
-
2013
- 2013-11-08 CN CN201310551868.6A patent/CN103604388B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN201572793U (en) * | 2009-12-01 | 2010-09-08 | 武汉钢铁(集团)公司 | Overproof cold-rolled sheet thickness automatic tracking and labeling apparatus |
CN101894324A (en) * | 2010-07-13 | 2010-11-24 | 大连斯博瑞科技有限公司 | System for controlling in and out of storage of steel plates |
CN202046471U (en) * | 2011-02-28 | 2011-11-23 | 宝山钢铁股份有限公司 | RFID (Radio Frequency Identification) labeling device for steel plate product |
CN102446355A (en) * | 2011-09-05 | 2012-05-09 | 西安理工大学 | Method for detecting target protruding from plane based on double viewing fields without calibration |
Non-Patent Citations (1)
Title |
---|
基于多目视觉的变形钢板三维测量方法;史金龙等;《计算机测量与控制》;20110925;第19卷(第9期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN103604388A (en) | 2014-02-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111340797B (en) | Laser radar and binocular camera data fusion detection method and system | |
CN103020632B (en) | The method for quickly identifying of localization for Mobile Robot monumented point in a kind of indoor environment | |
CN103439348B (en) | Remote controller key defect detection method based on difference image method | |
Siriborvornratanakul | An automatic road distress visual inspection system using an onboard in-car camera | |
CN108388641B (en) | Traffic facility map generation method and system based on deep learning | |
CN103324913B (en) | A kind of pedestrian event detection method of Shape-based interpolation characteristic sum trajectory analysis | |
CN103530590B (en) | DPM Quick Response Code recognition system | |
CN102914545B (en) | Gear defect detection method and system based on computer vision | |
CN105913093A (en) | Template matching method for character recognizing and processing | |
CN104657711B (en) | A kind of readings of pointer type meters automatic identifying method of robust | |
CN105373135A (en) | Method and system for guiding airplane docking and identifying airplane type based on machine vision | |
CN104766058A (en) | Method and device for obtaining lane line | |
CN104268857A (en) | Rapid sub pixel edge detection and locating method based on machine vision | |
CN110378957B (en) | Torpedo tank car visual identification and positioning method and system for metallurgical operation | |
CN111751380B (en) | Concrete dam crack inspection method based on light and small unmanned aerial vehicle | |
CN110197157B (en) | Pavement crack growth detection method based on historical crack data | |
CN105574527A (en) | Quick object detection method based on local feature learning | |
CN110910350A (en) | Nut loosening detection method for wind power tower cylinder | |
CN103438826A (en) | Three-dimensional measurement system and three-dimensional measurement method for steel plate with lasers and vision combined | |
CN114241469A (en) | Information identification method and device for electricity meter rotation process | |
CN106529548A (en) | Sub-pixel level multi-scale Harris corner point detection algorithm | |
CN104182752A (en) | Intelligent monitoring method of outdoor advertising board | |
CN103604388B (en) | The quick accurate positioning method of special marking thing in large-scale steel plate three-dimensional measurement | |
CN115984766A (en) | Rapid monocular vision three-dimensional target detection method for underground coal mine | |
CN103699876A (en) | Method and device for identifying vehicle number based on linear array CCD (Charge Coupled Device) images |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160302 Termination date: 20181108 |