CN107726990A - The collection of dot matrix grid image and recognition methods in a kind of Sheet metal forming strain measurement - Google Patents
The collection of dot matrix grid image and recognition methods in a kind of Sheet metal forming strain measurement Download PDFInfo
- Publication number
- CN107726990A CN107726990A CN201710842903.8A CN201710842903A CN107726990A CN 107726990 A CN107726990 A CN 107726990A CN 201710842903 A CN201710842903 A CN 201710842903A CN 107726990 A CN107726990 A CN 107726990A
- Authority
- CN
- China
- Prior art keywords
- grid image
- lattice grid
- exposure
- image
- dot matrix
- 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 61
- 239000011159 matrix material Substances 0.000 title claims abstract description 54
- 239000002184 metal Substances 0.000 title claims abstract description 36
- 238000005259 measurement Methods 0.000 title claims abstract description 23
- 238000000354 decomposition reaction Methods 0.000 claims description 22
- 230000008569 process Effects 0.000 claims description 9
- 238000004891 communication Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 230000004927 fusion Effects 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 230000003044 adaptive effect Effects 0.000 claims description 2
- 230000003287 optical effect Effects 0.000 abstract description 5
- 238000005516 engineering process Methods 0.000 abstract description 2
- 238000011017 operating method Methods 0.000 abstract 1
- 238000012360 testing method Methods 0.000 description 14
- 230000006870 function Effects 0.000 description 10
- 238000000691 measurement method Methods 0.000 description 5
- 238000010276 construction Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000010297 mechanical methods and process Methods 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 230000003321 amplification Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000015654 memory Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000007500 overflow downdraw method 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
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 collection of dot matrix grid image and recognition methods in a kind of Sheet metal forming strain measurement, belongs to optical 3-dimensional non-contact measuring technology field, and this method realizes that step includes the more exposure dot matrix grid images of 1) shooting；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 the technical field of optical three-dimensional non-contact measurement, and relates to a method for acquiring and identifying a dot matrix grid image in metal sheet forming strain measurement. Further relates to a method for acquiring and identifying a dot matrix grid image based on multi-exposure image fusion.
Background
The metal plate forming is one of important branches in the material processing technology, and is widely applied to various departments of national economy such as aerospace, automobiles, equipment manufacturing, electric appliances and the like. In the forming process of the metal plate, the forming condition of the metal plate needs to be analyzed by measuring the three-dimensional strain of the surface of the metal plate, so that the critical deformation part is monitored, the complex forming is solved, the stamping process is optimized, and the like. The three-dimensional strain measurement means and method for the surface of the metal plate mainly comprise a mechanical method, an electrical measurement method, an optical three-dimensional non-contact measurement method and the like. The optical three-dimensional non-contact measurement method is a main means for measuring the three-dimensional full-field strain of the metal plate forming at present.
Chinese invention patent ZL201110263622.X discloses a metal sheet forming strain measurement method based on optical three-dimensional non-contact measurement. The method adopts a digital camera or an industrial CCD camera to sample and shoot a dot matrix grid pattern prepared on the surface of the metal plate, obtains two or more grid images for three-dimensional reconstruction of the grid, and calculates the three-dimensional full-field strain of the surface of the metal plate according to the size change of the three-dimensional grid before and after forming. Compared with a mechanical method and an electrical measurement method, the method has the advantages of non-contact type, high measurement efficiency, high measurement precision, capability of obtaining three-dimensional full-field strain, suitability for large-size metal plate forming parts and the like. However, due to the strong light reflection characteristic of the surface of the metal plate, the difference between light and shade in the acquired dot matrix grid image is usually large, and particularly in some areas with serious light reflection, the imaging of the grid nodes in the image only remains a few parts or cannot be imaged, so that the identification rate of the grid nodes is low, and the measurement of three-dimensional strain is influenced.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention provides a method for acquiring and identifying a dot matrix grid image based on multi-exposure image fusion, which can eliminate the influence of strong light reflection on the surface of a metal plate in the sampling and shooting process, improve the shooting quality of the dot matrix grid image and improve the identification rate of the dot matrix grid image.
In order to achieve the above purposes, the technical scheme adopted by the invention is as follows:
a method for acquiring and identifying a dot matrix grid image in sheet metal forming strain measurement comprises the following steps:
step one, shooting a multi-exposure lattice grid image
Keeping a camera still, and shooting a plurality of dot matrix grid images with different exposure quantities in a black-and-white shooting mode;
step two, calculating a weight map
Respectively calculating the contrast and the exposure of each multi-exposure lattice grid image shot in the step one, and multiplying the calculated contrast and the calculated exposure of each multi-exposure lattice grid image to obtain a weight map of each multi-exposure lattice grid image;
step three, tower-shaped decomposition of the weight map
Carrying out Gaussian pyramid decomposition on the weight graph of each multi-exposure lattice grid image calculated in the step two;
step four, multi-exposure lattice grid image tower-shaped decomposition
Performing Laplacian pyramid decomposition on each multi-exposure lattice grid image shot in the step one, wherein the number of decomposed layers is the same as that of the weight map Gaussian pyramid decomposed in the step three;
step five, weighting by Laplacian pyramid
Multiplying the image on each layer of the Laplacian pyramid of each multi-exposure lattice grid image decomposed in the fourth step with the image on the corresponding layer of the Gaussian pyramid of the weight map of the multi-exposure lattice grid image decomposed in the third step to obtain a weighted Laplacian pyramid;
step six, Laplacian pyramid fusion
Adding the images on the same layer of the Laplacian pyramid of the plurality of weighted multi-exposure lattice grid images obtained in the step five;
step seven, reconstructing the lattice grid image
Performing inverse tower shape transformation on the Laplacian pyramid fused in the step six, and reconstructing a new lattice grid image;
step eight, binarization of the lattice grid image
Carrying out local self-adaptive binarization processing on the reconstructed dot matrix grid image in the step seven to obtain a binary image;
step nine, sub-pixel boundary identification
Adopting an 8-connected domain rule to identify the whole pixel boundary of the dot matrix grid nodes in the binary image obtained in the step eight; on the basis, identifying the sub-pixel boundary of the dot matrix grid node by adopting a space moment method;
step ten, ellipse fitting
And (4) iteratively fitting the sub-pixel boundaries of the dot matrix grid nodes identified in the step nine by adopting a least square method to obtain the central coordinates of the ellipse, thereby obtaining the coordinates of the dot matrix grid nodes.
Further, the method for acquiring and identifying the lattice grid image in the metal sheet forming strain measurement further comprises the steps of adjusting the focal length before shooting the multi-exposure lattice grid image in the step one, so that the lattice grid on the surface of the metal sheet can be clearly imaged, and locking the focal length after the adjustment is finished.
Further, when shooting the multi-exposure lattice grid image in the first step, the camera is adjusted to a black and white shooting mode, the camera is kept stable, the exposure is adjusted, and 3 or more than 3 lattice grid images from underexposure, normal exposure to overexposure are shot.
Further, the method for calculating the weight map in the second step is as follows:
2.1) filtering each multi-exposure lattice grid image shot in the step one by adopting a Laplace operator, and taking an absolute value of a filtering result to obtain a contrast factor of each multi-exposure lattice grid image;
2.2) for each multi-exposure lattice grid image shot in the step one, the exposure factor is calculated by adopting the following formula:
wherein e represents a natural constant, k represents the kth multi-exposure lattice grid image, (I, j) represents the position of a pixel point, and Ik(i, j) represents the gray value of the pixel point of the k-th multi-exposure lattice grid image at the (i, j) position, Ek(i, j) represents an exposure factor of the k-th multi-exposure lattice grid image at the (i, j) position, and sigma represents a kernel function width;
2.3) on the basis of the contrast factor calculated in the step 2.1) and the exposure factor calculated in the step 2.2), calculating the weight factor of each multi-exposure lattice grid image by adopting the following formula:
wherein,representing the weight factor, C, of the kth multi-exposure lattice grid image at the (i, j) positionk(i, j) represents the contrast factor of the k-th multi-exposure lattice grid image calculated in the step 2.1) at the position of (i, j);
2.4) carrying out normalization processing on the weight factors of each multi-exposure lattice grid image calculated in the step 2.3) to obtain a weight map of each multi-exposure lattice grid image, wherein the normalization method comprises the following steps:
wherein, Wk(i, j) represents the gray value of the pixel point of the weight map of the kth multi-exposure lattice grid image at the position of (i, j), N represents the number of the multi-exposure lattice grid image shot in the step one, s represents the sth multi-exposure lattice grid image,representing the weighting factor of the s-th multi-exposure lattice grid image calculated in step 2.3) at the (i, j) position.
Further, the laplacian h adopted in the step 2.2) is:
further, the method for calculating the number L of the tower decomposition layers in the third step is as follows:
where r and c represent the number of rows and columns of the multi-exposure lattice grid image, respectively, ln () represents a natural logarithmic function with a natural constant e as a base, and min (r, c) represents a minimum function of the number of rows and columns of the multi-exposure lattice grid image.
Further, the local adaptive binarization method of the dot matrix grid image in the eight steps is as follows:
8.1) rapidly binarizing the lattice grid image reconstructed in the seventh step by adopting a window-by-window binarization method to obtain a binary image;
8.2) carrying out connected region marking on the binary image obtained in the step 8.1), and further adopting a pixel-by-pixel binarization method to carry out binarization again on all pixel points in the connected region for the connected region with the number of the pixel points being more than a given threshold value tau:
a) calculating a binarization threshold value for any pixel point in the communication area which is larger than a given threshold value tau:
wherein (i, j) represents the pixel position, T (i, j) represents the binarization threshold of the pixel at the (i, j) position in the communication region,representing the average gray value of pixel points in a u multiplied by v local window taking (I, j) as the center in the reconstructed lattice grid image in the step seven, wherein u and v respectively represent the length size and the height size of the window, m and n represent two integer variables, I'0(m, n) represents the reconstructed in step sevenThe gray value of the pixel point of the lattice grid image at the (m, n) position,representing the standard deviation value of the gray value distribution of the pixel points in the u x v local window taking (i, j) as the center in the reconstructed dot matrix grid image in the step seven, wherein gamma and offset are two constants;
b) re-binarizing the pixel points according to the binarization threshold value calculated in the step a):
wherein I ' (I, j) represents the gray value, I ', of the pixel point at the (I, j) position of the binary image obtained in step 8.1) after the binarization is carried out again '0And (i, j) represents the gray value of the pixel point of the lattice grid image reconstructed in the step seven at the position of (i, j).
Further, the value of the threshold τ in the step 8.2) is τ 10000.
Further, in each iteration process in the step ten, the distance from each boundary point to the fitted ellipse and the standard deviation value sigma are calculated, and for the boundary points with the distance larger than 3sigma, the next iteration operation is not involved.
Compared with the prior art, the method has the following advantages:
(1) the method can eliminate the influence of strong light reflection on the surface of the metal plate and acquire the high-quality dot matrix grid image;
(2) the dot matrix grid image acquired by the method is high in quality, so that the recognition rate of the dot matrix grid image is correspondingly improved;
(3) the dot matrix grid image acquired by the method is high in quality, so that the identification precision of the dot matrix grid image is correspondingly improved;
(4) the method improves the identification rate and the identification precision of the dot matrix grid image, thereby indirectly improving the reconstruction precision of the three-dimensional grid and the measurement precision of the three-dimensional strain.
Drawings
FIG. 1 is a flow chart of the specific operation steps of the present invention.
FIG. 2a is a photograph of an underexposed lattice grid image of a cupping test piece.
FIG. 2b shows a normal exposure lattice grid image of a cupping test piece.
FIG. 2c shows an overexposed dot matrix grid image of a cupping test piece.
FIG. 3 is a reconstructed dot matrix grid image of a cupping test piece.
FIG. 4 shows the result of detecting the under-exposure lattice grid image of a certain cupping test piece.
FIG. 5 shows the result of detecting the normal exposure lattice grid image of a cupping test piece.
FIG. 6 shows the result of detecting the over-exposure lattice grid image of a certain cupping test piece.
FIG. 7 shows a reconstructed dot matrix grid image detection result of a certain cupping test piece.
Detailed Description
The invention is further described in detail below with reference to the drawings and examples, but the invention is not limited thereto.
The invention provides a method for acquiring and identifying a dot matrix grid image in sheet metal forming strain measurement, which is shown in figure 1. When the lattice grid image of a certain visual angle metal plate surface is collected and identified, in the first step, a multi-exposure lattice grid image is shot. Before shooting, the focal length is adjusted firstly, so that the lattice grids on the surface of the metal plate can be imaged clearly, and the focal length is locked after the adjustment is finished. During shooting, in order to reduce the color impurity, the camera is adjusted to a black-and-white shooting mode, the camera is kept stable, the exposure amount is adjusted, and 3 or more dot matrix grid images from underexposure, normal exposure to overexposure are shot.
And secondly, calculating a weight map. And calculating a weight map of each multi-exposure lattice grid image shot in the first step. The weight graph calculation flow comprises the following steps:
1) a contrast factor is calculated. And filtering each multi-exposure lattice grid image shot in the first step by adopting a Laplacian operator h, and taking an absolute value of a filtering result to obtain a contrast factor of each multi-exposure lattice grid image. The laplacian h used is:
2) an exposure factor is calculated. For each multi-exposure lattice grid image shot in the first step, calculating an exposure factor by adopting the following formula:
wherein E represents a natural constant, k represents the kth multi-exposure lattice grid image, (i, j) represents the position of a pixel point, and Ek(I, j) denotes an exposure factor at the (I, j) position of the kth multi-exposure lattice grid image, IkAnd (i, j) represents the gray value of a pixel point of the k-th multi-exposure lattice grid image at the position (i, j), wherein sigma represents the width of the kernel function, the value range of sigma is more than 0 and less than or equal to 0.5, and sigma is generally selected to be 0.2.
3) A weighting factor is calculated. On the basis of the contrast factor calculated in the step 1) and the exposure factor calculated in the step 2), calculating the weight factor of each multi-exposure lattice grid image by adopting the following formula:
wherein,representing the weight factor, C, of the kth multi-exposure lattice grid image at the (i, j) positionk(i, j) represents the contrast factor of the k-th multi-exposure lattice grid image calculated in step 1) at the (i, j) position.
4) And normalizing the weight factor. Normalizing the weight factor of each multi-exposure lattice grid image calculated in the step 3) to obtain a weight map of each multi-exposure lattice grid image, wherein the normalization formula is as follows:
wherein, Wk(i, j) represents the gray value of the pixel point of the weight map of the kth multi-exposure lattice grid image at the position of (i, j), N represents the number of the multi-exposure lattice grid image shot in the first step, s represents the sth multi-exposure lattice grid image,representing the weight factor of the s-th multi-exposure lattice grid image calculated in the step 3) at the (i, j) position.
And thirdly, tower-shaped decomposition of the weight map. And carrying out Gaussian pyramid decomposition on the weight graph of each multi-exposure lattice grid image calculated in the second step. The decomposition process comprises the following steps:
1) and calculating the number of tower decomposition layers. The number L of tower decomposition layers is calculated as follows:
wherein r and c represent the number of rows and columns of the multi-exposure lattice grid image, respectively, ln () represents a natural logarithmic function with a natural constant e as a base, and min (r, c) represents a minimum function of the number of rows and columns of the multi-exposure lattice grid image.
2) And constructing a Gaussian pyramid 0 level of the weight map. Taking the weight map of each multi-exposure lattice grid image calculated in the second step as an image on the 0 th layer of the Gaussian pyramid of the weight map:
Gk,0(i,j)＝Wk(i,j)，
wherein k represents the k-th multi-exposure lattice grid image, (i, j) represents the pixel position, Gk,0(i, j) represents the gray value of the pixel point of the image on the 0 th layer of the Gaussian pyramid of the weight graph of the k-th multi-exposure lattice grid image at the (i, j) position, Wk(i, j) represents the gray value of the pixel point of the weight map of the kth multi-exposure lattice grid image calculated in the second step at the position of (i, j).
3) And constructing a t-th layer of the Gaussian pyramid of the weight map. Convolving the image on the t-1 th layer of the Gaussian pyramid of the weight map with a 5 multiplied by 5 window function, and then performing interlaced alternate downsampling on the convolution result to obtain the image on the t-th layer of the Gaussian pyramid of the weight map:
where t denotes the t-th layer of the pyramid, Gk,t(i, j) represents the gray value of the pixel point of the image on the t-th layer of the weight map Gaussian pyramid of the kth multi-exposure lattice grid image at the (i, j) position, m and n represent two integer variables, Gk,t-1(2i + m,2j + n) represents the image of the image on the t-1 layer of the Gaussian pyramid of the weight map of the kth multi-exposure lattice grid image at the (2i + m,2j + n) positionThe gray value of a pixel point, λ (m, n), represents a 5 × 5 window function:
on the basis of the 0 th layer of the weight map Gaussian pyramid of the kth multi-exposure lattice grid image constructed in the step 2), the 1 st layer of the weight map Gaussian pyramid of the kth multi-exposure lattice grid image can be constructed according to the construction method of the t th layer of the weight map Gaussian pyramid, and by analogy, the L th layer of the weight map Gaussian pyramid of the kth multi-exposure lattice grid image can be constructed.
And fourthly, decomposing the multi-exposure lattice grid image tower shape. And performing Laplacian pyramid decomposition on each multi-exposure lattice grid image shot in the first step. The decomposition process comprises the following steps:
1) and carrying out Gaussian pyramid decomposition on the multi-exposure lattice grid image. According to the weight map tower-shaped decomposition method in the third step, Gaussian pyramid decomposition is carried out on each multi-exposure lattice grid image shot in the first step, and the image on the t-th layer of the decomposed Gaussian pyramid is recorded as G'k,tWherein k represents the kth multi-exposure lattice grid image;
2) and constructing an L-th layer of the Laplacian pyramid. Taking the image on the L-th layer of the Gaussian pyramid decomposed in the step 1) as an image on the L-th layer of the Laplacian pyramid:
LPk,L(i,j)＝G′k,L(i,j)，
wherein, (i, j) represents the position of the pixel point, L represents the number of tower decomposition layers calculated in the third step, and LPk,L(i, j) represents the gray value, G ', of the pixel point of the image on the L-th layer of the Laplacian pyramid of the k-th multi-exposure lattice grid image at the position of (i, j)'k,L(i, j) representing the gray value of a pixel point of the image on the Lth layer of the k-th multi-exposure lattice grid image Gaussian pyramid decomposed in the step 1) at the position of (i, j);
3) and constructing a t-th Laplacian pyramid layer. The Laplacian pyramid t-th layer construction method is as follows:
a) interpolating and amplifying the image on the t +1 th layer of the Gaussian pyramid decomposed in the step 1):
wherein t +1 represents the t +1 th layer of the pyramid,representing the gray value of the pixel point at the (i, j) position of the interpolated and enlarged image,m and n represent two integer variables,representing the image on the t +1 th layer of the Gaussian pyramid of the k-th multi-exposure lattice grid image decomposed in the step 1)The gray value of the pixel point at the location,represents a transition variable, and λ (m, n) represents the 5 × 5 window function described in the third step;
b) subtracting the image on the t-th layer of the Gaussian pyramid decomposed in the step 1) from the image subjected to interpolation and amplification in the step a):
wherein, LPk,t(i, j) represents the gray value, G ', of the pixel point of the image on the t-th layer of the Laplacian pyramid of the k-th multi-exposure lattice grid image at the (i, j) position'k,t(i, j) representing the gray value of a pixel point of the image on the t-th layer of the k-th multi-exposure lattice grid image Gaussian pyramid decomposed in the step 1) at the (i, j) position;
according to the construction method of the t-th Laplacian pyramid layer, the L-1-0 layers of the Laplacian pyramid can be constructed in sequence.
And fifthly, weighting the Laplacian pyramid. And weighting the Laplacian pyramid of each multi-exposure lattice grid image decomposed in the fourth step by adopting the following method:
LP′k,t(i,j)＝Gk,t(i,j)×LPk,t(i,j),
wherein k represents the kth multi-exposure lattice grid image, t represents the t-th pyramid layer, and (i, j) represents the pixel position and LP'k,t(i, j) represents the gray value of the pixel point of the weighted kth multi-exposure lattice grid image on the t-th layer of the Laplacian pyramid at the (i, j) position, Gk,t(i, j) represents the gray value of the pixel point of the image on the t-th layer of the weight map Gaussian pyramid of the kth multi-exposure lattice grid image decomposed in the third step at the (i, j) position, LPk,t(i, j) represents the gray value of the pixel point of the image on the t-th layer of the laplacian pyramid of the kth multi-exposure lattice grid image decomposed in the fourth step at the (i, j) position, and the symbol x represents the multiplication of the gray values of two pixel points.
And sixthly, fusing the Laplacian pyramid. The laplacian pyramid fusion method is as follows:
wherein k represents the kth multi-exposure lattice grid image, t represents the t-th layer of the pyramid, (i, j) represents the pixel position, and N is shownShowing the number of the multi-exposure lattice grid image photographed at the first step,represents the gray value, LP ', of the pixel point at the (i, j) position of the fused image on the t-th layer of the Laplacian pyramid'k,tAnd (i, j) represents the gray value of the pixel point of the image on the t-th layer of the laplacian pyramid of the kth multi-exposure lattice grid image weighted in the fifth step at the position of (i, j).
And seventhly, reconstructing the lattice grid image. And (4) performing inverse tower shape transformation on the Laplacian pyramid fused in the sixth step to construct a new lattice grid image. The inverse turriform transformation includes the steps of:
1) and restoring the L-th layer of the Gaussian pyramid. Taking the image on the Laplacian pyramid Lth layer fused in the sixth step as an image on the Gaplacian pyramid Lth layer to be restored:
wherein (I, j) represents the pixel position, L represents the tower decomposition layer number calculated in the third step, and I'L(i, j) represents the gray value of the pixel point of the restored image on the L-th layer of the Gaussian pyramid at the (i, j) position,representing the gray value of the pixel point of the image on the L-th layer of the Laplacian pyramid fused in the sixth step at the (i, j) position;
2) and restoring the t-th layer of the Gaussian pyramid. The method for recovering the t-th layer of the Gaussian pyramid comprises the following steps:
a) interpolating and amplifying the image on the t +1 th layer of the Laplacian pyramid fused in the sixth step:
wherein t +1 represents the t +1 th layer of the pyramid,representing the gray values of the pixel points of the interpolated and magnified image at the (i, j) position, m and n representing two integer variables, λ (m, n) representing the 5 × 5 window function described in the third step,representing the image on the t +1 th layer of the Laplacian pyramid fused in the sixth stepThe gray value of the pixel point at the location,representing a transition variable;
b) adding the image interpolated and amplified in the step a) and the image on the t-th layer of the Laplacian pyramid fused in the sixth step:
wherein, I't(i, j) represents the gray value of the pixel point of the restored image on the t-th layer of the Gaussian pyramid at the (i, j) position,representing the gray value of the pixel point of the image on the t-th layer of the Laplacian pyramid fused in the sixth step at the (i, j) position;
according to the Gaussian pyramid t-layer recovery method, the L-1 to 0 layers of the Gaussian pyramid and the image I 'on the recovered 0 layer of the Gaussian pyramid can be sequentially recovered'0Is the lattice grid image to be reconstructed.
And eighthly, performing self-adaptive binarization on the lattice grid image. The self-adaptive binarization process comprises the following steps:
1) and (3) rapidly binarizing the dot matrix grid image reconstructed in the step seven by adopting a window-by-window binarization method described in the literature 'circulation grid patterned surface grid format for sheet metal forming' (Shi B and Liang J, OPT LASER ENG,2012,50(9):1186-1195), so as to obtain a binary image.
2) Carrying out connected region marking on the binary image obtained in the step 1), and further adopting a pixel-by-pixel binarization method to carry out binarization again on all pixel points in the connected region in the region of which the number of the pixel points is greater than a given threshold value tau:
a) calculating a binarization threshold value for any pixel point in the communication area which is larger than a given threshold value tau:
wherein (i, j) represents the pixel position, T (i, j) represents the binarization threshold of the pixel at the (i, j) position in the communication region,the average gray scale value of the pixel points in the u × v local window centered at (I, j) in the reconstructed lattice grid image in the seventh step is represented, m and n represent two integer variables, u and v represent the length and height of the window, respectively, and u ═ v ═ 23 and I'0(m, n) represents the gray value of the pixel point of the lattice grid image reconstructed in the step seven at the (m, n) position,representing the pixel points in the u × v local window with (i, j) as the center in the reconstructed lattice grid image in the seventh stepThe standard deviation value of the gray value distribution of (1) is two constants, γ and offset are generally selected to be 0.1, offset is 5.0, the threshold τ is a positive integer, and τ is generally 10000;
b) re-binarizing the pixel points according to the binarization threshold value calculated in the step a):
wherein I ' (I, j) represents the gray value, I ', of the pixel point at the (I, j) position of the binary image obtained in step 8.1) after the binarization is carried out again '0(i, j) represents the gray value of the pixel point at the (i, j) position of the reconstructed lattice grid image in the step seven.
And step nine, identifying the boundary of the sub-pixel. The sub-pixel boundary identification method comprises the following steps:
1) adopting 8 connected domain rules to identify the whole pixel boundary of the dot matrix grid nodes in the binary image obtained in the eighth step;
2) and (2) further identifying the sub-pixel boundary of the dot matrix grid node by adopting a space moment method on the basis of the whole pixel boundary of the dot matrix grid node identified in the step 1).
And step ten, ellipse fitting. For the sub-pixel boundaries of the lattice grid nodes in the binary image identified in the ninth step, the ellipse center coordinates are fitted by using the Least squares iteration method described in the document "Least squares fitting of circles and equations" (W.Gander, G.H. Golub and R.Strebel, BIT Numerical Mathematics, 1994, 34: 558-. In order to reduce noise interference, in each iteration process, the distance from each boundary point to the fitted ellipse and the standard deviation value sigma are calculated, and the boundary points with the distance larger than 3sigma do not participate in the next iteration operation. And finally, the fitted ellipse center coordinates are coordinates of the lattice grid nodes.
The invention is described below by combining with a specific simulation experiment, wherein the method of the invention realizes corresponding algorithms on VS2010 and opengl platforms and runs on a PC with 3.4GHz and 16GB memories of an Intel i7-4770 CPU.
The 3 shot multi-exposure lattice grid images of a certain cupping test piece are shown in fig. 2 a-2 c, wherein fig. 2a shows a shot under-exposure lattice grid image of the certain cupping test piece, fig. 2b shows a shot normal exposure lattice grid image of the certain cupping test piece, and fig. 2c shows a shot over-exposure lattice grid image of the certain cupping test piece. Fig. 3 shows a new lattice grid image reconstructed by the method according to the 3 shot multi-exposure lattice grid images of a cupping test piece. As can be seen by comparing fig. 2 a-2 c with fig. 3, the lattice grid in the reconstructed lattice grid image is clearer. Fig. 4 to 7 show the dot matrix grid node detection results of the under-exposed dot matrix grid image, the normal-exposed dot matrix grid image, the over-exposed dot matrix grid image and the reconstructed dot matrix grid image of a certain cupping test piece, and the dot matrix grid node identification rates are 49.6%, 77.78%, 72.3% and 86.87% in sequence. Wherein white cross lines represent the coordinates of the detected lattice grid nodes. As can be seen from comparing fig. 4 to fig. 7, the number of lattice grid node detections in the reconstructed lattice grid image is the largest, and the lattice grid node recognition rate is the highest, which is 86.87%. The embodiment also shows that the method can eliminate the influence of strong light reflection on the surface of the metal plate, improve the collection quality of the dot matrix grid image and improve the recognition rate of the dot matrix grid image.
The embodiments of the present invention have been described above with reference to the drawings and examples, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the spirit of the present invention.
Claims (9)
1. A method for acquiring and identifying a lattice grid image in sheet metal forming strain measurement is characterized by comprising the following steps:
step one, shooting a multi-exposure lattice grid image
Keeping a camera still, and shooting a plurality of dot matrix grid images with different exposure quantities in a black-and-white shooting mode;
step two, calculating a weight map
Respectively calculating the contrast and the exposure of each multi-exposure lattice grid image shot in the step one, and multiplying the calculated contrast and the calculated exposure of each multi-exposure lattice grid image to obtain a weight map of each multi-exposure lattice grid image;
step three, tower-shaped decomposition of the weight map
Carrying out Gaussian pyramid decomposition on the weight graph of each multi-exposure lattice grid image calculated in the step two;
step four, multi-exposure lattice grid image tower-shaped decomposition
Performing Laplacian pyramid decomposition on each multi-exposure lattice grid image shot in the step one, wherein the number of decomposed layers is the same as that of the weight map Gaussian pyramid decomposed in the step three;
step five, weighting by Laplacian pyramid
Multiplying the image on each layer of the Laplacian pyramid of each multi-exposure lattice grid image decomposed in the fourth step with the image on the corresponding layer of the Gaussian pyramid of the weight map of the multi-exposure lattice grid image decomposed in the third step to obtain a weighted Laplacian pyramid;
step six, Laplacian pyramid fusion
Adding the images on the same layer of the Laplacian pyramid of the plurality of weighted multi-exposure lattice grid images obtained in the step five;
step seven, reconstructing the lattice grid image
Performing inverse tower shape transformation on the Laplacian pyramid fused in the step six, and reconstructing a new lattice grid image;
step eight, binarization of the lattice grid image
Carrying out local self-adaptive binarization processing on the reconstructed dot matrix grid image in the step seven to obtain a binary image;
step nine, sub-pixel boundary identification
Adopting an 8-connected domain rule to identify the whole pixel boundary of the dot matrix grid nodes in the binary image obtained in the step eight; on the basis, identifying the sub-pixel boundary of the dot matrix grid node by adopting a space moment method;
step ten, ellipse fitting
And (4) iteratively fitting the sub-pixel boundaries of the dot matrix grid nodes identified in the step nine by adopting a least square method to obtain the central coordinates of the ellipse, thereby obtaining the coordinates of the dot matrix grid nodes.
2. The method for acquiring and identifying the lattice grid image in the metal sheet forming strain measurement according to claim 1, further comprising adjusting the focal length before the step one of taking the multi-exposure lattice grid image so that the lattice grid on the surface of the metal sheet can be clearly imaged, and locking the focal length after the adjustment is completed.
3. The method for acquiring and identifying the lattice grid images in the metal sheet forming strain measurement according to claim 1, wherein when the multi-exposure lattice grid image is shot in the first step, the camera is adjusted to a black and white shooting mode, the camera is kept stable, the exposure is adjusted, and 3 or more than 3 lattice grid images from underexposure, normal exposure to overexposure are shot.
4. The method for acquiring and identifying the lattice grid image in the sheet metal forming strain measurement according to claim 1, wherein the method for calculating the weight map in the second step is as follows:
2.1) filtering each multi-exposure lattice grid image shot in the step one by adopting a Laplace operator, and taking an absolute value of a filtering result to obtain a contrast factor of each multi-exposure lattice grid image;
2.2) for each multi-exposure lattice grid image shot in the step one, the exposure factor is calculated by adopting the following formula:
wherein e represents a natural constant, k represents the kth multi-exposure lattice grid image, (I, j) represents the position of a pixel point, and Ik(i, j) represents the gray value of the pixel point of the k-th multi-exposure lattice grid image at the (i, j) position, Ek(i, j) represents an exposure factor of the k-th multi-exposure lattice grid image at the (i, j) position, and sigma represents a kernel function width;
2.3) on the basis of the contrast factor calculated in the step 2.1) and the exposure factor calculated in the step 2.2), calculating the weight factor of each multi-exposure lattice grid image by adopting the following formula:
wherein,representing the weight factor, C, of the kth multi-exposure lattice grid image at the (i, j) positionk(i, j) represents the contrast factor of the k-th multi-exposure lattice grid image calculated in the step 2.1) at the position of (i, j);
2.4) carrying out normalization processing on the weight factors of each multi-exposure lattice grid image calculated in the step 2.3) to obtain a weight map of each multi-exposure lattice grid image, wherein the normalization method comprises the following steps:
wherein, Wk(i, j) represents the gray value of the pixel point of the weight map of the kth multi-exposure lattice grid image at the position of (i, j), N represents the number of the multi-exposure lattice grid image shot in the step one, s represents the sth multi-exposure lattice grid image,representing the weighting factor of the s-th multi-exposure lattice grid image calculated in step 2.3) at the (i, j) position.
5. The method for acquiring and identifying the lattice grid image in the sheet metal forming strain measurement according to claim 4, wherein the Laplace operator h adopted in the step 2.2) is:
6. the method for acquiring and identifying the lattice grid image in the metal sheet forming strain measurement according to claim 1, wherein the number L of the tower-shaped decomposition layers in the third step is calculated as follows:
where r and c represent the number of rows and columns of the multi-exposure lattice grid image, respectively, ln () represents a natural logarithmic function with a natural constant e as a base, and min (r, c) represents a minimum function of the number of rows and columns of the multi-exposure lattice grid image.
7. The method for acquiring and identifying the lattice grid image in the metal sheet forming strain measurement according to claim 1, wherein the eight-step lattice grid image local adaptive binarization method is as follows:
8.1) rapidly binarizing the lattice grid image reconstructed in the seventh step by adopting a window-by-window binarization method to obtain a binary image;
8.2) carrying out connected region marking on the binary image obtained in the step 8.1), and further adopting a pixel-by-pixel binarization method to carry out binarization again on all pixel points in the connected region for the connected region with the number of the pixel points being more than a given threshold value tau:
a) calculating a binarization threshold value for any pixel point in the communication area which is larger than a given threshold value tau:
wherein (i, j) represents a pixel point positionT (i, j) represents the binarization threshold of the pixel point at the (i, j) position in the communication area,representing the average gray value of pixel points in a u multiplied by v local window taking (I, j) as the center in the reconstructed lattice grid image in the step seven, wherein u and v respectively represent the length size and the height size of the window, m and n represent two integer variables, I'0(m, n) represents the gray value of the pixel point of the lattice grid image reconstructed in the step seven at the (m, n) position,representing the standard deviation value of the gray value distribution of the pixel points in the u x v local window taking (i, j) as the center in the reconstructed dot matrix grid image in the step seven, wherein gamma and offset are two constants;
b) re-binarizing the pixel points according to the binarization threshold value calculated in the step a):
wherein I ' (I, j) represents the gray value, I ', of the pixel point at the (I, j) position of the binary image obtained in step 8.1) after the binarization is carried out again '0And (i, j) represents the gray value of the pixel point of the lattice grid image reconstructed in the step seven at the position of (i, j).
8. The method for acquiring and identifying the lattice grid image in the metal sheet forming strain measurement according to claim 7, wherein the value of the threshold τ in the step 8.2) is τ 10000.
9. The method for acquiring and identifying the lattice grid images in the metal sheet forming strain measurement according to claim 1, wherein in each iteration process in the step ten, the distance between each boundary point and the fitted ellipse and the standard deviation value sigma are calculated, and the boundary points with the distance larger than 3sigma do not participate in the next iteration operation.
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 true CN107726990A (en) | 2018-02-23 |
| CN107726990B 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 (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108507476A (en) * | 2018-04-27 | 2018-09-07 | 中国石油大学(北京) | Displacement field measurement 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 |
| CN111832433A (en) * | 2020-06-24 | 2020-10-27 | 科络克电子科技（上海）有限公司 | Device for extracting object characteristics from image and working method thereof |
| CN112935013A (en) * | 2021-02-10 | 2021-06-11 | 哈尔滨工业大学 | Method for measuring shear strain of viscous medium on plate blank |
| CN113112470A (en) * | 2021-04-08 | 2021-07-13 | 西安道法数器信息科技有限公司 | Cold rolling process steel bending detection and analysis method based on artificial intelligence |
| CN113155648A (en) * | 2021-05-18 | 2021-07-23 | 苏州大学 | Material microscopic deformation measuring method and system based on impact test |
| CN116297336A (en) * | 2023-03-27 | 2023-06-23 | 佛山市顺德区豪锐五金制品有限公司 | Method for detecting defect points on surface of metal sheet for outer layer protection |
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 (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108507476A (en) * | 2018-04-27 | 2018-09-07 | 中国石油大学(北京) | Displacement field measurement method, device, equipment and storage medium for material surface |
| 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 |
| CN111832433A (en) * | 2020-06-24 | 2020-10-27 | 科络克电子科技（上海）有限公司 | Device for extracting object characteristics from image and working method thereof |
| 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 |
| CN112935013A (en) * | 2021-02-10 | 2021-06-11 | 哈尔滨工业大学 | Method for measuring shear strain of viscous medium on plate blank |
| 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 |
| CN113112470A (en) * | 2021-04-08 | 2021-07-13 | 西安道法数器信息科技有限公司 | Cold rolling process steel bending detection and analysis method based on artificial intelligence |
| 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 |
| CN113155648A (en) * | 2021-05-18 | 2021-07-23 | 苏州大学 | Material microscopic deformation measuring method and system based on impact test |
| CN116297336A (en) * | 2023-03-27 | 2023-06-23 | 佛山市顺德区豪锐五金制品有限公司 | Method for detecting defect points on surface of metal sheet for outer layer protection |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107726990B (en) | 2019-06-21 |
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 | |
| CN108510504B (en) | Image segmentation method and device | |
| CN108510451B (en) | Method for reconstructing license plate based on double-layer convolutional neural network | |
| CN107507135B (en) | Image reconstruction method based on coding aperture and target | |
| EP3438929B1 (en) | Foreground and background detection method | |
| CN112686935A (en) | Airborne depth sounding radar and multispectral satellite image registration method based on feature fusion | |
| CN105405138B (en) | Waterborne target tracking based on conspicuousness detection | |
| CN112560619B (en) | Multi-focus image fusion-based multi-distance bird accurate identification method | |
| Yilmaz et al. | Metaheuristic pansharpening based on symbiotic organisms search optimization | |
| CN116664451B (en) | Measurement robot measurement optimization method based on multi-image processing | |
| CN114581318B (en) | Low-illumination image enhancement method and system | |
| CN112381742A (en) | Single image motion blur removing method and system | |
| CN115170416A (en) | Multi-exposure dynamic range enhancement method for low-light-level image | |
| CN111260735B (en) | External parameter calibration method for single-shot LIDAR and panoramic camera | |
| CN116758266A (en) | Reading method of pointer type instrument | |
| Tadic et al. | Edge-preserving Filtering and Fuzzy Image Enhancement in Depth Images Captured by Realsense Cameras in Robotic Applications. | |
| Jung et al. | Multispectral fusion of rgb and nir images using weighted least squares and convolution neural networks | |
| CN116029954A (en) | Image fusion method and device | |
| CN115761241A (en) | Image enhancement method and application thereof | |
| CN115511765A (en) | Multi-exposure image rapid fusion method based on double-scale decomposition | |
| Yi et al. | Attention-model guided image enhancement for robotic vision applications | |
| CN112116561B (en) | Power grid transmission line detection method and device based on image processing fusion network weight | |
| CN109345539B (en) | Self-adaptive Mean-Shift standing tree image segmentation method based on image abstraction | |
| Nasonov et al. | Image sharpening with blur map estimation using convolutional neural network |
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 | ||
| 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 |