CN108662977A - A kind of refractory brick geometric dimension measurement method - Google Patents

A kind of refractory brick geometric dimension measurement method Download PDF

Info

Publication number
CN108662977A
CN108662977A CN201810213913.XA CN201810213913A CN108662977A CN 108662977 A CN108662977 A CN 108662977A CN 201810213913 A CN201810213913 A CN 201810213913A CN 108662977 A CN108662977 A CN 108662977A
Authority
CN
China
Prior art keywords
refractory brick
image
noise reduction
geometric dimension
threshold
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.)
Withdrawn
Application number
CN201810213913.XA
Other languages
Chinese (zh)
Inventor
曹衍龙
张远松
杨将新
王敬
孙安顺
董献瑞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shandong Industrial Technology Research Institute of ZJU
Original Assignee
Shandong Industrial Technology Research Institute of ZJU
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shandong Industrial Technology Research Institute of ZJU filed Critical Shandong Industrial Technology Research Institute of ZJU
Publication of CN108662977A publication Critical patent/CN108662977A/en
Withdrawn legal-status Critical Current

Links

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
    • 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
    • G01B11/26—Measuring arrangements characterised by the use of optical techniques for measuring angles or tapers; for testing the alignment of axes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/12—Edge-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
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30132—Masonry; Concrete

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Geometry (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a kind of refractory brick geometric dimension measurement methods, including to original refractory brick Algorithm for Color Image Filtering noise reduction, obtain image after noise reduction；After noise reduction in image, it is partitioned into refractory brick region, obtains refractory brick image；Morphological scale-space is carried out to refractory brick image, so that refractory brick image border is seamlessly transitted, and remove the isolated spot noise at refractory brick image border, refractory brick image after being corrected；Edge detection is carried out to refractory brick image after amendment with Canny operators；Refractory brick shape straight line is extracted using Hough transform method and K Mean Methods, forms the outer profile of refractory brick.The present invention provides the refractory brick geometric dimension measurement methods that a kind of efficiency and precision improve.

Description

A kind of refractory brick geometric dimension measurement method
Technical field
The present invention relates to a kind of refractory brick geometric dimension measurement methods.
Background technology
Refractory brick is a kind of sizing refractory material for being in refractory clay or the firing of other refractory raw materials, is mainly used for building Smelting furnace or steel ladle are resistant to 1580 DEG C -1770 DEG C of high temperature.On the production line of refractory brick, before the offline vanning of product, All it is by manually using tape measure manual measurement, measurement result accuracy is not high, is susceptible to erroneous judgement, if individually for a long time Defective products is mixed into finished product by the gross, serious financial consequences can be brought to factory, or even seriously affect the production of steel.
Invention content
The object of the present invention is to provide the refractory brick geometric dimension measurement methods that a kind of efficiency and precision improve.
The technical solution adopted by the present invention to solve the technical problems is：
A kind of refractory brick geometric dimension measurement method, includes the following steps：
Step 1：To original refractory brick Algorithm for Color Image Filtering noise reduction, image after noise reduction is obtained；
Step 2：After noise reduction in image, it is partitioned into refractory brick region, obtains refractory brick image；
Step 3：Morphological scale-space is carried out to refractory brick image, so that refractory brick image border is seamlessly transitted, and remove fire resisting Isolated spot noise at brick image border, refractory brick image after being corrected；
Step 4：Edge detection is carried out to refractory brick image after amendment with Canny operators；
Step 5：Refractory brick shape straight line is extracted using Hough transform method and K Mean Methods, forms the outer of refractory brick Profile includes the following steps：
Step 5-1：The refractory brick shape straight-line detection after edge detection is fitted using Hough transform method a plurality of Straight line indicates straight line using polar coordinate system, then has：
Xcos θ+ysin θ=ρ (3-23)
ρ θ parameter spaces are divided into summing elements, wherein (ρmin,ρmax) and (θmin,θmax) parameter area is：-90°≤ θ≤90 ° and-D≤ρ≤D, wherein D is the diagonal maximum distance of image；
Step 5-2：Given sample set D={ x1,x2,…,xm, k Mean Methods are for cluster gained cluster C={ C1, C2,…,CKMinimum square error is divided,
Wherein, k is cluster number of clusters；CiFor cluster；It is cluster CiMean vector, 1≤i≤k；Between E is Gauge from.
Step 5-3：In conjunction with Hough transform method and K Mean Methods, randomly selects intersection point from multigroup intersection point and be defined as gathering Straight line is extracted on same side by class center, until the four edges of extraction refractory brick come out；
Step 6：Refractory brick shape straight-line intersection is sought, distance between two points are the shape geometric dimension of refractory brick.
Further, it is using bilateral filtering noise reduction, the calculation formula of bilateral filtering in step 1：
Wherein, k is used for carrying out result unitization, and c is the Gauss weight based on space length, and s is based on phase between pixel Like the Gauss weight of degree, f (ε) is current pixel.
Further, go out refractory brick region using gray processing and Ostu binarization segmentations in step 2, include the following steps：
Step 2-1：It is averaging the RGB three-component brightness in refractory brick coloured image to obtain gray value f (x, y),It is worth to refractory brick grey level histogram according to gray scale, wherein R (x, y) is Red channel components, G (x, y) are green channel components, and B (x, y) is blue channel component；
Step 2-2：According to refractory brick grey level histogram, the gradation intervals of refractory brick grey level histogram are divided into three classes, root Optimal threshold is obtained according to inter-class variance：
Wherein,For inter-class variance, k1,k2For preset threshold value,For optimal threshold.
Step 2-3：Refractory brick gray-scale map is divided according to optimal threshold.
Further, in step 3, Morphological scale-space includes carrying out opening operation and closed operation to refractory brick image, carries out out fortune It calculates removal and isolates spot noise；Carrying out closed operation keeps refractory brick image boundary smooth.
Further, the method for edge detection comprises the steps of in step 4：
Step 4-1：With Gaussian smoothing function to refractory brick image filtering after amendment, it is f to enable refractory brick image after correcting (x, y), Gaussian function are G (x, y), it is smooth after image be fs(x, y), thenfs(x, y)=G (x, y) * f(x,y)；
Step 4-2：Gradient magnitude image M and angle magnitude image θ is calculated, Its In, GxIndicate the gradient magnitude component of horizontal direction, GyIndicate the gradient magnitude component of vertical direction；
Step 4-3：Non-maxima suppression is carried out to gradient magnitude image M：In gradient direction, to a point, field Center pixel f (x, y) is compared with along the two of gradient line adjacent pixel, if the Grad of f (x, y) is less than or equal to along ladder The Grad for spending two adjacent pixels of line, then enable f (x, y)=0；
Step 4-4：Edge is detected and connected using dual threshold, is to mend with Low threshold image based on high threshold image It fills to connect image border.
Advantage is the present invention compared with prior art：
1, it uses filtering noise reduction, Threshold segmentation to obtain refractory brick region, Morphological scale-space to isolate to eliminate refractory brick image Noise, smooth boundary, then to refractory brick shape straight-line detection, refractory brick appearance profile can be accurately obtained.
2, statistical method i.e. K mean values machine learning method is introduced, is combined to obtain with traditional Hough transform method resistance to Firebrick shape geometric dimension, precision and computational efficiency significantly improve.
3, best to the preservation of refractory brick edge details using bilateral filtering, and noise reduction efficacy is high.
4, to determine that object under test region is not only advantageous to accurately analyze using gray processing and Ostu (big Tianjin method) to be measured The appearance and size for measuring object, is also beneficial to downscaled images process range, improvement method efficiency.
5, using Canny operators to edge detection, with low error rate, marginal point is positioned well and single side Edge point responds advantage.
Description of the drawings
Fig. 1 is that refractory brick is split to the refractory brick image to be formed from background.
Fig. 2 is the image carried out to refractory brick image before and after morphological operation, and (2a) is grasped to refractory brick morphological image Image before work, (2b) are the images after being operated to refractory brick morphological image.
Fig. 3 is the schematic diagram after the fitting of refractory brick image border after correcting.
Fig. 4 is a plurality of straight line fitted by Hough transform method.
Fig. 5 is refractory brick appearance and size.
Fig. 6 is refractory brick cut extraction result figure.
Fig. 7 is the schematic diagram that deep defects measurement obtains.
Fig. 8 is image capture module schematic diagram.
Fig. 9 is refractory brick measuring device schematic diagram.
Figure 10 is refractory brick measuring system flow chart.
Specific implementation mode
The present invention is further described with reference to the accompanying drawings and detailed description.
Embodiment 1
As shown in Figs. 1-5, a kind of refractory brick geometric dimension measurement method, includes the following steps：
Step 1：To original refractory brick Algorithm for Color Image Filtering noise reduction, image after noise reduction is obtained；
Step 2：After noise reduction in image, it is partitioned into refractory brick region, obtains refractory brick image；
Step 3：Morphological scale-space is carried out to refractory brick image, so that refractory brick image border is seamlessly transitted, and remove fire resisting Isolated spot noise at brick image border, refractory brick image after being corrected；
Step 4：Edge detection is carried out to refractory brick image after amendment with Canny operators；
Step 5：Refractory brick shape straight line is extracted using Hough transform method and K Mean Methods, forms the outer of refractory brick Profile includes the following steps：
Step 5-1：Consider x/y plane on a point (x, y) and slope-intercept form expression formula for y=ax+b straight line, Straight line is indicated using polar coordinate system, then is had：
Xcos θ+ysin θ=ρ (3-23)
Every sine curve is indicated through a particular point (x in x/y planek,yk) cluster straight line.Hough transformation is by ρ θ Parameter space is divided into so-called summing elements, wherein (ρmin,ρmax) and (θmin,θmax) it is desired parameter area, -90 ° ≤ θ≤90 ° and-D≤ρ≤D, wherein D is the diagonal maximum distance of image.Unit at coordinate (i, j) has accumulated value A (i, j), it corresponds to and parameter space coordinate (ρi,θj) associated square.Start these units being initialized as 0. Then, for the non-background dot (x of each of x/y planek,yk), it enables θ be equal to the subdivision value each allowed on θ axis, uses simultaneously Equation ρ=xkcosθ+ykSin θ releases corresponding ρ.It rounds up to ρ, obtains the immediate permission cell value along axis. If selecting a θpThe solution ρ being worth toq, then A (p, q)=A (p, q)+1 is enabled.After this process, the value P in A (i, j) will anticipate Taste, which in x/y plane, has P point to be located at straight line xcos θj+ysinθj=ρjOn.Subdivision quantity in ρ θ planes determines these points Synteny precision.
Step 5-2：Given sample set D={ x1,x2,…,xm, k Mean Methods are for cluster gained cluster C={ C1, C2,…,CKDivide minimum square error
Wherein, k is cluster number of clusters；CiFor cluster；It is cluster CiMean vector, 1≤i≤k；E is interval Distance.Sample is around the tightness degree of cluster mean vector in cluster to a certain extent, and E values are smaller, then Sample Similarity is got in cluster It is high.
Step 5-3：In conjunction with Hough transform method and K Mean Methods, randomly selects intersection point from multigroup intersection point and be defined as gathering Class center comes out same side as a type extraction straight line, until the four edges of extraction refractory brick, four edges The as appearance and size of refractory brick, as shown in Fig. 5.
Step 6：Refractory brick shape straight-line intersection is sought, distance between two points are the shape geometric dimension of refractory brick.
As shown in Figure 1, using bilateral filtering noise reduction in step 1, the calculation formula of bilateral filtering is：
Wherein, k is used for carrying out result unitization, and c is the Gauss weight based on space length, and s is based on phase between pixel Like the Gauss weight of degree, f (ε) is current pixel.
Go out refractory brick region using gray processing and Ostu binarization segmentations in step 2, includes the following steps：
Step 2-1：It is averaging the RGB three-component brightness in refractory brick coloured image to obtain gray value f (x, y),It is worth to refractory brick grey level histogram according to gray scale, wherein R (x, y) is Red channel components, G (x, y) are green channel components, and B (x, y) is blue channel component；
Step 2-2：According to refractory brick grey level histogram, the gradation intervals of refractory brick grey level histogram are divided into three classes, by 2 threshold values separate, and optimal threshold is obtained according to inter-class variance：
In formula,
And there is following relationship to set up：
P1m1+P2m2+P3m3=mG (3-10)
P1+P2+P3=1 (3-11)
Wherein,For inter-class variance, k1,k2For preset threshold value,For optimal threshold.
Step 2-3：Refractory brick gray-scale map is split according to optimal threshold, obtains refractory brick image, as shown in Figure 1.
In step 3, Morphological scale-space includes carrying out opening operation and closed operation to refractory brick image, carries out opening operation removal Isolated spot noise；Carrying out closed operation keeps refractory brick image boundary smooth, as shown in Figure 2.
The method of edge detection comprises the steps of in step 4：
Step 4-1：With Gaussian smoothing function to refractory brick image filtering after amendment, it is f to enable refractory brick image after correcting (x, y), Gaussian function are G (x, y), it is smooth after image be fs(x, y), thenfs(x, y)=G (x, y) * f(x,y)；
Step 4-2：Gradient magnitude image M and angle magnitude image θ is calculated, Its In, GxIndicate the gradient magnitude component of horizontal direction, GyIndicate the gradient magnitude component of vertical direction；
Step 4-3：Non-maxima suppression is carried out to gradient magnitude image M：In gradient direction, to a point, field Center pixel f (x, y) is compared with along the two of gradient line pixel, if the Grad of f (x, y) is less than or equal to along gradient line Two adjacent pixels Grad, then enable f (x, y)=0；
Step 4-4：It is handled with dual threshold and is detected with linking parsing and connect edge：Based on high threshold image, with Low threshold image is to supplement to connect the edge of image, referring to Fig. 3.
Embodiment 2
The present embodiment is progress refractory brick surface scratch identification on the basis of to refractory brick image into row threshold division, The acquisition of refractory brick image filters noise reduction process and to be partitioned into refractory brick region same as Example 1.
Based on the refractory brick surface scratch recognition methods of frequency filtering enhancing, include the following steps：
Step 1：Noise reduction is filtered to the original refractory brick coloured image collected by sensor, obtains noise reduction Image afterwards；After noise reduction in image, it is partitioned into refractory brick region, obtains refractory brick gray level image；
Step 2：, refractory brick gray level image is converted to refractory brick coloured image, to refractory brick coloured image into row of channels It decomposes, isolates the image in tri- channels R, G, B, obtain defect channel image, as shown in Figure 1；
Step 3：Two dimensional discrete Fourier transform is carried out to defect channel image, obtains the frequency domain of defect channel image Image；Convolution operation is carried out to frequency area image using the bandpass filter of sinusoidal shape, after obtaining frequency domain image filtering Image；Inverse Fourier transform is carried out to filtered refractory brick image, obtains the refractory brick image after inverse transformation；To inverse transformation Refractory brick image afterwards carries out threshold process and Morphological scale-space, obtains refractory brick threshold binary image；
Step 4：Two-pass scan method is used to refractory brick threshold binary image, marks institute present in refractory brick threshold binary image There is connected region；Judged and screened according to the feature of different connected regions, to identify scored area, as shown in Figure 6. Connected region refers to the image-region with the adjacent foreground pixel point composition of same pixel value and position in image.
It is using bilateral filtering noise reduction, the calculation formula of bilateral filtering in step 1：
Wherein, k is used for carrying out result unitization, and c is the Gauss weight based on space length, and s is based on phase between pixel Like the Gauss weight of degree, f (ε) is current pixel.
Go out refractory brick region, including following step using gray processing and Ostu (Otsu algorithm) binarization segmentation in step 1 Suddenly：
Step 1-1：It is averaging the RGB three-component brightness in refractory brick coloured image to obtain gray value f (x, y),It is worth to refractory brick grey level histogram according to gray scale, wherein R (x, y) is Red channel components, G (x, y) are green channel components, and B (x, y) is blue channel component；
Step 1-2：According to refractory brick grey level histogram, the gradation intervals of refractory brick grey level histogram are divided into three classes, root Optimal threshold is obtained according to inter-class variance：
In formula,
And there is following relationship to set up：
P1m1+P2m2+P3m3=mG (3-10)
P1+P2+P3=1 (3-11)
Obtain optimal threshold：
Wherein,For inter-class variance, k1,k2For preset threshold value,For optimal threshold；
Step 1-3：Refractory brick gray-scale map is split according to optimal threshold, obtains refractory brick area image.
In step 2, according to the following formula to refractory brick coloured image carry out channel decomposition f (x, y)=0.3f (x, y, R)+ 0.59f(x,y,G)+0.11f(x,y,B)；
In step 3, two dimensional discrete Fourier transform expression formula is as follows：
Wherein, f (x, y) is the digital picture that size is M × N, and F (u, v) is frequency domain as a result, x, y are spatial domain change Amount, u, v are frequency domain variable,
X=0,1,2 ..., M-1, y=0,1,2 ..., N-1, u=0,1,2 ..., M-1, v=0,1,2 ..., N-1.
In step 3, convolution operation is carried out to frequency domain rate image f (x, y) using function h (x, y), expression formula is：
Wherein, x=0,1,2 ..., M-1, y=0,1,2 ..., N-1；
In step 3, two-dimensional discrete Fourier inverse transformation expression formula is as follows：
Wherein, f (x, y) is the digital picture that size is M × N, and F (u, v) is frequency domain as a result, x, y are spatial domain change Amount, u, v are frequency domain variable,
X=0,1,2 ..., M-1, y=0,1,2 ..., N-1, u=0,1,2 ..., M-1, v=0,1,2 ..., N-1.
In step 4, connected region is obtained according to two-pass scan method, is included the following steps：
Step 4-1：First pass is carried out to refractory brick threshold binary image, assigns each location of pixels one label, is scanned The pixel set in the same connected region is endowed one or more different labels in the process, and merging belongs to the same connection Region but the label with different value；
Step 4-2：Second time scanning is carried out to refractory brick threshold binary image, the same label with relation of equality is marked The pixel of note is classified as a connected region and assigns an identical label.
Embodiment 3
The present embodiment is progress refractory brick deep defects identification on the basis of to refractory brick image into row threshold division, The acquisition of refractory brick image filters noise reduction process and to be partitioned into refractory brick region same as Example 1.
The recognition methods of refractory brick deep defects based on height histogram divion, includes the following steps：
Step 1 is filtered the original refractory brick coloured image collected by sensor noise reduction, obtains noise reduction Image afterwards；After noise reduction in image, it is partitioned into refractory brick region, obtains refractory brick gray level image, as shown in Figure 1；
Step 2 carries out plane fitting acquisition zero plane to refractory brick picture point cloud using least square method, and obtains original The dimensional parameters (height and width) of beginning refractory brick image are generated according to the dimensional parameters and zero plane of original refractory brick image Corresponding datum plane image；
Step 3 to original refractory brick image and datum plane image make poor, the point cloud number after acquisition slant correction According to figure；
Step 4 is filtered segmentation to the height histogram of the point cloud after slant correction, obtains the point cloud of set depth Information carries out connected component labeling to the point cloud information of set depth according to two-pass scan method, calculates the connection under each depth Region area obtains defective data, as shown in Figure 7.Connected region refer in image have same pixel value and position it is adjacent The image-region of foreground pixel point composition.
It is using bilateral filtering noise reduction, the calculation formula of bilateral filtering in step 1：
Wherein, k is used for carrying out result unitization, and c is the Gauss weight based on space length, and s is based on phase between pixel Like the Gauss weight of degree, f (ε) is current pixel.
Go out refractory brick region using gray processing and Ostu binarization segmentations in step 1, includes the following steps：
Step 1-1：It is averaging the RGB three-component brightness in refractory brick coloured image to obtain gray value f (x, y),It is worth to refractory brick grey level histogram according to gray scale, wherein R (x, y) is Red channel components, G (x, y) are green channel components, and B (x, y) is blue channel component；
Step 1-2：According to refractory brick grey level histogram, the gradation intervals of refractory brick grey level histogram are divided into three classes, root According to formulaFind optimal threshold, whereinFor inter-class variance, k1,k2It is default Threshold value,For optimal threshold；
Step 1-3：Refractory brick gray-scale map is split according to optimal threshold, obtains refractory brick area grayscale image.
In step 2, the fitting parameter α of zero plane, beta, gamma are obtained：
For a width 2D consecutive image f (x, y) (>=0), p+q rank squares mpqIt is defined as：
Wherein, p, q are non-negative integers, and for discretization digital picture, above formula is：
Wherein, (r0,c0) it is center-of-mass coordinate, and
Single order plane approximation method is described by following formula：
Image (r, c)=α (r-r0)+β(c-c0)+γ (4-13)
Wherein, r0To wait for the abscissa of fitted area, c0To wait for that the ordinate of fitted area, γ are to wait for putting down for fitted area Equal gray scale, F are the area of entire plane, and MRow is the Gray Moment along line direction, and MCol is the Gray Moment along column direction, Then have：
MRow=sum ((r-r0)*(Image(r,c)-γ))/F2 (4-14)
MCol=sum ((c-r0)*(Image(r,c)-γ))/F2 (4-15)
In step 2, the generation method of datum plane image is：According to fitting parameter α, beta, gamma, in conjunction with original refractory brick point The dimension information of cloud atlas Image (r, c) generates datum plane image Image (r, c)0。
In step 2, to original refractory brick point cloud chart Image (r, c) and datum plane image Image (r, c)0Made Difference obtains point cloud data figure Image'(r, c after slant correction),
Image'(r, c)=Image (r, c)-Image (r, c)0。
In step 4, segmentation is filtered to the point cloud level degree histogram after slant correction using height bandpass filter.
In step 4, connected region is obtained according to two-pass scan method, is included the following steps：
Step 4-1：First pass is carried out to the point cloud information of set depth, assigns each location of pixels one label, Pixel set in scanning process in the same connected region is endowed one or more different labels, and merging belongs to same Connected region but the label with different value；
Step 4-2：Second time scanning is carried out to the point cloud information of constant depth, by the same label with relation of equality The pixel marked is classified as a connected region and assigns an identical label.
Embodiment 4
The present embodiment is to carry out refractory brick surface inclination angle on the basis of to refractory brick image into row threshold division to survey Amount, the acquisition of refractory brick image filter noise reduction process and to be partitioned into refractory brick region same as Example 1.
The measurement method at the refractory brick surface inclination angle based on fit Plane normal vector, includes the following steps：
Step 1 is filtered the original refractory brick coloured image collected by sensor noise reduction, obtains noise reduction Image afterwards；After noise reduction in image, it is partitioned into refractory brick region, obtains refractory brick gray level image, as shown in Figure 1；
Step 2 converts refractory brick gray level image to refractory brick coloured image, using single order planar approach to refractory brick Coloured image surface area carries out approximate fits, obtains fit Plane；
Step 3 determines three not conllinear points, respectively (x in fit Plane1,y1,z1),(x2,y2,z2),(x3, y3,z3), generate two vectors
Step 4：Two vectors ask fork collection to obtain normal vector
Step 5：According to normal vectorSurface tiltangleθ is obtained,
Wherein
It is using bilateral filtering noise reduction, the calculation formula of bilateral filtering in step 1：
Wherein, k is used for carrying out result unitization, and c is the Gauss weight based on space length, and s is based between pixel The Gauss weight of similarity degree, f (ε) are current pixel.
Go out refractory brick region using gray processing and Ostu binarization segmentations in step 1, includes the following steps：
Step 1-1：It is averaging the RGB three-component brightness in refractory brick coloured image to obtain gray value f (x, y),It is worth to refractory brick grey level histogram according to gray scale, wherein R (x, y) is Red channel components, G (x, y) are green channel components, and B (x, y) is blue channel component；
Step 1-2：According to refractory brick grey level histogram, the gradation intervals of refractory brick grey level histogram are divided into three classes, by Two threshold values separate, according to formulaObtain optimal threshold, whereinThe side between class Difference, k1,k2For preset threshold value,For optimal threshold；
Step 1-3：Refractory brick gray-scale map is split according to optimal threshold, obtains refractory brick area image.
In step 2, the method for obtaining fit Plane is：
For a width 2D consecutive image f (x, y) (>=0), p+q rank squares mpqIt is defined as：
Wherein, p, q are non-negative integers, and for discretization digital picture, above formula is：
Wherein, (r0,c0) it is center-of-mass coordinate, and
Single order plane approximation method is described by following formula：
Image (r, c)=α (r-r0)+β(c-c0)+γ (4-13)
Wherein, r0To wait for the abscissa of fitted area, c0To wait for that the abscissa of fitted area, γ are to wait for putting down for fitted area Equal gray scale, F are the area of entire plane, and MRow is the Gray Moment along line direction, and MCol is the Gray Moment along column direction, Then have：
MRow=sum ((r-r0)*(Image(r,c)-γ))/F2 (4-14)
MCol=sum ((c-r0)*(Image(r,c)-γ))/F2 (4-15)
Wherein, α, beta, gamma are the fitting parameter of fit Plane.
Embodiment 5
As seen in figs. 8-10, the refractory brick measuring device based on machine vision, including image capture module 1, control module 2, there is pedestal 11, pedestal 11 to be equipped with three structure lights of carrying for image processing module 3 and feedback module 4, image capture module 1 The guide rail 13 of laser sensor 12, stepper motor 15 drive three structure light laser sensors 12 to move, hold by shaft coupling 16 The turntable 14 for carrying refractory brick is fixed on pedestal 11, and the surface to be measured of three structure light laser sensors 12 alignment refractory brick is simultaneously scanned 4 end faces of refractory brick to be measured；The input terminal of control module 2 is connect with image capture module 1, the output end point of control module 2 It is not connect with image processing module 3 and feedback module 4, feedback module 4 includes travel switch, and executes rejecting screening operation Manipulator 41；Image capture module 1 acquires refractory brick image information, and image processing module 3 obtains refractory brick image information, and Refractory brick image information is analyzed, control module 2 obtains analysis result, and analysis result is fed back to feedback by control module 2 Module 4.
As Figure 1-10 shows, the refractory brick measurement method based on machine vision, comprises the steps of：
Step 1, controller are that turntable 14 turns to the surface to be measured alignment of refractory brick to three structure light laser sensors 12 position makes guide rail 13 be translatable, and three structure light laser sensors 12 scan the surface to be measured of refractory brick, to obtain gray scale The original refractory brick coloured image of information and elevation information fusion；
Step 2, to original refractory brick Algorithm for Color Image Filtering noise reduction, obtain image after noise reduction；
Step 3 in image, is partitioned into refractory brick region after noise reduction, obtains refractory brick image；
Step 4 carries out Morphological scale-space to refractory brick image, so that refractory brick image border is seamlessly transitted, and remove fire resisting Isolated spot noise at brick image border, refractory brick image after being corrected；
Step 5 carries out edge detection with Canny operators to refractory brick image after amendment；
Step 6 carries out Hough transform method and the fitting of K Mean Methods to the discrete point that the edge detection of refractory brick obtains Analysis, calculates the actual size of refractory brick；
Step 7, the refractory brick image obtained by step 3 carry out plane using least square method to refractory brick picture point cloud Fitting obtains zero plane, the point cloud of the point cloud and zero plane that make refractory brick image check the mark obtain slant correction after put cloud number According to figure；
Step 8 is filtered segmentation to the height histogram for putting cloud after slant correction, obtains the point cloud letter of set depth Breath carries out connected component labeling to the point cloud information of set depth according to two-pass scan method, calculates the connected region under each depth Domain area, obtains defective data, and defective data uploads database and shows on a display screen；
Step 9：The surface of cloud datagram fitting refractory brick, obtains refractory brick table after the slant correction obtained according to step 7 The normal vector in face calculates the inclination angle on refractory brick surface, to judge whether the flatness of refractory brick meets the requirements, if inclination angle Less than given threshold value, then flatness is qualified, qualified that inclination angle is uploaded database and is shown on a display screen；
Step 10 makes turntable turn to next detection faces of refractory brick, repeats step 1-9.
It is using bilateral filtering noise reduction, the calculation formula of bilateral filtering in step 2：
Wherein, k is used for carrying out result unitization, and c is the Gauss weight based on space length, and s is based between pixel The Gauss weight of similarity degree, f (ε) are current pixel.
Go out refractory brick region using gray processing and Ostu binarization segmentations in step 3, includes the following steps：
Step 3-1：It is averaging the RGB three-component brightness in refractory brick coloured image to obtain gray value f (x, y),It is worth to refractory brick grey level histogram according to gray scale, wherein R (x, y) is Red channel components, G (x, y) are green channel components, B (x, y）For blue channel component；
Step 3-2：According to refractory brick grey level histogram, the method for obtaining optimal threshold includes：
{ 0,1,2 ..., L-1 } is enabled to indicate a width size for L different gray levels in the digital picture of MN pixels, ni Indicate that gray level is the pixel number of i.Sum of all pixels MN in image is MN=n0+n1+n2+…+nL-1, pi=ni/ MN, i=0, 1,2 ..., L-1, according to the grey level histogram of refractory material, gradation intervals can be divided into three classes, and (these three classes are by two thresholds Value separates), inter-class variance is given by：
Enable P1, P2, P3It is three
In formula,
And there is following relationship to set up：
P1m1+P2m2+P3m3=mG
P1+P2+P3=1
At this point, finding optimal threshold using following formula：
Wherein, piFor a certain gray probability, m1Indicate a kind of average gray, mGIndicate the average gray of whole image,For inter-class variance, k1,k2For preset threshold value,For optimal threshold；
Step 3-3：Refractory brick gray-scale map is split according to optimal threshold, obtains refractory brick area grayscale image.
Opening operation is carried out to refractory brick image in step 4, removes and isolates spot noise；Carry out closed operation smooth boundary.
The method of edge detection comprises the steps of in step 5：
Step 5-1：With Gaussian smoothing function to refractory brick image filtering after amendment, it is f to enable refractory brick image after correcting (x, y), Gaussian function are G (x, y), it is smooth after image be fs(x, y), thenfs(x, y)=G (x, y)*f(x,y)；
Step 5-2：Gradient magnitude image M and angle magnitude image θ is calculated, Its In, GxIndicate the gradient magnitude component of horizontal direction, GyIndicate the gradient magnitude component of vertical direction；
Step 5-3：Non-maxima suppression is carried out to gradient magnitude image M：In gradient direction, to a point, field Center pixel f (x, y) is compared with along the two of gradient line adjacent pixel, if the Grad of f (x, y) is less than or equal to along ladder The Grad for spending two adjacent pixels of line, then enable f (x, y)=0；
Step 5-4：Edge is detected and connected with dual threshold, is supplement with Low threshold image based on high threshold image To connect the edge of image.
In step 7, the acquisition methods of zero plane include：
Step 7-1：The geometric center position for solving refractory brick upper surface point cloud chart picture, according to the planar central solved Then position acquisition plane correction regional center carries out plane fitting to the region using least square method, approximating method is：
For a width 2D consecutive image f (x, y) (>=0), p+q rank squares mpqIt is defined as：
Wherein, p, q are non-negative integers, and for discretization digital picture, above formula is：
Wherein, (r0,c0) it is center-of-mass coordinate, and
Single order plane approximation method is described by following formula：
Image (r, c)=α (r-r0)+β(c-c0)+γ (4-13)
Wherein, r0And c0As wait for that the transverse and longitudinal coordinate of fitted area, γ are the average gray for waiting for fitted area, F is entire The area of plane, MRow are the Gray Moments along line direction, and MCol is the Gray Moment along column direction, then has：
MRow=sum ((r-r0)*(Image(r,c)-γ))/F2 (4-14)
MCol=sum ((c-r0)*(Image(r,c)-γ))/F2 (4-15)
Wherein, α, beta, gamma are respectively the fitting parameter of zero plane.
Step 7-2：The dimensional parameters for obtaining original refractory brick image, according to original refractory brick image size parameter and zero Plane generates corresponding datum plane image (virtual plane image)；
Step 7-3：It is poor to original refractory brick image and benchmark image plane make, and obtains the point cloud after slant correction Datagram.
The solution procedure at inclination angle is in step 9：
Step 9-1, three not conllinear points, respectively (x are determined in fit Plane1,y1,z1),(x2,y2,z2), (x3,y3,z3), generate two vectors
Step 9-2, two vectors ask fork collection to obtain normal vector
Step 9-3, according to normal vectorSurface tiltangleθ is obtained,
Wherein
In step 7-2, the generation method of datum plane image is：According to fitting parameter α, beta, gamma, in conjunction with original refractory brick The dimension information of point cloud chart Image (r, c) generates datum plane image Image (r, c)0。
In step 2, in step 7-3, to original refractory brick point cloud chart Image (r, c) and datum plane image Image (r, c)0It is poor make, and obtains point cloud data figure Image'(r, c after slant correction),
Image'(r, c)=Image (r, c)-Image (r, c)0。
Step 10：PLC signals are waited for, the scanning survey of lower one side is carried out, then handle data according to step 2-9.
Step 11：When 4 face test analysis finish, whether judging certified products, sent by measuring system image processing module Signal instructs manipulator to execute the action of next step rejecting screening to control module.
Content described in this specification embodiment is only enumerating to the way of realization of inventive concept, guarantor of the invention Shield range is not construed as being only limitted to the concrete form that embodiment is stated, protection scope of the present invention is also and in this field Technical staff according to present inventive concept it is conceivable that equivalent technologies mean.

Claims (5)

1. a kind of refractory brick geometric dimension measurement method, which is characterized in that include the following steps：
Step 1：To original refractory brick Algorithm for Color Image Filtering noise reduction, image after noise reduction is obtained；
Step 2：After noise reduction in image, it is partitioned into refractory brick region, obtains refractory brick image；
Step 3：Morphological scale-space is carried out to refractory brick image, so that refractory brick image border is seamlessly transitted, and remove refractory brick figure As the isolated spot noise of edge, refractory brick image after being corrected；
Step 4：Edge detection is carried out to refractory brick image after amendment with Canny operators；
Step 5：Refractory brick shape straight line is extracted using Hough transform method and K Mean Methods, forms the outer profile of refractory brick, Include the following steps：
Step 5-1：A plurality of straight line is fitted to the refractory brick shape straight-line detection after edge detection using Hough transform method, Straight line is indicated using polar coordinate system, then is had：
Xcos θ+ysin θ=ρ
ρ θ parameter spaces are divided into summing elements, wherein (ρmin,ρmax) and (θmin,θmax) parameter area is：-90°≤θ≤ 90 ° and-D≤ρ≤D, wherein D is the diagonal maximum distance of image；
Step 5-2：Given sample set D={ x1,x2,…,xm, k Mean Methods are for cluster gained cluster C={ C1,C2,…,CK} It divides and minimizes square error,
Wherein, k is cluster number of clusters；CiFor cluster；It is cluster CiMean vector, 1≤i≤k；E is spacing distance；
Step 5-3：In conjunction with Hough transform method and K Mean Methods, randomly selects intersection point from multigroup intersection point and be defined as in cluster Straight line is extracted on same side by the heart, until the four edges of extraction refractory brick come out；
Step 6：Refractory brick shape straight-line intersection is sought, distance between two points are the shape geometric dimension of refractory brick.
2. a kind of refractory brick geometric dimension measurement method according to claim 1, it is characterised in that：Using double in step 1 Side filters noise reduction, and the calculation formula of bilateral filtering is：
Wherein, k is used for carrying out result unitization, and c is the Gauss weight based on space length, and s is based on similar journey between pixel The Gauss weight of degree, f (ε) are current pixel.
3. a kind of refractory brick geometric dimension measurement method according to claim 2, it is characterised in that：Ash is used in step 2 Degreeization and Ostu binarization segmentations go out refractory brick region, include the following steps：
Step 2-1：It is averaging the RGB three-component brightness in refractory brick coloured image to obtain gray value f (x, y),It is worth to refractory brick grey level histogram according to gray scale, wherein R (x, y) is Red channel components, G (x, y) are green channel components, and B (x, y) is blue channel component；
Step 2-2：According to refractory brick grey level histogram, the gradation intervals of refractory brick grey level histogram are divided into three classes, according to class Between variance obtain optimal threshold：
Wherein,For inter-class variance, k1,k2For preset threshold value,For optimal threshold；
Step 2-3：Refractory brick gray-scale map is divided according to optimal threshold.
4. a kind of refractory brick geometric dimension measurement method according to claim 3, it is characterised in that：In step 3, morphology Processing includes carrying out opening operation and closed operation to refractory brick image, carries out opening operation removal and isolates spot noise；Carrying out closed operation makes Refractory brick image boundary is smooth.
5. a kind of refractory brick geometric dimension measurement method according to claim 4, it is characterised in that：Edge is examined in step 4 The method of survey comprises the steps of：
Step 4-1：With Gaussian smoothing function to refractory brick image filtering after amendment, it is f (x, y) to enable refractory brick image after correcting, Gaussian function is G (x, y), it is smooth after image be fs(x, y), thenfs(x, y)=G (x, y) * f (x, y)；
Step 4-2：Gradient magnitude image M and angle magnitude image θ is calculated, Wherein, Gx Indicate the gradient magnitude component of horizontal direction, GyIndicate the gradient magnitude component of vertical direction；
Step 4-3：Non-maxima suppression is carried out to gradient magnitude image M：In gradient direction, to a point, the center in field Pixel f (x, y) is compared with along the two of gradient line adjacent pixel, if the Grad of f (x, y) is less than or equal to along gradient line The Grad of two adjacent pixels then enables f (x, y)=0；
Step 4-4：Edge is detected and connected using dual threshold, is that supplement is come with Low threshold image based on high threshold image Connect image border.
CN201810213913.XA 2018-03-14 2018-03-15 A kind of refractory brick geometric dimension measurement method Withdrawn CN108662977A (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810208727 2018-03-14
CN2018102087277 2018-03-14

Publications (1)

Publication Number Publication Date
CN108662977A true CN108662977A (en) 2018-10-16

Family

ID=63785161

Family Applications (2)

Application Number Title Priority Date Filing Date
CN201810213913.XA Withdrawn CN108662977A (en) 2018-03-14 2018-03-15 A kind of refractory brick geometric dimension measurement method
CN201910195416.6A Pending CN110230978A (en) 2018-03-14 2019-03-14 A kind of refractory brick geometric dimension measurement method

Family Applications After (1)

Application Number Title Priority Date Filing Date
CN201910195416.6A Pending CN110230978A (en) 2018-03-14 2019-03-14 A kind of refractory brick geometric dimension measurement method

Country Status (1)

Country Link
CN (2) CN108662977A (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110349169A (en) * 2019-06-24 2019-10-18 杭州汇萃智能科技有限公司 A kind of line measurement method
CN110470218A (en) * 2019-08-15 2019-11-19 浙江工业大学之江学院 A kind of ball stud vision detection system and ball stud detection device
CN110634128A (en) * 2019-08-15 2019-12-31 浙江工业大学之江学院 Ball pin size detection method and device, computer equipment and storage medium
CN111504193A (en) * 2020-05-07 2020-08-07 山东菏鲁电子科技有限公司 Automatic detection method for data line plug size
CN111833317A (en) * 2020-06-30 2020-10-27 佛山科学技术学院 Industrial product specification detection method and equipment based on augmented reality
CN112337810A (en) * 2020-09-14 2021-02-09 西安工程大学 Visual guidance pearl sorting robot and sorting method thereof
CN113192088A (en) * 2021-05-19 2021-07-30 山东纬横数据科技有限公司 Radar image fixed background mask method based on remote sensing image threshold segmentation
CN114663713A (en) * 2022-05-20 2022-06-24 海门市博洋铸造有限公司 Metal wire drawing production control method and system based on Hough transform
CN115115664A (en) * 2022-08-25 2022-09-27 济宁景泽信息科技有限公司 Information acquisition system for measuring instrument

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112648920B (en) * 2019-10-12 2022-05-24 上海微电子装备（集团）股份有限公司 Mask opening size measuring method, mask plate stretching device and screen expanding machine
CN111127481A (en) * 2019-12-20 2020-05-08 上海吉七物联网科技有限公司 Image identification method and device based on TOF image communication area
CN111784722A (en) * 2020-07-15 2020-10-16 山东派蒙机电技术有限公司 Improved Canny lane line edge detection algorithm
CN113177949B (en) * 2021-04-16 2023-09-01 中南大学 Large-size rock particle feature recognition method and device
CN113362468B (en) * 2021-07-05 2022-06-03 上海大学 Dimension measuring method for hub of train wheel
CN114549483A (en) * 2022-02-25 2022-05-27 山东科汇电力自动化股份有限公司 Refractory brick size measurement method based on halcon monocular vision

Family Cites Families (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101793501B (en) * 2010-04-14 2011-04-20 华中科技大学 Transmission line ice coating status detection method based on image
JP2013036964A (en) * 2011-08-11 2013-02-21 Mitsutoyo Corp Image measurement apparatus and image measurement method
CN102914545B (en) * 2012-11-15 2014-10-01 潍坊学院 Gear defect detection method and system based on computer vision
CN103065149B (en) * 2012-12-21 2016-05-04 上海交通大学 Muskmelon Fruit phenotype is extracted and quantization method
CN104732536A (en) * 2015-03-18 2015-06-24 广东顺德西安交通大学研究院 Sub-pixel edge detection method based on improved morphology
CN104820986B (en) * 2015-04-28 2017-06-16 电子科技大学 A kind of cable online test method based on machine vision
CN105403147B (en) * 2015-11-20 2019-07-12 山东科技大学 One kind being based on Embedded bottle embryo detection system and detection method
CN105678768B (en) * 2016-01-08 2018-11-02 杭州电子科技大学 A kind of tyre surface detection method based on machine vision
CN105891186B (en) * 2016-05-12 2019-01-04 中山大学 A kind of polymer phase interface width statistical method
CN105891228A (en) * 2016-06-07 2016-08-24 江南工业集团有限公司 Optical fiber appearance defect detecting and outer diameter measuring device based on machine vision
CN106568424B (en) * 2016-10-27 2017-11-07 西安科技大学 Rocker arm of coal mining machine mining height detection method based on machine vision
CN107084666B (en) * 2017-05-10 2020-08-28 中国计量大学 Comprehensive detection method for size of brake pad based on machine vision
CN107742289A (en) * 2017-10-15 2018-02-27 哈尔滨理工大学 One kind is based on machine vision revolving body workpieces detection method

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110349169A (en) * 2019-06-24 2019-10-18 杭州汇萃智能科技有限公司 A kind of line measurement method
CN110470218A (en) * 2019-08-15 2019-11-19 浙江工业大学之江学院 A kind of ball stud vision detection system and ball stud detection device
CN110634128A (en) * 2019-08-15 2019-12-31 浙江工业大学之江学院 Ball pin size detection method and device, computer equipment and storage medium
CN111504193A (en) * 2020-05-07 2020-08-07 山东菏鲁电子科技有限公司 Automatic detection method for data line plug size
CN111833317A (en) * 2020-06-30 2020-10-27 佛山科学技术学院 Industrial product specification detection method and equipment based on augmented reality
CN112337810A (en) * 2020-09-14 2021-02-09 西安工程大学 Visual guidance pearl sorting robot and sorting method thereof
CN112337810B (en) * 2020-09-14 2024-06-18 西安工程大学 Vision guiding pearl sorting robot and sorting method thereof
CN113192088A (en) * 2021-05-19 2021-07-30 山东纬横数据科技有限公司 Radar image fixed background mask method based on remote sensing image threshold segmentation
CN114663713A (en) * 2022-05-20 2022-06-24 海门市博洋铸造有限公司 Metal wire drawing production control method and system based on Hough transform
CN114663713B (en) * 2022-05-20 2022-08-30 海门市博洋铸造有限公司 Metal wire drawing production control method and system based on Hough transform
CN115115664A (en) * 2022-08-25 2022-09-27 济宁景泽信息科技有限公司 Information acquisition system for measuring instrument
CN115115664B (en) * 2022-08-25 2022-11-18 济宁景泽信息科技有限公司 Information acquisition system for measuring instrument

Also Published As

Publication number Publication date
CN110230978A (en) 2019-09-13

Similar Documents

Publication Publication Date Title
CN108662977A (en) A kind of refractory brick geometric dimension measurement method
CN108596873A (en) The recognition methods of refractory brick deep defects based on height histogram divion
CN108896547A (en) Refractory brick measuring system based on machine vision
CN108550138A (en) Refractory brick surface scratch recognition methods based on frequency filtering enhancing
CN107392232B (en) Flotation working condition classification method and system
CN110084241B (en) Automatic ammeter reading method based on image recognition
CN107145905B (en) Image recognition detection method for looseness of elevator fastening nut
CN104101600B (en) Cross Section of CC Billet testing of small cracks method
CN107358627B (en) Fruit size detection method based on Kinect camera
CN109900711A (en) Workpiece, defect detection method based on machine vision
CN109993099A (en) A kind of lane line drawing recognition methods based on machine vision
CN109242870A (en) A kind of sea horizon detection method divided based on image with textural characteristics
CN108960237B (en) Reading identification method for pointer type oil level indicator
CN109540925B (en) Complex ceramic tile surface defect detection method based on difference method and local variance measurement operator
CN109409290B (en) Thermometer verification reading automatic identification system and method
CN110070526A (en) Defect inspection method based on the prediction of deep neural network temperature figure
CN115100206B (en) Printing defect identification method for textile with periodic pattern
CN105046700A (en) Brightness correction and color classification-based fruit surface defect detection method and system
CN110687121B (en) Intelligent online detection and automatic grading method and system for ceramic tiles
CN104537651A (en) Proportion detecting algorithm and system for cracks in road surface image
CN101315338A (en) Defect detecting system and method for glass product
CN116703911B (en) LED lamp production quality detecting system
CN116721391B (en) Method for detecting separation effect of raw oil based on computer vision
CN115345876B (en) Bolt thread defect detection method based on ultrasonic image
CN110096980A (en) Character machining identifying system

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
WW01 Invention patent application withdrawn after publication
WW01 Invention patent application withdrawn after publication

Application publication date: 20181016