CN107726990B - The acquisition of dot matrix grid image and recognition methods in a kind of Sheet metal forming strain measurement - Google Patents
The acquisition of dot matrix grid image and recognition methods in a kind of Sheet metal forming strain measurement Download PDFInfo
- Publication number
- CN107726990B CN107726990B CN201710842903.8A CN201710842903A CN107726990B CN 107726990 B CN107726990 B CN 107726990B CN 201710842903 A CN201710842903 A CN 201710842903A CN 107726990 B CN107726990 B CN 107726990B
- Authority
- CN
- China
- Prior art keywords
- dot matrix
- matrix grid
- grid image
- width
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000011159 matrix material Substances 0.000 title claims abstract description 214
- 238000000034 method Methods 0.000 title claims abstract description 65
- 239000002184 metal Substances 0.000 title claims abstract description 34
- 229910052751 metal Inorganic materials 0.000 title claims abstract description 34
- 238000005259 measurement Methods 0.000 title claims abstract description 24
- 238000000354 decomposition reaction Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 4
- 238000012804 iterative process Methods 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 230000003287 optical effect Effects 0.000 abstract description 5
- 238000005516 engineering process Methods 0.000 abstract description 3
- 238000011017 operating method Methods 0.000 abstract 1
- 238000012360 testing method Methods 0.000 description 19
- 230000006870 function Effects 0.000 description 10
- 230000008569 process Effects 0.000 description 5
- 238000000691 measurement method Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 2
- 239000010931 gold Substances 0.000 description 2
- 229910052737 gold Inorganic materials 0.000 description 2
- 238000010297 mechanical methods and process Methods 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 230000003321 amplification Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 210000004209 hair Anatomy 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000003913 materials processing Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 238000004080 punching Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/16—Measuring arrangements characterised by the use of optical techniques for measuring the deformation in a solid, e.g. optical strain gauge
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- 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
- G06T7/11—Region-based segmentation
-
- 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
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
-
- 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
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
The invention discloses the acquisition of dot matrix grid image and recognition methods in a kind of Sheet metal forming strain measurement, belong to optical 3-dimensional non-contact measuring technology field, and the realization step of this method, which includes that 1) shooting, exposes dot matrix grid image；2) weight map is calculated；3) weight map Pyramid transform；4) expose dot matrix grid image Pyramid transform more；5) laplacian pyramid assigns power；6) laplacian pyramid merges；7) dot matrix grid image reconstructs；8) dot matrix grid image binaryzation；9) sub-pix Boundary Recognition；10) ellipse fitting.By the operating procedure, the present invention can eliminate the influence of metal blank surface reflection, improve dot matrix grid image acquisition quality, improve the discrimination of dot matrix grid image.
Description
Technical field
The invention belongs to optical 3-dimensional non-contact measuring technology field, it is related in a kind of Sheet metal forming strain measurement
The acquisition of dot matrix grid image and recognition methods.Further relate to it is a kind of based on more exposure images fusion dot matrix grid image adopt
Collection and recognition methods.
Background technique
Sheet metal forming is one of the important branch in materials processing technology, aerospace, automobile, equipment manufacturing,
The each department of the national economy such as electric appliance is widely used.During Sheet metal forming, need by measuring metal
The three dimensional strain of panel surface shapes situation to analyze it, to monitor critical strain position, solve complicated forming, optimization punching press
Technique etc..The three dimensional strain measurement means and method on metal blank surface mainly include that Mechanical Method, electrical measuring method and optical 3-dimensional are non-
Contact type measurement method etc..Wherein, optical 3-dimensional non-contact type measuring method is current Sheet metal forming three-dimensional whole field strain measurement
Main means.
Chinese invention patent ZL201110263622.X discloses a kind of metal based on optical 3-dimensional non-contact measurement
Sheet forming strain measurement method.The point that this method uses digital camera or industrial CCD camera to prepare metal blank surface
Battle array lattice is sampled shooting, and the grid image for obtaining two width or two width or more is used for the three-dimensional reconstruction of grid, and root
According to the change in size of forming front and back three-dimensional grid, the three-dimensional whole field strain on metal blank surface is calculated.With Mechanical Method and electrical measurement
Method is compared, and this method has contactless, measurement efficiency is high, measurement accuracy is high, can get three-dimensional whole field to strain and be applicable to
The advantages that large scale Sheet metal forming product.However, due to the strong light-reflecting property on metal blank surface, in general, the point of acquisition
Light and shade differs greatly in battle array grid image, and especially in some reflective serious regions, the imaging of grid node in the picture is only protected
It stays seldom part or can not be imaged, so that grid node discrimination is low, and then influence the measurement of three dimensional strain.
Summary of the invention
To solve drawbacks described above existing in the prior art, the present invention provides a kind of points based on the fusion of more exposure images
The acquisition of battle array grid image and recognition methods, this method can eliminate the strong reflective influence in metal blank surface in sampling shooting process,
Improve dot matrix grid image shooting quality, improves the discrimination of dot matrix grid image.
To achieve the above objectives, the technical solution adopted by the present invention is that:
The acquisition of dot matrix grid image and recognition methods, include the following steps: in a kind of Sheet metal forming strain measurement
Step 1 shoots more exposure dot matrix grid images
It keeps camera motionless, the different dot matrix grid image of several width light exposures is shot under black and white screening-mode；
Step 2 calculates weight map
Contrast and exposure that each width captured by step 1 exposes dot matrix grid image more are calculated separately out, and will
The contrast that each width calculated exposes dot matrix grid image more is multiplied with exposure, obtains each width and exposes dot matrix grid chart more
The weight map of picture；
Step 3, weight map Pyramid transform
The weight map that each width calculated to step 2 exposes dot matrix grid image more carries out gaussian pyramid decomposition；
Step 4 expose dot matrix grid image Pyramid transform more
It exposes dot matrix grid image more to each width captured by step 1 and carries out Laplacian pyramid, decomposition
The number of plies is identical as the number of plies of weight map gaussian pyramid that step 3 is decomposed；
Step 5, laplacian pyramid assign power
Each width that step 4 is decomposed exposes the image on each layer of laplacian pyramid of dot matrix grid image more
The image in the weight map gaussian pyramid respective layer of dot matrix grid image is exposed with the width that step 3 is decomposed to be multiplied more, is obtained
Obtain entitled laplacian pyramid；
Step 6, laplacian pyramid fusion
Step 5 is obtained entitled several width to expose on the laplacian pyramid identical layer of dot matrix grid image more
Image addition；
Step 7, the reconstruct of dot matrix grid image
The laplacian pyramid merged to step 6 carries out inverse tower-shaped transformation, reconstructs the new dot matrix grid chart of a width
Picture；
Step 8, dot matrix grid image binaryzation
The dot matrix grid image reconstructed to step 7 carries out local auto-adaptive binary conversion treatment, obtains bianry image；
Step 9, sub-pix Boundary Recognition
The whole pixel side of dot matrix grid node in step 8 bianry image obtained is identified using 8 connected domain rules
Boundary；On this basis, the sub-pix boundary of dot matrix grid node is identified using space moments method；
Step 10, ellipse fitting
To the sub-pix boundary for the dot matrix grid node that step 9 is identified, ellipse is gone out using least square method iterative fitting
Centre coordinate, to obtain the coordinate of dot matrix grid node.
Further, the acquisition of dot matrix grid image with recognition methods further includes in the Sheet metal forming strain measurement,
Before carrying out the more exposure dot matrix grid images of step 1 shooting, first focus, so that the dot matrix grid on metal blank surface
It can clearly be imaged, focal length is locked after adjusting.
Further, when the step 1 shooting exposes dot matrix grid image more, camera is adjusted to black and white screening-mode, is protected
It holds camera to stablize, adjusts light exposure, shoot the dot matrix grid chart of 3 width or 3 width or more from under-exposure, normal exposure to overexposure
Picture.
Further, weight map calculation method is as follows in the step 2:
2.1) dot matrix grid image is exposed to each width captured by step 1 more, is filtered using Laplace operator
Wave takes absolute value filter result, just obtains the contrast factor that each width exposes dot matrix grid image more；
2.2) dot matrix grid image is exposed more to each width captured by step 1, the exposure factor uses following formula
It calculates:
Wherein, e indicates that natural constant, k indicate that kth width exposes dot matrix grid image more, and (i, j) indicates pixel position,
Ik(i, j) indicates that kth width exposes the gray value of pixel of the dot matrix grid image at the position (i, j), E morek(i, j) indicates the
K width exposes the exposure factor of the dot matrix grid image at the position (i, j) more, and σ indicates kernel function width；
2.3) it on the basis of step 2.1) contrast factor calculated and 2.2) the exposure factor calculated, adopts
The weight factor that each width exposes dot matrix grid image more is calculated with following formula:
Wherein,Indicate that kth width exposes weight factor of the dot matrix grid image at the position (i, j), C morek(i,j)
Indicate that step 2.1) kth width calculated exposes the comparison degree factor of the dot matrix grid image at the position (i, j) more；
2.4) place is normalized in the weight factor that each width calculated to step 2.3) exposes dot matrix grid image more
Reason obtains the weight map that each width exposes dot matrix grid image more, and method for normalizing is as follows:
Wherein, Wk(i, j) indicates that kth width exposes pixel of the weight map of dot matrix grid image at the position (i, j) more
Gray value, N indicate step 1 captured by more exposure dot matrix grid images width number, s indicate s width more expose dot matrix grid
Image,Indicate step 2.3) s width calculated expose more weight of the dot matrix grid image at the position (i, j) because
Son.
Further, the Laplace operator h used in the step 2.2) are as follows:
Further, the calculation method of Pyramid transform number of plies L is as follows in the step 3:
Wherein, r and c respectively indicates the line number and columns of more exposure dot matrix grid images, and ln () expression is with natural constant e
The natural logrithm function at bottom, min (r, c) indicate that the line number of more exposure dot matrix grid images and columns are minimized function.
Further, dot matrix grid image local auto-adaptive binarization method is as follows in the step 8:
8.1) quick binaryzation is carried out using the dot matrix grid image reconstructed by window binaryzation method to step 7, obtained
Bianry image；
8.2) connected component labeling is carried out to step 8.1) bianry image obtained, pixel number is greater than given
The connected region of threshold tau, further using binaryzation method pixel-by-pixel to all pixels point in connected region again binaryzation:
A) binarization threshold is calculated for any pixel point in the connected region greater than given threshold value τ:
Wherein, (i, j) indicates pixel position, and T (i, j) indicates two of the pixel in connected region at the position (i, j)
Value threshold value,In the dot matrix grid image that expression step 7 is reconstructed
The average gray value of the pixel in u × v local window centered on (i, j), u and v respectively indicate the length dimension of window
And height dimension, m and n indicate two integer variables, I '0The dot matrix grid image that (m, n) expression step 7 is reconstructed is at (m, n)
The gray value of pixel at position,It indicates
The grey value profile of the pixel in u × v local window in the dot matrix grid image that step 7 is reconstructed centered on (i, j)
Standard deviation value, γ and offset are two constants；
B) according to step a) binarization threshold calculated to pixel again binaryzation:
Wherein, I ' (i, j) indicates picture of the step 8.1) bianry image obtained at the position (i, j) after binaryzation again
The gray value of vegetarian refreshments, I '0The ash of pixel of the dot matrix grid image that (i, j) expression step 7 is reconstructed at the position (i, j)
Angle value.
Further, the value of threshold tau is τ=10000 in the step 8.2).
Further, in the step 10 in each iterative process, it is elliptical to being fitted to calculate each boundary point
Spacing and standard deviation value sigma are greater than the boundary point of 3sigma to spacing, are not involved in interative computation next time.
Compared with prior art, the method for the present invention has the advantage that
(1) the method for the present invention can eliminate the strong reflective influence in metal blank surface, collect the dot matrix grid chart of high quality
Picture；
(2) the dot matrix grid image quality of the method for the present invention acquisition is high, therefore the discrimination of dot matrix grid image is also corresponding
It improves；
(3) the dot matrix grid image quality of the method for the present invention acquisition is high, therefore the accuracy of identification of dot matrix grid image also phase
It should improve；
(4) the method for the present invention improves the discrimination and accuracy of identification of dot matrix grid image, therefore also improves indirectly
The reconstruction precision of three-dimensional grid and the measurement accuracy of three dimensional strain.
Detailed description of the invention
The flow chart of Fig. 1 concrete operation step of the present invention.
The under-exposure dot matrix grid image of certain cupping test specimen of Fig. 2 a shooting.
Certain cupping test specimen normal exposure dot matrix grid image of Fig. 2 b shooting.
Certain cupping test specimen overexposure dot matrix grid image of Fig. 2 c shooting.
Certain cupping test specimen dot matrix grid image of Fig. 3 reconstruct.
Dot matrix grid image testing result that Fig. 4 cupping test specimen is under-exposure.
Fig. 5 cupping test specimen normal exposure dot matrix grid image testing result.
Fig. 6 cupping test specimen overexposure dot matrix grid image testing result.
Certain cupping test specimen dot matrix grid image testing result of Fig. 7 reconstruct.
Specific embodiment
The invention will be described in further detail with reference to the accompanying drawings and examples, but is not intended as doing invention any limit
The foundation of system.
The present invention proposes the acquisition of dot matrix grid image and recognition methods, such as Fig. 1 in a kind of Sheet metal forming strain measurement
It is shown.When acquisition and the dot matrix grid image on identification a certain visual angle metal blank surface, the first step shoots and exposes dot matrix grid more
Image.It before shooting, first focuses, so that the dot matrix grid on metal blank surface can clearly be imaged, is locked after adjusting
Focal length.It is variegated in order to reduce when shooting, camera is adjusted to black and white screening-mode, camera is kept to stablize, adjusts light exposure, shooting 3
Dot matrix grid image more than width or 3 width from under-exposure, normal exposure to overexposure.
Second step calculates weight map.Calculate the weight map that each width captured by the first step exposes dot matrix grid image more.
Weight map calculation process includes the following steps:
1) contrast factor is calculated.Dot matrix grid image is exposed more to each width captured by the first step, using La Pula
This operator h is filtered, and filter result is taken absolute value, just obtain each width more expose dot matrix grid image contrast because
Son.Used Laplace operator h are as follows:
2) the exposure factor is calculated.Dot matrix grid image is exposed more to each width captured by the first step, using following public affairs
Formula calculates its exposure factor:
Wherein, e indicates that natural constant, k indicate that kth width exposes dot matrix grid image more, and (i, j) indicates pixel position,
Ek(i, j) indicates that kth width exposes the exposure factor of the dot matrix grid image at the position (i, j), I morek(i, j) indicates kth width
The gray value of pixel of more exposure dot matrix grid images at the position (i, j), σ indicate kernel function width, value range 0
σ≤0.5 <, generally selects σ=0.2.
3) weight factor is calculated.In 1) step contrast factor calculated and 2) the step exposure factor calculated
On the basis of, the weight factor that each width exposes dot matrix grid image more is calculated using following formula:
Wherein,Indicate that kth width exposes weight factor of the dot matrix grid image at the position (i, j), C morek(i,j)
Indicate the 1) step kth width calculated expose the comparison degree factor of the dot matrix grid image at the position (i, j) more.
4) weight factor normalizes.To the 3) step each width calculated expose more the weight factor of dot matrix grid image into
Row normalized obtains the weight map that each width exposes dot matrix grid image more, and the normalization formula of use is as follows:
Wherein, Wk(i, j) indicates that kth width exposes pixel of the weight map of dot matrix grid image at the position (i, j) more
Gray value, N indicate the first step captured by more exposure dot matrix grid images width number, s indicate s width more expose dot matrix grid
Image,Indicate the 3) step s width calculated expose weight factor of the dot matrix grid image at the position (i, j) more.
Third step, weight map Pyramid transform.Each width calculated to second step exposes the weight of dot matrix grid image more
Figure carries out gaussian pyramid decomposition.Decomposition process comprises the following steps:
1) the Pyramid transform number of plies is calculated.The calculation method of Pyramid transform number of plies L is as follows:
Wherein, r and c respectively indicates the line number and columns of more exposure dot matrix grid images, and ln () expression is with natural constant e
The natural logrithm function at bottom, min (r, c) indicate that the line number of more exposure dot matrix grid images and columns are minimized function.
2) the 0th layer of weight map gaussian pyramid is constructed.Second step each width calculated is exposed into dot matrix grid image more
Weight map as the image on the 0th layer of weight map gaussian pyramid:
Gk,0(i, j)=Wk(i, j),
Wherein, k indicates that kth width exposes dot matrix grid image more, and (i, j) indicates pixel position, Gk,0(i, j) indicates kth
Width exposes the ash of pixel of the image at the position (i, j) on the 0th layer of weight map Gauss pyramid of dot matrix grid image more
Angle value, Wk(i, j) indicates that second step kth width calculated exposes the weight map of dot matrix grid image at the position (i, j) more
The gray value of pixel.
3) t layers of weight map gaussian pyramid is constructed.By on weight map gaussian pyramid t-1 layer image and one 5
× 5 window functions carry out convolution, then convolution results are made interlacing and just obtain t layers of weight map gaussian pyramid every the down-sampled of column
On image:
Wherein, t indicates pyramidal t layers, Gk,t(i, j) indicates that kth width exposes the weight map height of dot matrix grid image more
The gray value of pixel of the image at the position (i, j) on this pyramid t layer, m and n indicate two integer variables, Gk,t-1
(2i+m, 2j+n) indicates that kth width exposes the image on the weight map gaussian pyramid t-1 layer of dot matrix grid image in (2i more
+ m, 2j+n) pixel at position gray value, λ (m, n) indicates 5 × 5 window functions:
The 2) kth width that step is constructed expose the 0th layer of weight map gaussian pyramid of basis of dot matrix grid image more
On, according to described t layers of building method of weight map gaussian pyramid, the power that kth width exposes dot matrix grid image more can be constructed
The 1st layer of multigraph gaussian pyramid, and so on, the weight map Gauss gold word that kth width exposes dot matrix grid image more can be constructed
L layers of tower.
4th step expose dot matrix grid image Pyramid transform more.Dot matrix net is exposed more to each width captured by the first step
Table images carry out Laplacian pyramid.Decomposition process comprises the following steps:
1) mostly exposure dot matrix grid image gaussian pyramid decomposes.According to weight map Pyramid transform side described in third step
Method exposes dot matrix grid image more to each width captured by the first step and carries out gaussian pyramid decomposition, and the height that will be decomposed
Image on this pyramid t layer is denoted as G 'k,t, wherein k indicates that kth width exposes dot matrix grid image more；
2) L layers of laplacian pyramid are constructed.Using the image on 1) gaussian pyramid L layer that step is decomposed as
Image on laplacian pyramid L layer:
LPk,L(i, j)=G 'k,L(i, j),
Wherein, (i, j) indicates pixel position, and L indicates the third step Pyramid transform number of plies calculated, LPk,L(i, j) table
Show that kth width exposes pixel of the image at the position (i, j) on dot matrix grid image laplacian pyramid L layer more
Gray value, G 'k,L(i, j) indicates the 1st) the kth width that is decomposed of step exposes on dot matrix grid image gaussian pyramid L layer more
The gray value of pixel of the image at the position (i, j)；
3) t layers of laplacian pyramid are constructed.T layers of building method of laplacian pyramid are as follows:
A) image interpolation on the 1) gaussian pyramid t+1 layer that step is decomposed is amplified:
Wherein, t+1 indicates t+1 layers of pyramid,Indicate the amplified image of interpolation at the position (i, j)
The gray value of pixel,M and n
Indicate two integer variables,Indicating the, 1) to expose dot matrix grid image high more for kth width that step is decomposed
Image on this pyramid t+1 layer existsThe gray value of pixel at position,Table
Show that a transition variable, λ (m, n) indicate 5 × 5 window functions described in third step；
B) by 1) gaussian pyramid t layer that step is decomposed image with a) the amplified image of step interpolation do and subtract
Method operation:
Wherein, LPk,t(i, j) indicates that kth width exposes the image on dot matrix grid image laplacian pyramid t layer more
The gray value of pixel at the position (i, j), G 'k,t(i, j) indicates the 1st) the kth width that is decomposed of step exposes dot matrix grid more
The gray value of pixel of the image at the position (i, j) on image gaussian pyramid t layer；
According to t layers of building method of the laplacian pyramid, L-1 layers of laplacian pyramid can be successively constructed
To the 0th layer.
5th step, laplacian pyramid assign power.The each width decomposed to the 4th step exposes dot matrix grid image more
Laplacian pyramid carries out tax power with the following method:
LP′k,t(i, j)=Gk,t(i,j)×LPk,t(i,j),
Wherein, k indicates that kth width exposes dot matrix grid image more, and t indicates t layers of pyramid, and (i, j) indicates pixel point
It sets, LP 'k,t(i, j) indicates to assign the image on the mostly exposure dot matrix grid image laplacian pyramid t layer of the kth width after weighing
The gray value of pixel at the position (i, j), Gk,t(i, j) indicates that the kth width that third step is decomposed exposes dot matrix grid more
The gray value of pixel of the image at the position (i, j) on the weight map gaussian pyramid t layer of image, LPk,t(i, j) table
Show that kth width that the 4th step is decomposed exposes the image on dot matrix grid image laplacian pyramid t layer in the position (i, j) more
The gray value of the gray value of the pixel at place, two pixels of symbol × expression is multiplied.
6th step, laplacian pyramid fusion.Laplacian pyramid fusion method is as follows:
Wherein, k indicates that kth width exposes dot matrix grid image more, and t indicates t layers of pyramid, and (i, j) indicates pixel point
It setting, N indicates the width number of more exposure dot matrix grid images captured by the first step,Indicate fused Laplce's gold
The gray value of pixel of the image at the position (i, j) on word tower t layer, LP 'k,t(i, j) indicates the 5th step institute entitled the
K width exposes the gray scale of pixel of the image at the position (i, j) on dot matrix grid image laplacian pyramid t layer more
Value.
7th step, the reconstruct of dot matrix grid image.The laplacian pyramid merged to the 6th step carries out inverse tower-shaped transformation,
Construct the new dot matrix grid image of a width.Inverse tower-shaped transformation includes the following steps:
1) restore L layers of gaussian pyramid.The image on laplacian pyramid L layer that 6th step is merged as
Image on gaussian pyramid L layer to be restored:
Wherein, (i, j) indicates pixel position, and L indicates the third step Pyramid transform number of plies calculated, I 'L(i, j) is indicated
The gray value of pixel of the image on gaussian pyramid L layer restored at the position (i, j),Indicate the 6th
Walk the gray value of pixel of the image on merged laplacian pyramid L layer at the position (i, j)；
2) restore t layers of gaussian pyramid.T layers of restoration methods of gaussian pyramid are as follows:
A) the image interpolation amplification on the laplacian pyramid t+1 layer for being merged the 6th step:
Wherein, t+1 indicates t+1 layers of pyramid,Indicate the amplified image of interpolation at the position (i, j)
The gray value of pixel, m and n indicate that two integer variables, λ (m, n) indicate 5 × 5 window functions described in third step,Table
Show that the image on laplacian pyramid t+1 layer that the 6th step is merged existsThe ash of pixel at position
Angle value,Indicate a transition variable；
B) by the image on a) laplacian pyramid t layer that the amplified image of step interpolation is merged with the 6th step
Do add operation:
Wherein, I 't(i, j) indicates pixel of the image at the position (i, j) on restored gaussian pyramid t layer
Gray value,Indicate the image on laplacian pyramid t layer that the 6th step is merged at the position (i, j)
The gray value of pixel；
According to t layers of restoration methods of the gaussian pyramid, it can successively restore L-1 layers to the 0th layer of gaussian pyramid, institute
Image I ' on the 0th layer of gaussian pyramid of recovery0Dot matrix grid image exactly to be reconstructed.
8th step, dot matrix grid image self-adaption binaryzation.Self-adaption binaryzation process comprises the following steps:
1) document " Circular grid pattern based surface strain measurement is used
system for sheet metal forming”(Shi B and Liang J,OPT LASER ENG,2012,50(9):
The dot matrix grid image reconstructed by window binaryzation method to step 7 described in 1186-1195) carries out quick binaryzation,
Obtain bianry image.
2) to the 1) step bianry image obtained carry out connected component labeling, given threshold value is greater than for pixel number
The region of τ, further using binaryzation method pixel-by-pixel to all pixels point in connected region again binaryzation:
A) binarization threshold is calculated for any pixel point in the connected region greater than given threshold value τ:
Wherein, (i, j) indicates pixel position, and T (i, j) indicates two of the pixel in connected region at the position (i, j)
Value threshold value,It indicates in dot matrix grid image that the 7th step is reconstructed
The average gray value of the pixel in u × v local window centered on (i, j), m and n indicate two integer variables, u and v
The length dimension and height dimension for respectively indicating window, usually take u=v=23, I '0(m, n) indicates the point that step 7 is reconstructed
The gray value of pixel of the battle array grid image at the position (m, n),Indicate the dot matrix grid chart that the 7th step is reconstructed
The standard deviation value of the grey value profile of the pixel in u × v local window as in centered on (i, j), γ and offset
For two constants, γ=0.1, offset=5.0 are generally selected, threshold tau is positive integer, generally takes τ=10000；
B) according to step a) binarization threshold calculated to pixel again binaryzation:
Wherein, I ' (i, j) indicates picture of the step 8.1) bianry image obtained at the position (i, j) after binaryzation again
The gray value of vegetarian refreshments, I '0The ash of pixel of the dot matrix grid image that (i, j) expression step 7 is reconstructed at the position (i, j)
Angle value,.
9th step, sub-pix Boundary Recognition.Sub-pix Boundary Recognition method is as follows:
1) the whole pixel side of dot matrix grid node in the 8th step bianry image obtained is identified using 8 connected domain rules
Boundary；
2) further using space moments method on the basis of the 1) whole pixel boundary of dot matrix grid node that step is identified
Identify the sub-pix boundary of dot matrix grid node.
Tenth step, ellipse fitting.The sub-pix boundary for the dot matrix grid node in bianry image that 9th step is identified,
Using document " Least squares fitting of circles and ellipses " (W.Gander, G.H.Golub
And R.Strebel, BIT Numerical Mathematics, 1994,34:558-578) least-squares iteration described in
Method fits elliptical center coordinate.In order to reduce noise jamming, in each iterative process, calculates each boundary point and extremely intend
The elliptical spacing and standard deviation value sigma closed, the boundary point of 3sigma is greater than to spacing, is not involved in iteration next time
Operation.The elliptical center coordinate being finally fitted is the coordinate of dot matrix grid node.
Below in conjunction with physical simulation experiment, the present invention will be described, and wherein the method for the present invention is flat in VS2010 and opengl
Realized on platform corresponding algorithm and Intel i7-4770CPU 3.4GHz, 16GB memory PC machine on run.
3 width cupping test specimens of shooting expose dot matrix grid image as shown in Fig. 2 a- Fig. 2 c more, wherein Fig. 2 a is shown
The under-exposure dot matrix grid image of certain cupping test specimen of shooting, Fig. 2 b show certain cupping test specimen normal exposure dot matrix grid of shooting
Image, Fig. 2 c show certain cupping test specimen overexposure dot matrix grid image of shooting.Fig. 3 is shown according to 3 captured width
The new dot matrix grid image of the width that cupping test specimen mostly exposure dot matrix grid image is reconstructed using the method for the present invention.Pass through comparison
Fig. 2 a- Fig. 2 c and Fig. 3 can be seen that dot matrix grid in reconstructed dot matrix grid image and be more clear.Fig. 4 to Fig. 7 is shown
It certain cupping test specimen is under-exposure dot matrix grid image, normal exposure dot matrix grid image, overexposure dot matrix grid image and is reconstructed
Dot matrix grid node testing result in dot matrix grid image, dot matrix grid node discrimination is followed successively by 49.6%, 77.78%,
72.3%, 86.87%.Wherein, white cross hairs indicates the coordinate of dot matrix grid node detected.Schemed by comparison diagram 4-
7 as can be seen that the number that dot matrix grid node detects in the dot matrix grid image reconstructed is most, and dot matrix grid node identifies
Rate highest is 86.87%.Alternatively bright through this embodiment, it is strong reflective that the method for the present invention can eliminate metal blank surface
It influences, improves dot matrix grid image acquisition quality, improve the discrimination of dot matrix grid image.
Embodiments of the present invention are described above in conjunction with drawings and examples, but the present invention is not limited to above-mentioned realities
Mode is applied, it within the knowledge of one of ordinary skill in the art, can also be before not departing from present inventive concept
It puts and makes a variety of changes.
Claims (9)
1. the acquisition of dot matrix grid image and recognition methods in a kind of Sheet metal forming strain measurement, which is characterized in that including under
State step:
Step 1 shoots more exposure dot matrix grid images
It keeps camera motionless, the different dot matrix grid image of several width light exposures is shot under black and white screening-mode；
Step 2 calculates weight map
It calculates separately out contrast and exposure that each width captured by step 1 exposes dot matrix grid image more, and will calculate
Each width expose the contrast of dot matrix grid image more and be multiplied with exposure, obtain the dot matrix grid image of exposure more than each width
Weight map；
Step 3, weight map Pyramid transform
The weight map that each width calculated to step 2 exposes dot matrix grid image more carries out gaussian pyramid decomposition；
Step 4 expose dot matrix grid image Pyramid transform more
It exposes dot matrix grid image more to each width captured by step 1 and carries out Laplacian pyramid, the number of plies of decomposition
It is identical as the number of plies of weight map gaussian pyramid that step 3 is decomposed；
Step 5, laplacian pyramid assign power
Each width that step 4 is decomposed exposes image and step on each layer of laplacian pyramid of dot matrix grid image more
The image that rapid three width decomposed expose in the weight map gaussian pyramid respective layer of dot matrix grid image more is multiplied, and is assigned
The laplacian pyramid of power；
Step 6, laplacian pyramid fusion
Step 5 is obtained into the figure on entitled several width mostly laplacian pyramid identical layer of exposure dot matrix grid image
As being added；
Step 7, the reconstruct of dot matrix grid image
The laplacian pyramid merged to step 6 carries out inverse tower-shaped transformation, reconstructs the new dot matrix grid image of a width；
Step 8, dot matrix grid image binaryzation
The dot matrix grid image reconstructed to step 7 carries out local auto-adaptive binary conversion treatment, obtains bianry image；
Step 9, sub-pix Boundary Recognition
The whole pixel boundary of dot matrix grid node in step 8 bianry image obtained is identified using 8 connected domain rules；?
On the basis of this, the sub-pix boundary of dot matrix grid node is identified using space moments method；
Step 10, ellipse fitting
To the sub-pix boundary for the dot matrix grid node that step 9 is identified, elliptical center is gone out using least square method iterative fitting
Coordinate, to obtain the coordinate of dot matrix grid node.
2. the acquisition of dot matrix grid image and identification side in a kind of Sheet metal forming strain measurement according to claim 1
Method, it is characterised in that further include before carrying out the more exposure dot matrix grid images of step 1 shooting, first focusing, so that golden
The dot matrix grid for belonging to panel surface can clearly be imaged, and focal length is locked after adjusting.
3. the acquisition of dot matrix grid image and identification side in a kind of Sheet metal forming strain measurement according to claim 1
Method, which is characterized in that when shooting more exposure dot matrix grid images in the step 1, camera is adjusted to black and white screening-mode, is protected
It holds camera to stablize, adjusts light exposure, shoot the dot matrix grid chart of 3 width or 3 width or more from under-exposure, normal exposure to overexposure
Picture.
4. the acquisition of dot matrix grid image and identification side in a kind of Sheet metal forming strain measurement according to claim 1
Method, which is characterized in that weight map calculation method is as follows in the step 2:
2.1) dot matrix grid image is exposed to each width captured by step 1 more, is filtered using Laplace operator, it will
Filter result takes absolute value, and obtains the contrast factor that each width exposes dot matrix grid image more；
2.2) dot matrix grid image is exposed more to each width captured by step 1, the exposure factor uses following formula meter
It calculates:
Wherein, e indicates that natural constant, k indicate that kth width exposes dot matrix grid image more, and (i, j) indicates pixel position, Ik(i,
J) indicate that kth width exposes the gray value of pixel of the dot matrix grid image at the position (i, j), E morek(i, j) indicates that kth width is more
The exposure factor of the dot matrix grid image at the position (i, j) is exposed, σ indicates kernel function width；
2.3) it on the basis of step 2.1) contrast factor calculated and step 2.2) the exposure factor calculated, uses
Following formula calculates the weight factor that each width exposes dot matrix grid image more:
Wherein,Indicate that kth width exposes weight factor of the dot matrix grid image at the position (i, j), C morek(i, j) is indicated
Step 2.1) kth width calculated exposes the comparison degree factor of the dot matrix grid image at the position (i, j) more；
2.4) weight factor that each width calculated to step 2.3) exposes dot matrix grid image more is normalized, and obtains
The weight map that each width exposes dot matrix grid image more is obtained, method for normalizing is as follows:
Wherein, Wk(i, j) indicates that kth width exposes the gray scale of pixel of the weight map of dot matrix grid image at the position (i, j) more
Value, N indicate that the width number of more exposure dot matrix grid images captured by step 1, s indicate that s width exposes dot matrix grid image more,Indicate that step 2.3) s width calculated exposes weight factor of the dot matrix grid image at the position (i, j) more.
5. the acquisition of dot matrix grid image and identification side in a kind of Sheet metal forming strain measurement according to claim 4
Method, which is characterized in that the Laplace operator h used in the step 2.1) are as follows:
6. the acquisition of dot matrix grid image and identification side in a kind of Sheet metal forming strain measurement according to claim 1
Method, which is characterized in that the calculation method of Pyramid transform number of plies L is as follows in the step 3:
Wherein, r and c respectively indicates the line number and columns of more exposure dot matrix grid images, and ln () is indicated using natural constant e the bottom of as
Natural logrithm function, min (r, c) indicate that the line number of more exposure dot matrix grid images and columns are minimized function.
7. the acquisition of dot matrix grid image and identification side in a kind of Sheet metal forming strain measurement according to claim 1
Method, which is characterized in that dot matrix grid image local auto-adaptive binarization method is as follows in the step 8:
8.1) quick binaryzation is carried out using the dot matrix grid image reconstructed by window binaryzation method to step 7, obtains two-value
Image；
8.2) connected component labeling is carried out to step 8.1) bianry image obtained, given threshold value is greater than for pixel number
The connected region of τ, further using binaryzation method pixel-by-pixel to all pixels point in connected region again binaryzation:
A) binarization threshold is calculated for any pixel point in the connected region greater than given threshold value τ:
Wherein, (i, j) indicates pixel position, and T (i, j) indicates the binaryzation of the pixel in connected region at the position (i, j)
Threshold value,Indicate in the dot matrix grid image that is reconstructed of step 7 with
The average gray value of the pixel in u × v local window centered on (i, j), u and v respectively indicate window length dimension and
Height dimension, m and n indicate two integer variables, I '0The dot matrix grid image that (m, n) expression step 7 is reconstructed is in the position (m, n)
The gray value of the pixel at place is set,Indicate step
The grey value profile of the pixel in u × v local window in the rapid seven dot matrix grid images reconstructed centered on (i, j)
Standard deviation value, γ and offset are two constants；
B) according to step a) binarization threshold calculated to pixel again binaryzation:
Wherein, I ' (i, j) indicates pixel of the step 8.1) bianry image obtained at the position (i, j) after binaryzation again
Gray value, I '0The gray value of pixel of the dot matrix grid image that (i, j) expression step 7 is reconstructed at the position (i, j).
8. the acquisition of dot matrix grid image and identification side in a kind of Sheet metal forming strain measurement according to claim 7
Method, which is characterized in that the value of threshold tau is τ=10000 in the step 8.2).
9. the acquisition of dot matrix grid image and identification side in a kind of Sheet metal forming strain measurement according to claim 1
Method, which is characterized in that in the step 10 in each iterative process, calculate each boundary point to the elliptical spacing being fitted
And standard deviation value sigma, the boundary point of 3sigma is greater than to spacing, is not involved in interative computation next time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710842903.8A CN107726990B (en) | 2017-09-18 | 2017-09-18 | The acquisition of dot matrix grid image and recognition methods in a kind of Sheet metal forming strain measurement |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710842903.8A CN107726990B (en) | 2017-09-18 | 2017-09-18 | The acquisition of dot matrix grid image and recognition methods in a kind of Sheet metal forming strain measurement |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107726990A CN107726990A (en) | 2018-02-23 |
| CN107726990B true CN107726990B (en) | 2019-06-21 |
Family
ID=61207482
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710842903.8A Active CN107726990B (en) | 2017-09-18 | 2017-09-18 | The acquisition of dot matrix grid image and recognition methods in a kind of Sheet metal forming strain measurement |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107726990B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116297336B (en) * | 2023-03-27 | 2024-01-23 | 钢研纳克(沈阳)检测技术有限公司 | Method for detecting defect points on surface of metal sheet for outer layer protection |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108507476B (en) * | 2018-04-27 | 2020-08-07 | 中国石油大学(北京) | Displacement field measuring method, device, equipment and storage medium for material surface |
| CN109753766A (en) * | 2018-12-29 | 2019-05-14 | 中链科技有限公司 | Image artifacts verification information processing method and verification method based on block chain |
| CN111832433B (en) * | 2020-06-24 | 2023-12-29 | 奇点微(上海)光电科技有限公司 | Device for extracting object characteristics from image and working method thereof |
| CN112935013B (en) * | 2021-02-10 | 2022-01-07 | 哈尔滨工业大学 | Method for measuring shear strain of viscous medium on plate blank |
| CN113112470B (en) * | 2021-04-08 | 2023-12-15 | 揭阳市柏亿不锈钢有限公司 | Cold rolling process steel bending detection analysis method based on artificial intelligence |
| CN113155648B (en) * | 2021-05-18 | 2022-10-11 | 苏州大学 | Material micro-deformation measurement method and system based on impact test |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20000040421A (en) * | 1998-12-18 | 2000-07-05 | Inst Of Ind Science Pohang | System for measuring camber of slab using pyramid image |
| CN101226635A (en) * | 2007-12-18 | 2008-07-23 | 西安电子科技大学 | Multisource image anastomosing method based on comb wave and Laplace tower-shaped decomposition |
| CN101447072A (en) * | 2009-01-06 | 2009-06-03 | 覃征 | pyramidal empirical modal analyze image merge method |
| CN104050639A (en) * | 2014-06-16 | 2014-09-17 | 西安电子科技大学 | Multi-view dense point cloud data fusion method based on two-sided filter |
| CN104202538A (en) * | 2014-09-10 | 2014-12-10 | 浙江广播电视集团 | Double-registration method for different-exposure images in wide dynamic camera |
| CN104835130A (en) * | 2015-04-17 | 2015-08-12 | 北京联合大学 | Multi-exposure image fusion method |
-
2017
- 2017-09-18 CN CN201710842903.8A patent/CN107726990B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20000040421A (en) * | 1998-12-18 | 2000-07-05 | Inst Of Ind Science Pohang | System for measuring camber of slab using pyramid image |
| CN101226635A (en) * | 2007-12-18 | 2008-07-23 | 西安电子科技大学 | Multisource image anastomosing method based on comb wave and Laplace tower-shaped decomposition |
| CN101447072A (en) * | 2009-01-06 | 2009-06-03 | 覃征 | pyramidal empirical modal analyze image merge method |
| CN104050639A (en) * | 2014-06-16 | 2014-09-17 | 西安电子科技大学 | Multi-view dense point cloud data fusion method based on two-sided filter |
| CN104202538A (en) * | 2014-09-10 | 2014-12-10 | 浙江广播电视集团 | Double-registration method for different-exposure images in wide dynamic camera |
| CN104835130A (en) * | 2015-04-17 | 2015-08-12 | 北京联合大学 | Multi-exposure image fusion method |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116297336B (en) * | 2023-03-27 | 2024-01-23 | 钢研纳克(沈阳)检测技术有限公司 | Method for detecting defect points on surface of metal sheet for outer layer protection |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107726990A (en) | 2018-02-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107726990B (en) | The acquisition of dot matrix grid image and recognition methods in a kind of Sheet metal forming strain measurement | |
| CN111046880B (en) | Infrared target image segmentation method, system, electronic equipment and storage medium | |
| CN110263705A (en) | Towards two phase of remote sensing technology field high-resolution remote sensing image change detecting method | |
| CN108875821A (en) | The training method and device of disaggregated model, mobile terminal, readable storage medium storing program for executing | |
| CN103955922B (en) | Method for detecting flaws of printed fabric based on Gabor filter | |
| CN107229918A (en) | A kind of SAR image object detection method based on full convolutional neural networks | |
| CN109903282B (en) | Cell counting method, system, device and storage medium | |
| CN109753996B (en) | Hyperspectral image classification method based on three-dimensional lightweight depth network | |
| CN110334727B (en) | Intelligent matching detection method for tunnel cracks | |
| CN105335952B (en) | Matching power flow computational methods and device and parallax value calculating method and equipment | |
| CN101976436A (en) | Pixel-level multi-focus image fusion method based on correction of differential image | |
| CN105405138A (en) | Water surface target tracking method based on saliency detection | |
| Zhao et al. | An adaptation of CNN for small target detection in the infrared | |
| Cheng et al. | Fast and lightweight network for single frame structured illumination microscopy super-resolution | |
| CN105118035A (en) | Self-adaptive optical spot signal extraction method based on sparse representation | |
| CN113658147B (en) | Workpiece size measuring device and method based on deep learning | |
| CN110321869A (en) | Personnel's detection and extracting method based on Multiscale Fusion network | |
| CN107358625B (en) | SAR image change detection method based on SPP Net and region-of-interest detection | |
| CN116763295B (en) | Livestock scale measuring method, electronic equipment and storage medium | |
| CN116385401B (en) | High-precision visual detection method for textile defects | |
| Bhatnagar et al. | Multi-sensor fusion based on local activity measure | |
| CN116823863A (en) | Infrared image contour extraction method and device | |
| CN116843906A (en) | Target multi-angle intrinsic feature mining method based on Laplace feature mapping | |
| Thuan et al. | Edge-focus thermal image super-resolution using generative adversarial network | |
| CN116259087A (en) | Low-resolution face recognition method |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20221026 Address after: 710086 Room 006, F1901, 19/F, Building 4-A, Xixian Financial Port, Fengdong New City Energy Jinmao District, Xixian New District, Xi'an, Shaanxi Patentee after: Xi'an Baochuang Suwei Intelligent Technology Co.,Ltd. Address before: 710065 No. 2 Taibai South Road, Yanta District, Xi'an, Shaanxi Patentee before: XIDIAN University |
|
| TR01 | Transfer of patent right |