CN101819162A - Empty bottle wall defect detection method and device - Google Patents
Empty bottle wall defect detection method and deviceDownload PDFInfo
- Publication number
- CN101819162A CN101819162ACN201010170952ACN201010170952ACN101819162ACN 101819162 ACN101819162 ACN 101819162ACN 201010170952 ACN201010170952 ACN 201010170952ACN 201010170952 ACN201010170952 ACN 201010170952ACN 101819162 ACN101819162 ACN 101819162A
- Authority
- CN
- China
- Prior art keywords
- bottle wall
- image
- bottle
- empty bottle
- reflex mirror
- 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.)
- Pending
Links
- 238000001514detection methodMethods0.000titleclaimsabstractdescription37
- 230000007547defectEffects0.000titleclaimsabstractdescription30
- 238000012545processingMethods0.000claimsabstractdescription22
- 238000000034methodMethods0.000claimsabstractdescription18
- 238000004458analytical methodMethods0.000claimsabstractdescription9
- 238000007781pre-processingMethods0.000claimsabstractdescription7
- 230000011514reflexEffects0.000claimsdescription68
- 230000002950deficientEffects0.000claimsdescription15
- 230000005540biological transmissionEffects0.000claimsdescription9
- 238000006243chemical reactionMethods0.000claimsdescription9
- 238000012546transferMethods0.000claimsdescription6
- 230000009466transformationEffects0.000claimsdescription6
- 238000012935AveragingMethods0.000claimsdescription3
- 241000920471Lucilia caesarSpecies0.000claimsdescription3
- 230000002146bilateral effectEffects0.000claimsdescription3
- 239000000835fiberSubstances0.000claimsdescription3
- 238000003709image segmentationMethods0.000claimsdescription3
- 238000010606normalizationMethods0.000claimsdescription3
- 230000001131transforming effectEffects0.000claimsdescription3
- 238000004519manufacturing processMethods0.000description7
- 238000012360testing methodMethods0.000description4
- 238000010586diagramMethods0.000description3
- 239000011521glassSubstances0.000description3
- 238000013461designMethods0.000description2
- 230000010354integrationEffects0.000description2
- 208000037656Respiratory SoundsDiseases0.000description1
- 235000013405beerNutrition0.000description1
- 235000013361beverageNutrition0.000description1
- 238000010276constructionMethods0.000description1
- 230000007812deficiencyEffects0.000description1
- 238000005516engineering processMethods0.000description1
- 206010016256fatigueDiseases0.000description1
- 238000005286illuminationMethods0.000description1
- 238000012856packingMethods0.000description1
- 238000005406washingMethods0.000description1
Images
Landscapes
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Image Processing (AREA)
Abstract
Translated fromChineseDescription
Technical field
The present invention relates to a kind of bottle wall defect detection method and device, especially a kind of empty bottle wall defect detection method and device.
Background technology
At present, beer, beverage production enterprise more and more pay attention to the quality of product, yet the quality of bottle is difficult to reach the set goal on enterprise's production line, usually has defectives such as crackle.Especially the returnable bottle that utilizes again of repeated washing has the trade mark that does not wash to adhere on the bottle wall easily, therefore need comprehensively detect the bottle wall.Enterprise adopts manual detection more now, and bottle is when being installed in the other light test box in carrier chain road, and underproof artificial removal is found in naked-eye observation.There is following shortcoming in this traditional detection method:
1. accuracy of detection is low, can not satisfy the requirement of client to product quality, difficult tested the finding of tiny flaw, and the security of packing to food brings hidden danger;
2. detection efficiency is low, can not satisfy the demand of automatic production line high-speed production;
3. testing staff's fatiguability can not keep examination criteria;
4. manual work is unfavorable for the integrated of the information that realizes.
Existing detection technique, must make the glass bottle and jar on the production line detect station suitably pause and rotation at least 360 degree again as Chinese patent 02133618.0 " glass bottle and jar detection method and glass bottle and jar pick-up unit ", obviously do not meet the requirement of production line high-speed operation, infeasible in actual applications.Chinese patent 200710028027.1 " a kind of detection method of empty bottle mouth defect and device " detection method can not be transplanted to the detection of bottle wall because of the characteristics of bottle wall self: adopt light source can obtain circular plane bottleneck image over against the method for reflection, but it is inapplicable that the three-dimensional bottle of unconspicuous 360 degree wall is obviously reflected in transmission, and the style characteristic difference of bottleneck bottle wall image also makes disposal route have different separately characteristics.
Summary of the invention
The objective of the invention is for overcoming above-mentioned the deficiencies in the prior art, a kind of easy realization robotization is provided, discerns accurately, satisfy empty bottle wall defect detection method and device that high-speed production lines requires.
For achieving the above object, the present invention adopts following technical proposals:
A kind of empty bottle wall defect detection method may further comprise the steps:
A. the marginal point that scans on the bottleneck is right, utilizes the method for the definite bottle of symmetry analysis binding site relation wall each several part central axis to carry out a bottle wall location, obtains the position of body processing region, bottle wall surveyed area is divided carried out subarea processing;
B. to the positioned area by adopting grey stretching method to pre-processing image data, increase the contrast and the brightness of image;
C. adopt maximum variance between clusters that image is cut apart, obtain target information;
D. the bottle wall image after over-segmentation is carried out connectivity analysis, extract the characteristic of each defective, judge according to centroid position, figure's ratio and the area features of connected domain whether each detected connected domain is real defective.
Bottle wall basis on location bottleneck edge in the described steps A positions tracking, and concrete tracking step is as follows:
1) because in side wall image, only produce fluctuation in the horizontal direction, thus when the location horizontal ordinate of a computing center, at first determine one group of n bar (n=5 altogether, 6,, 20) and the position of scanning straight line, guarantees that every scans the both sides of the edge that straight line can both pass bottleneck;
2) set up some array P0, P1 and P2;
3) carry out bilateral scanning along every scanning straight line, suddenly change the from low to high point of intensity maximum of gray-scale value on the search straight line, with on every straight line from left to right scanning and from right to left the point that obtains of scanning be designated as respectively
,
, i=1 is to the integer of n, with point
Order leaves among the array P1 point in
Order leaves among the array P2;
4) utilize formula (1) promptly to a pair of marginal point on every scanning straight line
,
Horizontal ordinate
With
Average, utilize formula (2) to ask distance between every pair of marginal point, be designated as
, calculate a group switching centre point horizontal ordinate reference value by these paired marginal points that scan
(1)
5) combination
The distribution situation of analytic centre's point horizontal ordinate reference value: each position
All should with the corresponding scope of detection bottle in fluctuate, if exceeded this scope, illustrate that this is the pseudo-edge point to having a point in the marginal point at least, this moment can remove this horizontal ordinate reference value to the central point of a correspondence;
6) in the reference center point set of finally choosing, obtain the final calculated value of central point by the method for averaging, determine the position of body processing region then.
Grey stretching method among the described step B may further comprise the steps pre-processing image data: parameter is seen Fig. 7, and the x axle is the gray-scale value before the conversion, and the y axle is the gray-scale value after the conversion.
1) determine the greyscale transformation function, establish (
,
), (
,
) be the point of determining on the transforming function transformation function, certain gray values of pixel points is respectively Gray1 and Gray2 in the image of conversion front and back;
2) when Gray1 0～
Between the time, the order
3) exist as Gray1
～
Between the time, the order
4) exist as Gray1
In the time of between～255, order
It is 0 rank and the 1 rank square that utilizes grey level histogram that maximum variance between clusters among the described step C is cut apart image, dynamically determines the image segmentation threshold value according to the maximum variance between target and the background,
If the pixel number of piece image is N, it has L(L is natural number) individual gray level (0,1 ..., L-1), gray level is that the pixel number of i is
, so, to image histogram normalization, probability density distribution is arranged:
Wherein
, if entire image is with gray level t(0＜t＜L-1) as threshold value, establish threshold value t image is divided into
WithTwo classes,
With
Represent object and background respectively, and
With
Difference correspondinggrey scale level 0,1 ... t } and t+1, t+2 ... L-1 } pixel,With
The probability that class takes place is respectively:
(5)
Wherein,
With
Average be respectively:
(6)
Wherein
,, can verify the following formula establishment,
(8)
Can define the class internal variance at this:
(9)
Inter-class variance:
(10)
According to the maximum between-cluster variance criterion, obtain optimum threshold value gray level
Need satisfy following formula:
It is optimal threshold
Make the inter-class variance maximum.
Whether each the detected connected domain of judging among the described step D is that real defective is:
At first extract the parameter that the connected domain algorithm returns, set up a template pixel, promptly the pel array that breach returned of empty bottle wall 1mm * 1mm is H * W, and this array is a normal value in fixing camera system;
The breach corresponding parameters template of corresponding empty bottle wall Amm * Bmm size is (A * H) * (B * W);
The pixels tall coefficient that returns then, return the pixel wide coefficient, return the elemental area coefficient, return the pixel-intensive degree and compare with the examination criteria of setting, if the parameter of extracting is not in critical field, then think the defective image of bottle wall, decision signal is passed to controller, is rejected by device for eliminating.
A kind of empty bottle wall defect pick-up unit, comprise inlet bottle wall detecting unit, outlet bottle wall detecting unit and speed difference connecting gear, described inlet bottle wall detecting unit and outlet bottle wall detecting unit lay respectively at the two ends of speed difference connecting gear, described inlet bottle wall detecting unit and outlet bottle wall detecting unit comprise photoelectric sensor respectively, camera, backlight and tertiary reflex mirror, photoelectric sensor links to each other with controller, controller is connected with the rejecting mechanism of outlet behind the detecting unit, backlight is arranged at detected bottle wall one side, tertiary reflex mirror and camera are arranged at the opposite side of detected bottle wall, the tertiary reflex mirror is corresponding with detected bottle wall and camera respectively, camera and image processing system are electrically connected, and image processing system is electrically connected with controller.
Described speed difference connecting gear comprises two parallel travelling belts up and down, the equal diameters of the distance between two travelling belts and detected bottle, and two travelling belts are arranged at respectively on the different belt wheels, and belt wheel is connected with motor by transmission shaft.
Described controller is industrial computer CPU.
Described backlight is the led light source that is connected to the stroboscopic controller, the led light source color tunable: select white light during white bottle for use, green bottle is with green, palm fibre bottle red light source.
Described image processing system is the industrial computer that has image pick-up card.
This detection system adopts the flat LED light source to carry out back lighting in bottle one side in the bottle wall detects, image carries out realizing that the multi-angle of bottle wall is Polaroid after the tertiary reflex through the tertiary reflex mirror: by each mirror angle in the first order reflector group, collect the image of the bottle of many group different angles, image enters camera through second level reflector group, the reflection of third level catoptron two-stage successively then, and system has realized the clear and near undistorted image by a camera collecting bottle wall circumference 240 degree thus.The design of tertiary reflex mirror has reduced the acquisition system occupation space, makes the total system structure compact more, has strengthened the integration of system.
Tertiary reflex mirror structure comprises first order reflection mirror, secondary reflex mirror, tertiary reflex mirror, empty bottle is positioned on the incident direction of first order reflection mirror, the first order reflection mirror is positioned on the incident direction of secondary reflex mirror, the secondary reflex mirror is positioned on the incident direction of tertiary reflex mirror, camera is positioned on the reflection direction of tertiary reflex mirror, the tertiary reflex mirror is positioned on the reflection direction of secondary reflex mirror, and the secondary reflex mirror is positioned on the reflection direction of first order reflection mirror.The empty bottle wall reflected light enters the first order reflection mirror, enters the secondary reflex mirror by the first order reflection mirror reflection, enters the tertiary reflex mirror by the secondary reflex mirror reflection again, enters camera by the tertiary reflex mirror reflection then, catches the empty bottle wall image by camera.The first order reflection mirror comprises two groups, and every group has two parallel staggered catoptrons, and two groups of both sides that are positioned at the secondary reflex mirror respectively become with the empty bottle direction of transfer
() angle, the secondary reflex mirror comprises two catoptrons, becomes between two catoptrons
(
) angle, each becomes two secondary reflex mirrors with the first order reflection mirror of this side
(
) angle, the tertiary reflex mirror plane is parallel with the empty bottle direction of transfer, becomes with ground
() angle, camera is positioned at the top of tertiary reflex mirror, becomes with the tertiary reflex mirror() angle.
When arriving speed difference translator unit, empty bottle is clamped by belt, unsettled transmission.Because two belts have certain velocity contrast, empty bottle can rotate in this part transmission course, the belt speed difference that is provided with just makes empty bottle revolve after by this part and turn 90 degrees, and the detection blind area of inlet bottle wall test section is presented in the surveyed area of outlet bottle wall fully.
Empty bottle is realized 90 degree rotations in speed difference conveyer principle as shown in Figure 3, the length velocity relation of belt conveyor is V among the figure1V2, and become fixed proportion.Turn 90 degrees thereby realized empty bottle accurately revolved.
Detection method provided by the invention can be applied in the checkout equipment of industrial flow-line easily, thereby realizes the automatic high-speed of empty bottle wall defect is accurately detected.Pick-up unit provided by the invention is simple in structure, can realize accurately the empty bottle that has bottle wall defect is rejected automatically at a high speed.Judge that with graphical analysis bottle wall defect makes equipment simple and reliable.The higher order reflection structure can make camera once obtain the image information of empty bottle 240 degree, behind connecting gear, empty bottle revolves and turn 90 degrees, and inlet bottle wall detects and outlet bottle wall detects with regard to passable to complete image information like this, and this method can be finished and handle fast and judge bottle wall defect information.This device highest detection speed is 72000 bottles/hour, can detect in the surveyed area minimum dimension and be long 4mm, and the dirt of wide 4mm, the rejecting rate of defective empty bottle is more than 99.95%, the mistake rejecting rate of qualified empty bottle is controlled at below 0.3%.
Description of drawings
Fig. 1 and Fig. 2 are respectively the front elevation and the left views of speed difference connecting gear;
Fig. 3 is a bottle wall pick-up unit work synoptic diagram
Fig. 4 is bottle wall three part division figure;
Fig. 5 is a bottle wall higher order reflection floor map;
Fig. 6 is a bottle wall higher order reflection schematic perspective view;
Fig. 7 is the grey stretching method synoptic diagram;
Fig. 8 is the entire system block diagram;
Wherein, 1. drive pulley, 2. travelling belt, 3. an inlet bottle wall detecting unit, 4. export bottle wall detecting unit, 5. speed difference connecting gear, 6. empty bottle, 7. belt speed V1,8. belt speed V2,9. bottleneck, 10. shoulder, 11, body, 12, camera, 13. third level catoptrons, 14. first order catoptrons, 15. second level catoptron, 16. backlights, 17. empty bottle working direction.
Embodiment
The present invention is further described below in conjunction with drawings and Examples.
Shown in Fig. 1-8, the hardware construction of system is made up of travelling belt 2, photoelectric sensor,CCD camera 12, image processing system, industrial computer,backlight 16 and supporting stroboscopic controller thereof.
The principle of work of this device is:
Large-area flat-plate LED-backlit source 16 is adopted in the illumination of bottle wall, places bottle wall one side, and adjustable color light passes a bottle wall at a certain angle, andCCD camera 12 is on the light source opposite, and the opposite side of bottle is by multistage facetted mirrors collecting bottle wall image.The detection of bottle wall is made up of an inlet bottlewall detecting unit 3 and an outlet bottlewall detecting unit 4, is responsible for entrance and exit bottle wall image acquisition and processing respectively, eliminates the bottle wall and detects the comprehensive detection of blind area realization to empty bottle.View data is delivered to industrial computer and is carried out analysis and judgement, obtains being used to distinguish qualified empty bottle signal with defective and passes through control panel card control device for eliminating.Simultaneously, the software of man-machine interface is integrated data, realizes various functions such as data presentation, setting, filing.
The empty bottle wall defect pick-up unit, comprise inlet bottlewall detecting unit 3, outlet bottlewall detecting unit 4 and speed difference connecting gear 5, described inlet bottlewall detecting unit 3 and outlet bottlewall detecting unit 4 lay respectively at the two ends of speed difference connecting gear 5, described inlet bottlewall detecting unit 3 and outlet bottlewall detecting unit 4 comprise photoelectric sensor respectively,camera 12,backlight 16 and tertiary reflex mirror, photoelectric sensor links to each other with controller, controller is connected with the rejecting mechanism of outlet behind the detectingunit 4,backlight 16 is arranged at detected bottle wall one side, tertiary reflex mirror andcamera 12 are arranged at the opposite side of detected bottle wall, the tertiary reflex mirror is corresponding with detected bottle wall andcamera 12 respectively,camera 12 and image processing system are electrically connected, and image processing system is electrically connected with controller.
Described speed difference connecting gear 5 comprises up and down the equal diameters of distance and detected bottle between two parallel travelling belt 2, two travelling belts 2, and two travelling belts 2 are arranged at respectively on the different drive pulleys 1, and drivepulley 1 is connected with motor by transmission shaft.
Described controller is industrial computer CPU.
Describedbacklight 16 is the led light sources that are connected to the stroboscopic controller, the led light source color tunable: select white light during white bottle for use, green bottle is with green, palm fibre bottle red light source.
Described image processing system is the industrial computer that has image pick-up card.
This detection system adopts the flat LED light source to carry out back lighting in bottle one side in the bottle wall detects, image carries out realizing that the multi-angle of bottle wall is Polaroid after the tertiary reflex through the tertiary reflex mirror, each mirror angle in thefirst order catoptron 14, collect the image (image that comprisesbottleneck 9,shoulder 10 and body 11) of the bottle of many group different angles, this image is delivered to camera lens throughsecond level catoptron 15 andthird level catoptron 13 again, and system has realized the clear and near undistorted image by a camera collecting bottle wall circumference 240 degree thus.The design of tertiary reflex mirror has reduced the acquisition system occupation space, makes the total system structure compact more, has strengthened the integration of system.
Tertiary reflex mirror structure comprises first order reflection mirror 14, secondary reflex mirror 15, tertiary reflex mirror 13, empty bottle is positioned on the incident direction of first order reflection mirror 14, first order reflection mirror 14 is positioned on the incident direction of secondary reflex mirror 15, secondary reflex mirror 15 is positioned on the incident direction of tertiary reflex mirror 13, camera 12 is positioned on the reflection direction of tertiary reflex mirror 13, tertiary reflex mirror 13 is positioned on the reflection direction of secondary reflex mirror 15, and secondary reflex mirror 15 is positioned on the reflection direction of first order reflection mirror 14.The empty bottle wall reflected light enters first order reflection mirror 14, enter secondary reflex mirror 15 by 14 reflections of first order reflection mirror, enter tertiary reflex mirror 13 by 15 reflections of secondary reflex mirror again, enter camera 12 by 13 reflections of tertiary reflex mirror then, catch the empty bottle wall image by camera 12.First order reflection mirror 14 comprises two groups, and every group has two parallel staggered catoptrons, and two groups of both sides that are positioned at secondary reflex mirror 14 respectively become with the empty bottle direction of transfer
() angle, secondary reflex mirror 15 comprises two catoptrons, becomes between two catoptrons
(
) angle, each becomes two secondary reflex mirrors with the first order reflection mirror of this side
(
) angle, the tertiary reflex mirror plane is parallel with the empty bottle direction of transfer, becomes with ground(
) angle, camera 12 is positioned at the top of tertiary reflex mirror 13, with 13 one-tenth in tertiary reflex mirror
(
) angle.
When arriving speed difference translator unit, empty bottle is transmitted is with 2 to clamp, unsettled transmission.Because two transmission have certain velocity contrast,empty bottle 6 can rotate in this part transmission course, travelling belt 2 velocity contrasts that are provided with just make empty bottle revolve after by this part and turn 90 degrees, and the detection blind area of inlet bottle wall test section is presented in the surveyed area of outlet bottle wall fully.
Empty bottle is realized 90 degree rotations in speed difference conveyer principle as shown in Figure 2, the length velocity relation along emptybottle working direction 17 belt conveyors among the figure is belt speed V17〉belt speed V28, and become fixed proportion.Turn 90 degrees thereby realizedempty bottle 6 accurately revolved.
Empty bottle wall defect detection method may further comprise the steps:
A. the marginal point that scans on thebottleneck 9 is right, utilizes the method for the definite bottle of symmetry analysis binding site relation wall each several part central axis to carry out a bottle wall location, obtains the position ofbody 11 processing regions, bottle wall surveyed area is divided carried out subarea processing;
B. to the positioned area by adopting grey stretching method to pre-processing image data, increase the contrast and the brightness of image;
C. adopt maximum variance between clusters that image is cut apart, obtain target information;
D. the bottle wall image after over-segmentation is carried out connectivity analysis, extract the characteristic of each defective, judge according to centroid position, figure's ratio and the area features of connected domain whether each detected connected domain is real defective.
Bottle wall basis on location bottleneck edge in the described steps A positions tracking, and concrete tracking step is as follows:
1) because in side wall image, only produce fluctuation in the horizontal direction, thus when the location horizontal ordinate of a computing center, at first determine one group of n bar (n=5 altogether, 6,, 20) and the position of scanning straight line, guarantees that every scans the both sides of the edge that straight line can both passbottleneck 9;
2) set up some array P0, P1 and P2;
3) carry out bilateral scanning along every scanning straight line, suddenly change the from low to high point of intensity maximum of gray-scale value on the search straight line, with on every straight line from left to right scanning and from right to left the point that obtains of scanning be designated as respectively
,
, i=1 is to the integer of n, with point
Order leaves among the array P1 point in
Order leaves among the array P2;
4) utilize formula (1) promptly to a pair of marginal point on every scanning straight line
,
Horizontal ordinate
With
Average, utilize formula (2) to ask distance between every pair of marginal point, be designated as
, calculate a group switching centre point horizontal ordinate reference value by these paired marginal points that scan
(2)
5) combination
The distribution situation of analytic centre's point horizontal ordinate reference value: each position
All should with the corresponding scope of detection bottle in fluctuate, if exceeded this scope, illustrate that this is the pseudo-edge point to having a point in the marginal point at least, this moment can remove this horizontal ordinate reference value to the central point of a correspondence;
6) in the reference center point set of finally choosing, obtain the final calculated value of central point by the method for averaging, determine the position of body processing region then.
Grey stretching method among the described step B may further comprise the steps pre-processing image data: parameter is seen Fig. 7, and the x axle is the gray-scale value before the conversion, and the y axle is the gray-scale value after the conversion.
1) determine the greyscale transformation function, establish (
,
), (
,
) be the point of determining on the transforming function transformation function, certain gray values of pixel points is respectively Gray1 and Gray2 in the image of conversion front and back;
2) when Gray1 0～
Between the time, the order
3) exist as Gray1
～
Between the time, the order
4) exist as Gray1
In the time of between～255, order
It is 0 rank and the 1 rank square that utilizes grey level histogram that maximum variance between clusters among the described step C is cut apart image, dynamically determines the image segmentation threshold value according to the maximum variance between target and the background,
If the pixel number of piece image is N, it has L(L is natural number) individual gray level (0,1 ..., L-1), gray level is that the pixel number of i is
, so
, to image histogram normalization, probability density distribution is arranged:
Wherein
, if entire image is with gray level t(0＜t＜L-1) as threshold value, establish threshold value t image is divided into
With
Two classes,
WithRepresent object and background respectively, and
With
Difference correspondinggrey scale level 0,1 ... t } and t+1, t+2 ... L-1 } pixel,
With
The probability that class takes place is respectively:
Wherein
,
With
Average be respectively:
(6)
Wherein,
, can verify the following formula establishment,
(8)
Can define the class internal variance at this:(9)
Inter-class variance:
(10)
According to the maximum between-cluster variance criterion, obtain optimum threshold value gray level
Need satisfy following formula:
It is optimal threshold
Make the inter-class variance maximum.
Whether each the detected connected domain of judging among the described step D is that real defective is:
At first extract the parameter that the connected domain algorithm returns, set up a template pixel, promptly the pel array that breach returned of empty bottle wall 1mm * 1mm is H * W, and this array is a normal value in fixing camera system;
The breach corresponding parameters template of corresponding empty bottle wall Amm * Bmm size is (A * H) * (B * W);
The pixels tall coefficient that returns then, return the pixel wide coefficient, return the elemental area coefficient, return the pixel-intensive degree and compare with the examination criteria of setting, if the parameter of extracting is not in critical field, then think the defective image of bottle wall, decision signal is passed to controller, is rejected by device for eliminating.
Claims (10)
1. an empty bottle wall defect detection method is characterized in that, may further comprise the steps:
A. the marginal point that scans on the bottleneck is right, utilizes the method for the definite bottle of symmetry analysis binding site relation wall each several part central axis to carry out a bottle wall location, obtains the position of body processing region, bottle wall surveyed area is divided carried out subarea processing;
B. to the positioned area by adopting grey stretching method to pre-processing image data, increase the contrast and the brightness of image;
C. adopt maximum variance between clusters that image is cut apart, obtain target information;
D. the bottle wall image after over-segmentation is carried out connectivity analysis, extract the characteristic of each defective, judge according to centroid position, figure's ratio and the area features of connected domain whether each detected connected domain is real defective.
2. empty bottle wall defect detection method according to claim 1 is characterized in that: the bottle wall basis on location bottleneck edge in the described steps A positions tracking, and concrete tracking step is as follows:
1) because in side wall image, only produce fluctuation in the horizontal direction, thus when the location horizontal ordinate of a computing center, at first determine one group of position of n bar scanning straight line altogether, wherein, n is 5 to 20 integer; Guarantee that every scanning straight line can both pass the both sides of the edge of bottleneck;
2) set up some array P0, P1 and P2;
3) carry out bilateral scanning along every scanning straight line, suddenly change the from low to high point of intensity maximum of gray-scale value on the search straight line, with on every straight line from left to right scanning and from right to left the point that obtains of scanning be designated as respectively
,
, i=1 is to the integer of n, with point
Order leaves among the array P1 point in
Order leaves among the array P2;
4) utilize formula (1) promptly to a pair of marginal point on every scanning straight line
,
Horizontal ordinate
WithAverage, utilize formula (2) to ask distance between every pair of marginal point, be designated as, calculate a group switching centre point horizontal ordinate reference value by these paired marginal points that scan
5) combination
The distribution situation of analytic centre's point horizontal ordinate reference value: each positionAll should with the corresponding scope of detection bottle in fluctuate, if exceeded this scope, illustrate that this is the pseudo-edge point to having a point in the marginal point at least, this moment can remove this horizontal ordinate reference value to the central point of a correspondence;
6) in the reference center point set of finally choosing, obtain the final calculated value of central point by the method for averaging, determine the position of body processing region then.
3. empty bottle wall defect detection method according to claim 1 is characterized in that: the grey stretching method among the described step B may further comprise the steps pre-processing image data: the x axle is the gray-scale value before the conversion, and the y axle is the gray-scale value after the conversion;
1) determine the greyscale transformation function, establish (,
), (
,
) be the point of determining on the transforming function transformation function, certain gray values of pixel points is respectively Gray1 and Gray2 in the image of conversion front and back;
2) when Gray1 0～
Between the time, the order
3) exist as Gray1
～
Between the time, the order
4) exist as Gray1In the time of between～255, order
4. empty bottle wall defect detection method according to claim 1, it is characterized in that: it is 0 rank and the 1 rank square that utilizes grey level histogram that the maximum variance between clusters among the described step C is cut apart image, dynamically determine the image segmentation threshold value according to the maximum variance between target and the background
If the pixel number of piece image is N, it has L gray level is 0,1 ..., L-1, L are natural number, gray level is that the pixel number of i is
, so
, to image histogram normalization, probability density distribution is arranged:
Wherein
, if entire image with gray level t as threshold value, wherein, 0＜t＜L-1 establishes threshold value t image is divided into
With
Two classes,
With
Represent object and background respectively, andWith
Difference corresponding grey scale level 0,1 ... t } and t+1, t+2 ... L-1 } pixel,
With
The probability that class takes place is respectively:
Wherein
,With
Average be respectively:
Wherein
,
, can verify the following formula establishment,
Can define the class internal variance at this:
Inter-class variance:
According to the maximum between-cluster variance criterion, obtain optimum threshold value gray level
Need satisfy following formula:
It is optimal threshold
Make the inter-class variance maximum.
5. empty bottle wall defect detection method according to claim 1 is characterized in that: whether each the detected connected domain of judging among the described step D is that real defective is:
At first extract the parameter that the connected domain algorithm returns, set up a template pixel, promptly the pel array that breach returned of empty bottle wall 1mm * 1mm is H * W, and this array is a normal value in fixing camera system;
The breach corresponding parameters template of corresponding empty bottle wall Amm * Bmm size is (A * H) * (B * W);
Then to the pixels tall coefficient that returns, return the pixel wide coefficient, return the elemental area coefficient, return the pixel-intensive degree and compare with the examination criteria of setting, if the parameter of extracting is not in critical field, then think the defective image of bottle wall, decision signal is passed to controller, is rejected by device for eliminating.
6. empty bottle wall defect pick-up unit, it is characterized in that: comprise inlet bottle wall detecting unit, outlet bottle wall detecting unit and speed difference connecting gear, described inlet bottle wall detecting unit and outlet bottle wall detecting unit lay respectively at the two ends of speed difference connecting gear, described inlet bottle wall detecting unit and outlet bottle wall detecting unit comprise photoelectric sensor respectively, camera, backlight and tertiary reflex mirror, photoelectric sensor links to each other with controller, controller is connected with the rejecting mechanism of outlet behind the detecting unit, backlight is arranged at detected bottle wall one side, tertiary reflex mirror and camera are arranged at the opposite side of detected bottle wall, the tertiary reflex mirror is corresponding with detected bottle wall and camera respectively, camera and image processing system are electrically connected, and image processing system is electrically connected with controller.
7. empty bottle wall defect pick-up unit according to claim 6, it is characterized in that, described speed difference connecting gear comprises two parallel travelling belts up and down, the equal diameters of the distance between two travelling belts and detected bottle, two travelling belts are arranged at respectively on the different belt wheels, and belt wheel is connected with motor by transmission shaft.
8. empty bottle wall defect pick-up unit according to claim 6 is characterized in that: described controller is industrial computer CPU; Described image processing system is the industrial computer that has image pick-up card.
9. empty bottle wall defect pick-up unit according to claim 6 is characterized in that: described backlight is the led light source that is connected to the stroboscopic controller, the led light source color tunable: select white light during white bottle for use, green bottle is with green, palm fibre bottle red light source.
10. empty bottle wall defect pick-up unit according to claim 6, it is characterized in that: described tertiary reflex mirror comprises first order reflection mirror, secondary reflex mirror, tertiary reflex mirror, the incident direction of first order reflection mirror faces empty bottle, the first order reflection mirror is positioned on the incident direction of secondary reflex mirror, the secondary reflex mirror is positioned on the incident direction of tertiary reflex mirror, and camera is positioned on the reflection direction of tertiary reflex mirror; Described first order reflection mirror comprises two groups, and every group has two parallel staggered catoptrons, and two groups of both sides that are positioned at the secondary reflex mirror respectively become with the empty bottle direction of transfer
Angle, wherein,
The secondary reflex mirror comprises two catoptrons, becomes between two catoptrons
Angle, wherein,
Each becomes described two secondary reflex mirrors with the first order reflection mirror of this side
Angle, wherein,
The tertiary reflex mirror plane is parallel with the empty bottle direction of transfer, becomes with groundAngle, wherein,
Camera is positioned at the top of tertiary reflex mirror, becomes with the tertiary reflex mirror
Angle, wherein,
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010170952ACN101819162A (en) | 2010-05-13 | 2010-05-13 | Empty bottle wall defect detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010170952ACN101819162A (en) | 2010-05-13 | 2010-05-13 | Empty bottle wall defect detection method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101819162Atrue CN101819162A (en) | 2010-09-01 |
Family
ID=42654337
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010170952APendingCN101819162A (en) | 2010-05-13 | 2010-05-13 | Empty bottle wall defect detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101819162A (en) |
Cited By (33)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101825582A (en)* | 2010-05-19 | 2010-09-08 | 山东明佳包装检测科技有限公司 | Method and device for detecting wall of cylindrical transparent bottle |
| CN102213681A (en)* | 2011-04-01 | 2011-10-12 | 哈尔滨工业大学（威海） | Novel method for detecting sewage in anti-skidding region at bottom of glass bottle |
| CN102682293A (en)* | 2012-05-14 | 2012-09-19 | 山东大学 | Method and system for identifying salient-point mould number of revolution-solid glass bottle based on images |
| CN102708368A (en)* | 2012-05-04 | 2012-10-03 | 湖南大学 | Method for positioning bottle bodies on production line based on machine vision |
| CN102842131A (en)* | 2012-07-10 | 2012-12-26 | 中联重科股份有限公司 | Method and equipment for monitoring defects of target object |
| CN102998316A (en)* | 2012-12-20 | 2013-03-27 | 山东大学 | Transparent liquid impurity detection system and detection method thereof |
| CN103364414A (en)* | 2012-04-05 | 2013-10-23 | 庞红斌 | Special optical system for detecting bodies of glass bottles |
| CN103630542A (en)* | 2012-08-27 | 2014-03-12 | Ntn株式会社 | Defect detecting apparatus, defect correction device and defect detecting method |
| CN103698343A (en)* | 2013-12-30 | 2014-04-02 | 上海瑞伯德智能系统科技有限公司 | Pencil nib defect detection device |
| CN103718026A (en)* | 2011-06-10 | 2014-04-09 | Khs有限责任公司 | Empty bottle inspection |
| CN103884650A (en)* | 2014-03-28 | 2014-06-25 | 北京大恒图像视觉有限公司 | Multi-photosource linear array imaging system and method |
| CN104535006A (en)* | 2015-01-21 | 2015-04-22 | 杭州电子科技大学 | Bottle cap gap width estimation method by using transmission type illuminating and imaging system |
| CN104657977A (en)* | 2014-12-10 | 2015-05-27 | 天津普达软件技术有限公司 | Method for positioning center of bottle body of bottle parison |
| CN105073285A (en)* | 2013-03-14 | 2015-11-18 | 费南泰克控股有限公司 | Device and method for transporting and examining fast-moving objects to be treated |
| CN105301010A (en)* | 2015-09-23 | 2016-02-03 | 广东暨通信息发展有限公司 | Bottle body defect detection device for glass bottle |
| CN106611402A (en)* | 2015-10-23 | 2017-05-03 | 腾讯科技（深圳）有限公司 | Image processing method and device |
| CN106651837A (en)* | 2016-11-14 | 2017-05-10 | 中国科学院自动化研究所 | White glass plate surface edge breakage defect detecting method |
| CN106705839A (en)* | 2016-12-07 | 2017-05-24 | 广州道注塑机械股份有限公司 | Fast moving bottle pre-form size precision measuring device |
| CN106705840A (en)* | 2016-12-07 | 2017-05-24 | 广州道注塑机械股份有限公司 | Bottle pre-form size fast measuring device |
| CN107064724A (en)* | 2012-03-01 | 2017-08-18 | 株式会社东芝 | Defect detecting device and defect inspection method |
| CN107107121A (en)* | 2014-12-12 | 2017-08-29 | 诺尔顿沃特福德有限公司 | Be recessed detection device and method |
| CN108375579A (en)* | 2018-02-21 | 2018-08-07 | 重庆环视高科技有限公司 | A kind of medicine bottle detection method |
| CN108709892A (en)* | 2018-07-04 | 2018-10-26 | 杭州智感科技有限公司 | Detecting system and its method |
| CN108802055A (en)* | 2018-08-22 | 2018-11-13 | 苏州西斯派克检测科技有限公司 | A kind of translucent or opaque empty bottle outer wall testing agency |
| CN108872258A (en)* | 2018-06-25 | 2018-11-23 | 成都市金鼓药用包装有限公司 | A kind of empty bottle method of inspection and its empty bottle checking machine |
| CN109064439A (en)* | 2018-06-15 | 2018-12-21 | 杭州舜浩科技有限公司 | Single-sided illumination formula light guide plate shadow defect extracting method based on subregion |
| CN109525840A (en)* | 2018-12-18 | 2019-03-26 | 凌云光技术集团有限责任公司 | The detection method of minor defect on a kind of imager chip |
| CN109615606A (en)* | 2018-11-09 | 2019-04-12 | 华南理工大学 | A rapid classification method for point, line and surface defects of flexible IC substrates |
| CN109632808A (en)* | 2018-12-05 | 2019-04-16 | 深圳大学 | Seamed edge defect inspection method, device, electronic equipment and storage medium |
| CN110517233A (en)* | 2019-08-15 | 2019-11-29 | 浙江赤霄智能检测技术有限公司 | A kind of defect classification learning system and its classification method based on artificial intelligence |
| CN111351754A (en)* | 2020-04-23 | 2020-06-30 | 广州番禺职业技术学院 | Bottle bottom defect detection system and method |
| CN111808367A (en)* | 2020-07-09 | 2020-10-23 | 浙江七色鹿色母粒有限公司 | Improvement method for plastic PPR silver grain whitening defect |
| CN118864437A (en)* | 2024-08-13 | 2024-10-29 | 中山市迪朗食品科技有限公司 | A method, device and medium for visually detecting beverage packaging defects |
- 2010
- 2010-05-13CNCN201010170952Apatent/CN101819162A/enactivePending
Cited By (47)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101825582A (en)* | 2010-05-19 | 2010-09-08 | 山东明佳包装检测科技有限公司 | Method and device for detecting wall of cylindrical transparent bottle |
| CN102213681B (en)* | 2011-04-01 | 2013-01-02 | 哈尔滨工业大学（威海） | Novel method for detecting sewage in anti-skidding region at bottom of glass bottle |
| CN102213681A (en)* | 2011-04-01 | 2011-10-12 | 哈尔滨工业大学（威海） | Novel method for detecting sewage in anti-skidding region at bottom of glass bottle |
| CN103718026B (en)* | 2011-06-10 | 2017-10-10 | Khs有限责任公司 | Empty bottle is checked |
| CN103718026A (en)* | 2011-06-10 | 2014-04-09 | Khs有限责任公司 | Empty bottle inspection |
| CN107064724A (en)* | 2012-03-01 | 2017-08-18 | 株式会社东芝 | Defect detecting device and defect inspection method |
| CN103364414A (en)* | 2012-04-05 | 2013-10-23 | 庞红斌 | Special optical system for detecting bodies of glass bottles |
| CN102708368B (en)* | 2012-05-04 | 2014-01-15 | 湖南大学 | A method for positioning bottles on a production line based on machine vision |
| CN102708368A (en)* | 2012-05-04 | 2012-10-03 | 湖南大学 | Method for positioning bottle bodies on production line based on machine vision |
| CN102682293A (en)* | 2012-05-14 | 2012-09-19 | 山东大学 | Method and system for identifying salient-point mould number of revolution-solid glass bottle based on images |
| CN102842131A (en)* | 2012-07-10 | 2012-12-26 | 中联重科股份有限公司 | Method and equipment for monitoring defects of target object |
| CN102842131B (en)* | 2012-07-10 | 2015-04-29 | 中联重科股份有限公司 | Method and equipment for monitoring defects of target object |
| CN103630542A (en)* | 2012-08-27 | 2014-03-12 | Ntn株式会社 | Defect detecting apparatus, defect correction device and defect detecting method |
| CN102998316A (en)* | 2012-12-20 | 2013-03-27 | 山东大学 | Transparent liquid impurity detection system and detection method thereof |
| CN105073285B (en)* | 2013-03-14 | 2018-01-19 | 费南泰克控股有限公司 | Device and method for transporting and inspecting fast-moving pending goods |
| CN105073285A (en)* | 2013-03-14 | 2015-11-18 | 费南泰克控股有限公司 | Device and method for transporting and examining fast-moving objects to be treated |
| CN103698343A (en)* | 2013-12-30 | 2014-04-02 | 上海瑞伯德智能系统科技有限公司 | Pencil nib defect detection device |
| CN103884650A (en)* | 2014-03-28 | 2014-06-25 | 北京大恒图像视觉有限公司 | Multi-photosource linear array imaging system and method |
| CN104657977A (en)* | 2014-12-10 | 2015-05-27 | 天津普达软件技术有限公司 | Method for positioning center of bottle body of bottle parison |
| CN104657977B (en)* | 2014-12-10 | 2017-09-22 | 天津普达软件技术有限公司 | A kind of method for positioning bottle base body center |
| CN107107121A (en)* | 2014-12-12 | 2017-08-29 | 诺尔顿沃特福德有限公司 | Be recessed detection device and method |
| CN104535006B (en)* | 2015-01-21 | 2017-10-27 | 杭州电子科技大学 | A kind of bottle cap gap width evaluation method of utilization transmission-type illumination imaging systems |
| CN104535006A (en)* | 2015-01-21 | 2015-04-22 | 杭州电子科技大学 | Bottle cap gap width estimation method by using transmission type illuminating and imaging system |
| CN105301010A (en)* | 2015-09-23 | 2016-02-03 | 广东暨通信息发展有限公司 | Bottle body defect detection device for glass bottle |
| CN106611402A (en)* | 2015-10-23 | 2017-05-03 | 腾讯科技（深圳）有限公司 | Image processing method and device |
| CN106611402B (en)* | 2015-10-23 | 2019-06-14 | 腾讯科技（深圳）有限公司 | Image processing method and device |
| CN106651837A (en)* | 2016-11-14 | 2017-05-10 | 中国科学院自动化研究所 | White glass plate surface edge breakage defect detecting method |
| CN106651837B (en)* | 2016-11-14 | 2019-10-15 | 中国科学院自动化研究所 | Detection method for chipping defects on the surface of white glass plate |
| CN106705840A (en)* | 2016-12-07 | 2017-05-24 | 广州道注塑机械股份有限公司 | Bottle pre-form size fast measuring device |
| CN106705839A (en)* | 2016-12-07 | 2017-05-24 | 广州道注塑机械股份有限公司 | Fast moving bottle pre-form size precision measuring device |
| CN108375579A (en)* | 2018-02-21 | 2018-08-07 | 重庆环视高科技有限公司 | A kind of medicine bottle detection method |
| CN108375579B (en)* | 2018-02-21 | 2021-05-07 | 重庆环视高科技有限公司 | Medicine bottle detection method |
| CN109064439A (en)* | 2018-06-15 | 2018-12-21 | 杭州舜浩科技有限公司 | Single-sided illumination formula light guide plate shadow defect extracting method based on subregion |
| CN109064439B (en)* | 2018-06-15 | 2020-11-24 | 杭州舜浩科技有限公司 | Partition-based extraction method for shadow defect of single-side incident light guide plate |
| CN108872258A (en)* | 2018-06-25 | 2018-11-23 | 成都市金鼓药用包装有限公司 | A kind of empty bottle method of inspection and its empty bottle checking machine |
| CN108709892A (en)* | 2018-07-04 | 2018-10-26 | 杭州智感科技有限公司 | Detecting system and its method |
| CN108802055A (en)* | 2018-08-22 | 2018-11-13 | 苏州西斯派克检测科技有限公司 | A kind of translucent or opaque empty bottle outer wall testing agency |
| CN109615606A (en)* | 2018-11-09 | 2019-04-12 | 华南理工大学 | A rapid classification method for point, line and surface defects of flexible IC substrates |
| CN109615606B (en)* | 2018-11-09 | 2023-01-06 | 华南理工大学 | Rapid classification method for point-line-surface defects of flexible IC substrate |
| CN109632808A (en)* | 2018-12-05 | 2019-04-16 | 深圳大学 | Seamed edge defect inspection method, device, electronic equipment and storage medium |
| CN109632808B (en)* | 2018-12-05 | 2021-11-09 | 深圳大学 | Edge defect detection method and device, electronic equipment and storage medium |
| CN109525840A (en)* | 2018-12-18 | 2019-03-26 | 凌云光技术集团有限责任公司 | The detection method of minor defect on a kind of imager chip |
| CN110517233A (en)* | 2019-08-15 | 2019-11-29 | 浙江赤霄智能检测技术有限公司 | A kind of defect classification learning system and its classification method based on artificial intelligence |
| CN111351754A (en)* | 2020-04-23 | 2020-06-30 | 广州番禺职业技术学院 | Bottle bottom defect detection system and method |
| CN111808367A (en)* | 2020-07-09 | 2020-10-23 | 浙江七色鹿色母粒有限公司 | Improvement method for plastic PPR silver grain whitening defect |
| CN111808367B (en)* | 2020-07-09 | 2023-05-30 | 浙江七色鹿色母粒有限公司 | Improvement method for plastic PPR silver grain whitening defect |
| CN118864437A (en)* | 2024-08-13 | 2024-10-29 | 中山市迪朗食品科技有限公司 | A method, device and medium for visually detecting beverage packaging defects |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101819162A (en) | Empty bottle wall defect detection method and device | |
| CN201689063U (en) | Empty bottle wall defect inspection device | |
| CN105181714B (en) | A kind of bottling liquid Impurity Detection System and its detection method based on image procossing | |
| EP3460765B1 (en) | Banknote management method and system | |
| CN103913468B (en) | Many defects of vision checkout equipment and the method for large-scale LCD glass substrate on production line | |
| CN108896574B (en) | Bottled liquor impurity detection method and system based on machine vision | |
| CN100458422C (en) | Glass Bottle and can detecting method and detecting device | |
| CN101825582B (en) | Method and device for detecting wall of cylindrical transparent bottle | |
| CN105738294B (en) | The spike fruit automatic detection device and method of imaging are regarded based on monocular more | |
| CN109001212A (en) | A kind of stainless steel soup ladle defect inspection method based on machine vision | |
| CN109584239B (en) | High-light object surface defect detection system and method based on reflected light | |
| CN101063662A (en) | Method for detecting empty bottle bottom defect and device for detecting empty bottle bottom defect based on DSP | |
| CN101105459A (en) | Empty bottle mouth defect inspection method and device | |
| CN102507598A (en) | High-speed unordered capsule defect detecting system | |
| CN105865570A (en) | Machine vision-based glass bottle liquid level detection method | |
| CN103792699A (en) | TFT-LCD Mura defect machine vision detecting method based on B spline surface fitting | |
| CN101685006A (en) | Automatic detection system for vision of contact part feature machine | |
| CN111474179A (en) | Lens surface cleanliness detection device and method | |
| CN102175692A (en) | System and method for detecting defects of fabric gray cloth quickly | |
| CN110813795B (en) | Device and method for detecting breakage of chemical fiber spindle paper tube on line | |
| US20060098191A1 (en) | Optical inspection of container walls | |
| CN108267455A (en) | plastic film printing character defect detecting device and method | |
| CN100547394C (en) | Fruit quality detection system based on image information fusion technology | |
| CN111551559A (en) | LCD (liquid Crystal display) liquid crystal screen defect detection method based on multi-view vision system | |
| CN107742119A (en) | A device and method for object contour extraction and matching based on background imaging |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication | Open date:20100901 |