CN105572143B - The detection method of rolled material surface periodic defect in calender line - Google Patents
The detection method of rolled material surface periodic defect in calender line Download PDFInfo
- Publication number
- CN105572143B CN105572143B CN201510952057.6A CN201510952057A CN105572143B CN 105572143 B CN105572143 B CN 105572143B CN 201510952057 A CN201510952057 A CN 201510952057A CN 105572143 B CN105572143 B CN 105572143B
- Authority
- CN
- China
- Prior art keywords
- defect
- periodicity
- rolled material
- material surface
- source images
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/89—Investigating the presence of flaws or contamination in moving material, e.g. running paper or textiles
- G01N21/892—Investigating the presence of flaws or contamination in moving material, e.g. running paper or textiles characterised by the flaw, defect or object feature examined
- G01N21/8922—Periodic flaws
Landscapes
- Engineering & Computer Science (AREA)
- Textile Engineering (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention discloses a kind of detection method of rolled material surface periodic defect in calender line, including step:S1 gathers the defects of rolled material surface to be detected source images;The characteristics of image of defect in S2 extraction defect source images;S3 carries out periodicity defect to defect source images and judges in advance;S4 matches the adjacent doubtful periodicity defect of current longitudinal region using image correlation matching method, according to the characteristics of image of doubtful periodicity defect to determine periodicity defect one by one.The present invention can rapidly and accurately detect the periodicity defect on rolled material surface, the omission factor of various environment periodical defects detections and false defect false drop rate can be improved, suitable for any material.
Description
Technical field
The invention belongs to rolled material surfaces in material surface technical field of optical detection more particularly to a kind of calender line
The detection method of periodicity defect.
Background technology
Periodicity defect is the major defect caused by preceding process disturbances in calendering production process, and identification early can be reduced
Defect production rate.Surface defects detection based on machine vision is had been applied to recognition methods among every field, such as
The identification of paper products surface defect image, textile defect identification, steel strip surface defect identification etc..But material surface periodicity defect
Detection method it is seldom, there is also following problems for current existing detection method:
(1) Chinese patent of Patent No. ZL200980112171.0《Periodicity defect detection device and its method》Middle public affairs
A kind of magnetic field detection method has been opened, i.e., magnetic field has first been converted into image, then periodicity defect identification is carried out based on image.It examines in the magnetic field
Survey method detection rates are relatively low, and are only applicable to metal surface.
(2) Chinese patent of Patent No. ZL201010205357.5《The detection method of defects in periodic texture images
And device》A kind of detection method based on machine vision is disclosed, this method easily will be in the more material surface image of false defect
Aperiodicity defect is mistaken for periodicity defect;When the color and pattern of part periodic defect change, this method meeting
Cause the missing inspection of periodicity defect;When periodicity defect is discontinuous to be occurred, this method can be known one group of periodicity defect by mistake
It Wei not two groups of periodicity defects or something lost inspection.
With the extensive use of the detection method based on machine vision, it is periodical to use it for product surface in calender line
The identification of defect has important theory significance and realistic meaning.
Background technology is related to following bibliography:
[1]S Zhao,H Huang,et al.Estimating periodic length and defect
coordinates from a series of integer values[J].Measurement,2009,42(4):516-
523.
[2]Changhyun Park,S Choi,et al.Vision-based inspection for periodic
defects in steel wire rod production[J].Optical Engineering,2010,49(1):
017202-017202-10.
[3]FG Bulnes,U Ruben,et al.An efficient method for defect detection
during the manufacturing of web materials[J].Journal of Intelligent
Manufacturing,2014:1-15.
[4]FG Bulnes,U Ruben,et al.Vision-Based Sensor for Early Detection of
Periodical Defects in Web Materials[J].Sensors,2012,12(8):10788-10809.
[5]FG Bulnes,R Usamentiaga,et al.Periodic defects in steel strips:
detection through a vision-base technique.Industry Applications Magazine
IEEE,2013,19:39-46.
The content of the invention
For the deficiencies in the prior art, the present invention provides rolled material surface periodics in a kind of calender line
The detection method of defect the present invention is based on realizing of Robot Vision, can quick and precisely identify the periodicity defect on rolled material surface,
And suitable for any material.
In order to solve the above technical problems, the present invention adopts the following technical scheme that:
The detection method of rolled material surface periodic defect in calender line, including step:
S1 gathers the defects of rolled material surface to be detected source images;
S2 extraction defect source images in defect characteristics of image, the characteristics of image include position feature, shape characteristic,
Color characteristic, centroid feature and section feature;
S3 carries out periodicity defect to defect source images and judges that this step further comprises in advance:
3.1 preset the spacing range of adjacent periods defect as [η according to actual conditions1,η2];
Defect source images are divided into n sections of wide d, long 3 η by 3.2 in the width direction2Longitudinal region, 10mm<d<300mm;
3.3 judge defect number in longitudinal region, and the longitudinal region of 3 is reached to defect number and performs step 3.4;It is if all vertical
Into region, defect number is not up to 3, then defect source images are determined as aperiodicity defect;
If number is not less than 3 [h- Δs h, h+ Δ h] scope the defects of for width position in 3.4 current longitudinal regions, then
Step 3.5 is performed to current longitudinal region;Otherwise, current longitudinal region is determined as aperiodicity defect;
The position relationship matrix P of the 3.5 current longitudinal region defects of structure, and judge doubtful periodicity defect, it is specially:
(a) R=1, j=1, k=2 are initialized;
(b) column vector P is judgedRWhether middle element meets condition (a) 2-5%<PRk/PRj<2+5%;Or (b) 3-5%<PRk/
PRj<3+5%, if satisfied, current j-th of defect of longitudinal region is determined as first doubtful periodicity defect, first doubtful week
The defects of after phase property defect, is denoted as doubtful periodicity defect, and step S4 is performed to current longitudinal region;Otherwise, step is performed
(c);
(c) judge whether k is (m-1), if so, performing step (d);Otherwise, k is made to add 1, performs step (c);
(d) judge whether j is (m-2), if so, performing step (e);Otherwise, j is made to add 1, k=2, then performs step
(c);
(e) judge whether R is (m-1), if so, current longitudinal region is without doubtful periodicity defect;Otherwise, R is made to add 1, j=
1, k=2, then perform step (b);
S4 is according to the characteristics of image of doubtful periodicity defect, using image correlation matching method to the phase of current longitudinal region
Adjacent doubtful periodicity defect is matched one by one, to determine periodicity defect;
It is above-mentioned, position relationship matrix
PidjCurrent j-th of defect of longitudinal region is represented in the position of rolling direction, j=1,2 ... m, m are current longitudinal region
Defect number;PRRepresent the column vector that R is arranged in position relationship matrix P;PRk、PRj、PRlColumn vector P is represented respectivelyRMiddle kth, j, l are a
Element;H is current first defect of longitudinal region in the position of defect source images width;Δ h is the width of periodicity defect
Direction position deviation is spent, Δ h is empirical value.
It is described using source images the defects of machine vision detection device acquisition rolled material surface to be detected in step S1
Machine vision detection device include CCD camera and light source, light source irradiation rolled material surface;The covering calendering of CCD camera visual angle
Material width, source images the defects of for obtaining rolled material surface.
As a kind of specific embodiment, machine vision detection device includes four CCD cameras and two light sources, two light
The defects of rolled material upper and lower surface is irradiated in source respectively, and four CCD cameras shoot rolled material upper and lower surface respectively source images.
Above-mentioned shape characteristic includes defect topmost away from defect source images distance from top, defect away from left end defect source images
Left side distance, defect center are away from distance, defect length, flaw height, defect length-breadth ratio, defect face on the left of defect source images
It is a variety of in product, defect perimeter, defect concentration, defect perimeter ratio, defect rounding.
Brightness of the above-mentioned color characteristic including defect, gray scale, contrast, brightness storehouse q+r+s areas, darkness storehouse q+r+s faces
It is a variety of in product, mean intensity, brightness mean intensity, darkness mean intensity and strength range.
Above-mentioned centroid feature includes the horizontal centroid position, longitudinal centroid position and barycenter ratio of defect.
Step S4 is specially:
According to the characteristics of image of doubtful periodicity defect, the adjacent doubtful periodicity defect of current longitudinal region is sequentially calculated
Similitude, if similitude is not less than predetermined threshold value T_Cor, adjacent doubtful periodicity defect is determined as periodicity defect;It is no
Then, it is not periodicity defect;Threshold value T_Cor is empirical value, takes 0.5~1.
To reduce the detection time of periodicity defect, the present invention can also carry out increment according to fixed periodicity defect
Defect matches, and is specially:
In same longitudinal region, newly-increased defect meets the newly-increased defect of condition (1) simultaneously and previous definite periodicity lacks
The position deviation for being trapped in width position is less than position deviation Δ h;(2) increase defect newly and previous definite periodicity defect exists
In the range of the distance [δ-Δ δ, δ+Δ δ] of rolling direction;(3) image for increasing defect and previous definite periodicity defect newly is special
The similitude of sign vector is not less than threshold value T_Cor;It then increases defect newly and is determined as periodicity defect;
Above-mentioned, δ is according to cycle of fixed periodicity defect, and Δ δ is periodic deviation, in value and calender line most
The roller diameter of big roller is related.
Compared to the prior art, the present invention has following features and advantageous effect:
1st, it can rapidly and accurately detect the periodicity defect on rolled material surface, various environment periodicals can be improved and lacked
The omission factor of detection and false defect false drop rate are fallen into, suitable for any material.
2nd, easy to operate, algorithm linking is good, and input data is few, convenient for applying and safeguarding, suitable for industrial applications.
Description of the drawings
Fig. 1 is the particular flow sheet of the present invention;
Fig. 2 is the schematic diagram that machine vision device gathers defect source images;
Fig. 3 is shape characteristic schematic diagram;
Fig. 4 is the pre- decision flowchart of periodicity defect;
Fig. 5 divides schematic diagram for defect source images longitudinal region;
Fig. 6 is the distribution schematic diagram of periodicity defect;
Fig. 7~11 are the defects of embodiment image.
Specific embodiment
Below in conjunction with attached drawing, specific embodiments of the present invention is described in detail.
See Fig. 1, the method for the present invention is as follows:
Step 1, the defects of rolled material surface to be detected being gathered by machine vision detection device source images.
See Fig. 2, the machine vision detection device includes CCD camera, image processing unit and light source, and CCD camera is used
Come source images the defects of obtaining rolled material surface.CCD camera visual angle should cover rolled material width, and light source is irradiated in calendering material
Expect surface, CCD camera gather the defects of source images be transmitted to image processing unit.In the present embodiment, a CCD camera corresponds to a light
CCD camera shooting is respectively adopted in source, rolled material upper and lower surface.
Subsequent step is based on defect source images recognition cycle defect.
Step 2, the characteristics of image of defect in defect source images is extracted, the characteristics of image includes position feature, pattern
Feature, color characteristic, centroid feature and section feature.
(1) position of the position feature, that is, defect in defect source images.
(2) shape characteristic.
Common shape characteristic mainly has:Defect is topmost away from defect source images distance from top, defect away from left end defect
Distance on the left of source images, defect center away from distance on the left of defect source images, defect length, flaw height, defect length-breadth ratio, lack
Fall into area, defect perimeter, defect concentration, defect perimeter ratio, defect rounding.Partial Feature is as shown in Figure 3.
Defect perimeter
Wherein, f (x, y) limits pixel for threshold value, and value is 0 or 1;B is defect borderline region.Pixel (x, y) is pure white
When, f (x, y) value 0;When pixel (x, y) is ater, f (x, y) value 1.
Defect concentration Density is the ratio of defect area and its Bounding Box area, this value can represent defect compared with it
The density of Bounding Box, Density=(Area/ defect Bounding Boxs area) × 100%, wherein, Area is defect area.
Defect perimeter ratio Perimeter ratio represent defect perimeter Defect Perimeter and its Bounding Box perimeter
Between matching degree, Perimeter ratio=Defect Perimeter/ defect Bounding Box perimeters.
Bounding Box is the rectangular area obtained by region of interesting extraction, and defect part, region of interest are included in the region
Domain is extracted as routine techniques, and Gaussian processes, local thresholding method etc. can be used and extract.
Defect rounding be using between 0.0 to 1.0 digital representation defect with completely justify similar degree, defect rounding=
(defect circumscribed circle diameter-defect inscribed circle diameter)/defect circumscribed circle diameter.
(2) color characteristic
Color is the coloured image bottom, most intuitively one of physical features, the color characteristic that the present invention uses mainly have:
The brightness of defect, gray scale, contrast, brightness storehouse q+r+s areas, darkness storehouse q+r+s areas, mean intensity, brightness mean intensity,
Darkness mean intensity, strength range.
Brightness storehouse q+r+s areas and darkness storehouse q+r+s areas represent respectively defect pixel belong to brightness storehouse and darkness storehouse q,
R, the area of s, wherein, q, r, s are marked with sequence number 1,2,3, q, r, s any combination, if the sequence number of q, r, s mark is consistent, only
Remember one group.Such as, brightness storehouse 1+1+2 areas, that is, brightness storehouse 1+2 areas.Q, r, s represent brightness range respectively, usually with gray value
Based on 128, when being labeled as 1, the brightness range of expression is [0,20], is labeled as the brightness range represented when 2 as [20,40],
The brightness range represented when 3 is labeled as [40,60].For example, the defects of brightness storehouse 1+1+2 areas, that is, brightness is 128 pixel faces
Product subtracts the defects of brightness is located at [128+0,128+20] and [128+20,128+40] elemental area;Brightness storehouse 1+1+2 areas are
The defects of the defects of brightness is 128 elemental area is located at [128+0,128+20] and [128+20,128+40] plus brightness pixel
Area.
Mean intensity represents the average gray level intensity value of defect pixel.
Brightness mean intensity and darkness mean intensity represent respectively defect bright pixel and dark pixel average gray level it is strong
Angle value.Gray value is less than 128 pixel, that is, bright pixel, is otherwise dark pixel.
Strength range represents the strength range between the most bright intensity of gray level of defect pixel and most dark intensity.
(3) centroid feature
The description of barycenter characteristic is based on defect barycenter attribute, the horizontal centroid position including defect, longitudinal centroid position and matter
Heart ratio totally 3 kinds of features.
Making defect Bounding Box primitive definition, (x, y) represents defect Bounding Box location of pixels for g (x, y), and defect Bounding Box is big
It is small to represent horizontal centroid position and longitudinal centroid position respectively for M × N, I, J.
So horizontal centroid position meets condition:
Longitudinal centroid position meets condition:
Position feature and section feature are conventional characteristics of image, and this will not be repeated here.
Step 3, as constraint, it is pre- to carry out periodicity defect to defect source images for position relationship using between adjacent periods defect
Judge.
Fig. 4 is the idiographic flow schematic diagram of this step.Since the spacing of rolled material surface periodic defect must have most
Big value, the spacing range that adjacent periods defect is preset according to actual conditions is [η1,η2], η1For possible minimum period, cycle
Cycle δ >=η of property defect1.Defect source images are divided into n sections of longitudinal regions in the width direction, see Fig. 5, each section of longitudinal region
Width is d, 10mm<d<300mm, each section of longitudinal region length are 3 η2。
This step the specific implementation process is as follows:
3.1 judge defect number in each section of longitudinal region, and the longitudinal region of 3 is reached to defect number and performs step 3.2;If institute
There is longitudinal region defect number to be not up to 3, then judge aperiodicity defect.
I-th section of longitudinal region is denoted as id regions, i=1, and 2 ... n.
The width position uniformity of defect judges in 3.2 longitudinal regions.
It is that for first defect in the position of defect source images width, Δ h is artificial default week in longitudinal region to make h
Phase property defect width position deviation, Δ h be the empirical value based on actual conditions;hidjRepresent j-th of regions of id defect
In the position of width.
To id regions, it is assumed thatT={ hidj∈ [h- Δs h, h+ Δ h] }, j=1,2 ... m, m, that is, id area
The defects of domain number.If first prime number is not less than 3 in set T, then step 3.3 is performed to the id regions;Otherwise, it is determined that id regions
Aperiodicity defect.
The periodicity defect of 3.3 longitudinal regions judges in advance.
Define PidjIt is j-th of id regions defect in the position of rolling direction, j=1,2 ... m, m are defect in id regions
Number.
Build the position relationship matrix P of id area defects:
In this step, judge that doubtful periodicity lacks by the multiple proportion of element in inquiring position relational matrix P column vectors
It falls into, is specially:
The position relationship matrix P R column vectors arranged are denoted as PR, R=1,2 ... (m-1), if PRMiddle presence:
(1) 2-5%<PRk/PRj<2+5%, i.e. PRk/PRj≈ 2, PRkAnd PRjColumn vector P is represented respectivelyRMiddle kth and j member
Element, k>J, then determine that for current j-th of defect of longitudinal region, it is denoted as in the position of rolling direction for doubtful periodicity defect
Pidj;
Or (2) 3-5%<PRl/PRj<3+5%, i.e. PRl/PRj≈ 3, PRlRepresent column vector PRIn l-th of element, l>J, then
Determine that it, for current j-th of defect of longitudinal region, is denoted as P by doubtful periodicity defect in the position of rolling directionidj。
In above-mentioned decision process, once defective be determined as doubtful periodicity defect, then terminate to judge, perform step 4.
First doubtful periodicity defect and its position P in id regions are obtained by above-mentioned judgementidj, first doubtful week
The defects of after phase property defect, is denoted as doubtful periodicity defect.Doubtful periodicity defect is the further inspection of subsequent periodic defect
It surveys and basis is provided, and the detection range on rolling direction is reduced by way of being phased out aperiodicity defect, so as to have
Effect improves the pre- judgement efficiency of periodicity defect.
Step 4 is performed to there are the longitudinal regions of doubtful periodicity defect.
Step 4, the adjacent doubtful periodicity defect of each longitudinal region is carried out respectively using image correlation matching method by
One matching, to determine periodicity defect.
This step is used for image correlation matching method can increase substantially the cycle for the definite of doubtful periodicity defect
The detection order of accuarcy and detection speed of property defect.
The characteristics of image that defect is extracted according to step 2 obtains the characteristics of image of doubtful periodicity defect, and forms doubtful week
Image feature vector G=(the g of phase property defect1,g2,...gt).Take doubtful periodicity defect Pirl, remember that its image feature vector is
Gl=(gl1,gl2,...glt), l is doubtful periodicity defect number in current longitudinal region.
For each longitudinal region, the similitude between adjacent doubtful periodicity defect is sequentially calculated based on image feature vector
Cor:
Cor=C/ σ (4)
Wherein:
C represents adjacent doubtful periodicity defect characteristics of image
The covariance of vector, GnkAnd G(n+1)kK-th yuan in respectively adjacent doubtful periodicity defect image feature vector
Element, GnaFor a-th of element in the image feature vector of one of adjacent doubtful periodicity defect image, G(n+1)bFor adjacent doubtful week
B-th of element in two image feature vector of phase property defect image;
For variance.
Given threshold T_Cor, as Cor >=T_Cor, current adjacent doubtful periodicity defect is determined as periodicity defect.
When it is implemented, threshold value T_Cor is the empirical value manually set, the value generally in the range of 0.5~1.
Cycle δ=P of identified periodicity defectRj, the roller diameter phase of periodic deviation Δ δ and maximum roller in calender line
It closes, less than the 5% of roller diameter, the Δ δ if roller diameter size is 500mm<25mm.
Step 5, increment defect matches.
To having there is the region of periodicity defect, the matching to periodicity defect can be passed through.This step is to subtract
The repetition of few step 1~4, quickly judges to have determined that whether periodicity defect region periodicity defect in calender line stops
Occur, high degree reduces the detection time of periodicity defect.
The defects of meeting following all conditions i.e. newly-increased periodicity defect:
(1) defect is increased newly in the position deviation of width position and previous definite periodicity defect width to be less than
Position deviation Δ h.
(2) increase newly defect rolling direction position and previous definite periodicity defect rolling direction positional distance [δ-
Δ δ, δ+Δ δ] in the range of.
(3) similitude Cor >=T_Cor of the special image feature vector with previous definite periodicity defect of defect is increased newly.
Embodiment
Steel surface periodicity defect detects
In the present embodiment, iron and steel material quality is cold rolled silicon steel, 5 groups of defect image data is taken, including different types of periodicity
Defect image and aperiodicity defect image, are shown in Fig. 7~11.Table 1 show defects detection in embodiment as a result, missing inspection number refers to week
Phase property defect is mistaken for the quantity of aperiodicity defect, and flase drop number refers to the quantity that aperiodicity defect is mistaken for periodicity defect.
Fig. 6 show periodicity defect distribution schematic diagram, wherein, figure (a) is distributed for preferable periodicity defect, due to pressure
Prolong the variation of the extraneous factors such as technique, acquisition environment, vibration equipment, other two kinds common periodicity defect distributions can be caused:
Scheme the offset-type periodicity defect distribution shown in room type periodicity defect distribution and the figure (c) shown in (b).Room type is periodical
The discontinuous appearance of defect distribution, that is, periodicity defect, it may appear that single or continuous multiple rooms.Offset-type periodicity defect is distributed
Generally caused by strip running deviation, part periodic defect deviates.The parameter used in the present embodiment is shown in Table 2, testing result
It is shown in Table 1.The result shows that in the present embodiment, the omission factor of periodicity defect detection less than 8%, false drop rate be less than 4%, it can be achieved that
Detection in real time, does not interfere with production process.
1 defects detection result of table
2 parameter of table and threshold value
Title | Symbol | Value |
Periodicity defect spacing | [η1,η2] | [200mm,20000mm] |
Longitudinal region width | d | 100mm |
Allow defect width position deviation | Δh | 15mm |
Use characteristics of image number | t | 75 |
Allow defect periodic deviation | Δδ | 15mm |
Periodicity defect similarity threshold | T_Cor | 0.8 |
Claims (8)
1. the detection method of rolled material surface periodic defect in calender line, which is characterized in that including:
S1 gathers the defects of rolled material surface to be detected source images;
The characteristics of image of defect, the characteristics of image include position feature, shape characteristic, color in S2 extraction defect source images
Feature, centroid feature and section feature;
S3 carries out periodicity defect to defect source images and judges that this step further comprises in advance:
3.1 preset the spacing range of adjacent periods defect as [η according to actual conditions1,η2];
Defect source images are divided into n sections of wide d, long 3 η by 3.2 in the width direction2Longitudinal region, 10mm<d<300mm;
3.3 judge defect number in longitudinal region, and the longitudinal region of 3 is reached to defect number and performs step 3.4;If all longitudinal regions
Defect number is not up to 3 in domain, then defect source images are determined as aperiodicity defect;
If number is not less than 3 [h- Δs h, h+ Δ h] scope the defects of for width position in 3.4 current longitudinal regions, then to working as
Preceding longitudinal region performs step 3.5;Otherwise, current longitudinal region is determined as aperiodicity defect;
The position relationship matrix P of the 3.5 current longitudinal region defects of structure, and judge doubtful periodicity defect, it is specially:
(a) R=1, j=1, k=2 are initialized;
(b) column vector P is judgedRWhether middle element meets condition (a) 2-5%<PRk/PRj<2+5%;Or (b) 3-5%<PRk/PRj<3
+ 5%, if satisfied, current j-th of defect of longitudinal region is determined as first doubtful periodicity defect, first doubtful periodicity
The defects of after defect, is denoted as doubtful periodicity defect, and step S4 is performed to current longitudinal region;Otherwise, step (c) is performed;
(c) judge whether k is (m-1), if so, performing step (d);Otherwise, k is made to add 1, performs step (c);
(d) judge whether j is (m-2), if so, performing step (e);Otherwise, j is made to add 1, k=2, then performs step (c);
(e) judge whether R is (m-1), if so, current longitudinal region is without doubtful periodicity defect;Otherwise, R is made to add 1, j=1, k
=2, then perform step (b);
S4 doubts the adjacent of current longitudinal region using image correlation matching method according to the characteristics of image of doubtful periodicity defect
It is matched one by one like periodicity defect, to determine periodicity defect;
It is above-mentioned, position relationship matrixPidjRepresent current
J-th of defect of longitudinal region is in the position of rolling direction, j=1, and 2 ... m, m are current longitudinal region defect number;PRRepresent position
Put the column vector that R is arranged in relational matrix P;PRk、PRjColumn vector P is represented respectivelyRMiddle kth, j element;H is current longitudinal region
First defect is in the position of defect source images width;Δ h is the width position deviation of periodicity defect, and Δ h is
Empirical value.
2. the detection method of rolled material surface periodic defect in calender line as described in claim 1, it is characterised in that:
In step S1, using source images the defects of machine vision detection device acquisition rolled material surface to be detected, the machine
Device vision inspection apparatus includes CCD camera and light source, light source irradiation rolled material surface;CCD camera visual angle covers rolled material
Width, source images the defects of for obtaining rolled material surface.
3. the detection method of rolled material surface periodic defect in calender line as claimed in claim 2, it is characterised in that:
The machine vision detection device includes two CCD cameras and two light sources, and two light sources irradiate rolled material respectively
The defects of upper and lower surface, two CCD cameras shoot rolled material upper and lower surface respectively source images.
4. the detection method of rolled material surface periodic defect in calender line as described in claim 1, it is characterised in that:
It is left away from defect source images away from defect source images distance from top, defect left end topmost that the shape characteristic includes defect
Lateral extent, defect center away from distance on the left of defect source images, defect length, flaw height, defect length-breadth ratio, defect area,
It is a variety of in defect perimeter, defect concentration, defect perimeter ratio, defect rounding;
The defect perimeter ratio Perimeter ratio=Defect Perimeter/ defect Bounding Box perimeters, wherein,
Defect Perimeter represent defect perimeter;
The defect rounding=(defect circumscribed circle diameter-defect inscribed circle diameter)/defect circumscribed circle diameter.
5. the detection method of rolled material surface periodic defect in calender line as described in claim 1, it is characterised in that:
The brightness of the color characteristic including defect, gray scale, contrast, brightness storehouse q+r+s areas, darkness storehouse q+r+s areas,
It is a variety of in mean intensity, brightness mean intensity, darkness mean intensity and strength range;
The brightness storehouse q+r+s areas and the darkness storehouse q+r+s areas are respectively that defect pixel belongs to brightness storehouse and darkness storehouse
Q, the area of r, s, wherein, q, r, s are marked with sequence number 1,2,3, q, r, s any combination, if the sequence number of q, r, s mark is consistent,
Only one group of note;
The mean intensity is the average gray level intensity value of defect pixel;
The brightness mean intensity and the darkness mean intensity are respectively the bright pixel of defect and the average gray level of dark pixel
Intensity value;
The strength range represents the strength range between the most bright intensity of gray level of defect pixel and most dark intensity.
6. the detection method of rolled material surface periodic defect in calender line as described in claim 1, it is characterised in that:
The centroid feature includes the horizontal centroid position, longitudinal centroid position and barycenter ratio of defect.
7. the detection method of rolled material surface periodic defect in calender line as described in claim 1, it is characterised in that:
Step S4 is specially:
According to the characteristics of image of doubtful periodicity defect, the phase of the adjacent doubtful periodicity defect of current longitudinal region is sequentially calculated
Like property, if similitude is not less than predetermined threshold value T_Cor, adjacent doubtful periodicity defect is determined as periodicity defect;Otherwise, no
For periodicity defect;Threshold value T_Cor is empirical value, takes 0.5~1.
8. the detection method of rolled material surface periodic defect in calender line as described in claim 1, it is characterised in that:
Further include step:
Increment defect matching is carried out according to fixed periodicity defect, is specially:
In same longitudinal region, newly-increased defect meets the newly-increased defect of condition (1) simultaneously and previous definite periodicity defect exists
The position deviation of width position is less than position deviation Δ h;(2) increase defect newly and previous definite periodicity defect is rolling
In the range of the distance [δ-Δ δ, δ+Δ δ] in direction;(3) increase newly the characteristics of image of defect and previous definite periodicity defect to
The similitude of amount is not less than threshold value T_Cor;It then increases defect newly and is determined as periodicity defect;
Above-mentioned, δ is the cycle of fixed periodicity defect, and Δ δ is periodic deviation, the roller of maximum roller in value and calender line
Diameter is related.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510952057.6A CN105572143B (en) | 2015-12-17 | 2015-12-17 | The detection method of rolled material surface periodic defect in calender line |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510952057.6A CN105572143B (en) | 2015-12-17 | 2015-12-17 | The detection method of rolled material surface periodic defect in calender line |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105572143A CN105572143A (en) | 2016-05-11 |
CN105572143B true CN105572143B (en) | 2018-05-25 |
Family
ID=55882519
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510952057.6A Active CN105572143B (en) | 2015-12-17 | 2015-12-17 | The detection method of rolled material surface periodic defect in calender line |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105572143B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109493311B (en) * | 2017-09-08 | 2022-03-29 | 上海宝信软件股份有限公司 | Irregular defect picture pattern recognition and matching method and system |
CN113933312B (en) * | 2021-09-30 | 2024-07-05 | 杭州百子尖科技股份有限公司 | Method for judging periodicity rule of surface defects in real time |
CN116519708B (en) * | 2023-03-28 | 2023-10-31 | 钛玛科(北京)工业科技有限公司 | Method and device for detecting surface defects of coiled material |
CN116559183B (en) * | 2023-07-11 | 2023-11-03 | 钛玛科(北京)工业科技有限公司 | Method and system for improving defect judging efficiency |
CN116754567B (en) * | 2023-08-18 | 2023-11-10 | 钛玛科(北京)工业科技有限公司 | Periodic defect detection method, device and equipment for copper foil material |
CN118392872A (en) * | 2024-04-28 | 2024-07-26 | 北京海云捷迅科技股份有限公司 | Surface defect statistics method and device for metal surface and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102007400A (en) * | 2008-03-31 | 2011-04-06 | 杰富意钢铁株式会社 | Periodic defect detecting device and method for the same |
CN102099672A (en) * | 2008-07-18 | 2011-06-15 | 旭硝子株式会社 | Image data processing apparatus and method for defect inspection, defect inspecting apparatus and method using the image data processing apparatus and method, board-like body manufacturing method using the defect inspecting apparatus and method, and storage medium |
CN102279191A (en) * | 2010-06-13 | 2011-12-14 | 中钞特种防伪科技有限公司 | Detection method and apparatus for defects in periodic texture images |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4516884B2 (en) * | 2005-04-28 | 2010-08-04 | 新日本製鐵株式会社 | Periodic defect inspection method and apparatus |
EP2929330A2 (en) * | 2012-12-10 | 2015-10-14 | Uster Technologies AG | Detection of a periodic structure in a moving elongated textile material |
-
2015
- 2015-12-17 CN CN201510952057.6A patent/CN105572143B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102007400A (en) * | 2008-03-31 | 2011-04-06 | 杰富意钢铁株式会社 | Periodic defect detecting device and method for the same |
CN102099672A (en) * | 2008-07-18 | 2011-06-15 | 旭硝子株式会社 | Image data processing apparatus and method for defect inspection, defect inspecting apparatus and method using the image data processing apparatus and method, board-like body manufacturing method using the defect inspecting apparatus and method, and storage medium |
CN102279191A (en) * | 2010-06-13 | 2011-12-14 | 中钞特种防伪科技有限公司 | Detection method and apparatus for defects in periodic texture images |
Non-Patent Citations (3)
Title |
---|
Estimating periodic length and defect coordinates from a series of integer values;Sanxing Zhao et al.;《Measurement》;20091231;第42卷(第4期);第516-523页 * |
Periodic Defects in Steel Strips:Detection through a Vision-based Technique;Francisco G.Bulnes et al.;《Idustry Application Magazine》;20130430;第19卷(第2期);第39-46页 * |
Vision-Based Sensor for Early Detection of Periodical Defects in Web Materials;Francisco G.Bulnes et al.;《Sensors》;20120831;第12卷(第8期);第10788-10809页 * |
Also Published As
Publication number | Publication date |
---|---|
CN105572143A (en) | 2016-05-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105572143B (en) | The detection method of rolled material surface periodic defect in calender line | |
Zhu et al. | Yarn-dyed fabric defect detection based on autocorrelation function and GLCM | |
Çelik et al. | Development of a machine vision system: real-time fabric defect detection and classification with neural networks | |
CN104458755B (en) | Multi-type material surface defect detection method based on machine vision | |
CN104268505B (en) | Fabric Defects Inspection automatic detecting identifier and method based on machine vision | |
CN108416766B (en) | Double-side light-entering type light guide plate defect visual detection method | |
CN110009618B (en) | Shaft part surface quality detection method and device | |
CN104574353B (en) | The surface defect decision method of view-based access control model conspicuousness | |
CN107389701A (en) | A kind of PCB visual defects automatic checkout system and method based on image | |
CN108090494B (en) | Textile defect identification method based on Gabor filter and support vector machine | |
CN104574389A (en) | Battery piece chromatism selection control method based on color machine vision | |
CN104198497B (en) | Surface defect detection method based on visual saliency map and support vector machine | |
CN108491788A (en) | A kind of intelligent extract method and device for financial statement cell | |
CN109544522A (en) | A kind of Surface Defects in Steel Plate detection method and system | |
CN109087286A (en) | A kind of detection method and application based on Computer Image Processing and pattern-recognition | |
CN111266315A (en) | Ore material online sorting system and method based on visual analysis | |
CN107369155A (en) | A kind of cloth surface defect detection method and its system based on machine vision | |
CN102654464A (en) | Copper strip surface defect detection system based on multi-feature fuzzy recognition | |
CN106780526A (en) | A kind of ferrite wafer alligatoring recognition methods | |
CN111047655A (en) | High-definition camera cloth defect detection method based on convolutional neural network | |
Samarawickrama et al. | Matlab based automated surface defect detection system for ceremic tiles using image processing | |
CN110781913B (en) | Zipper cloth belt defect detection method | |
CN105069778B (en) | Based on the industrial products detection method of surface flaw that target signature notable figure builds | |
CN110763700A (en) | Method and equipment for detecting defects of semiconductor component | |
Shi et al. | A faster-rcnn based chemical fiber paper tube defect detection method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |