CN108885193A - Failure detector and the defect inspection method for utilizing failure detector - Google Patents
Failure detector and the defect inspection method for utilizing failure detector Download PDFInfo
- Publication number
- CN108885193A CN108885193A CN201780019559.0A CN201780019559A CN108885193A CN 108885193 A CN108885193 A CN 108885193A CN 201780019559 A CN201780019559 A CN 201780019559A CN 108885193 A CN108885193 A CN 108885193A
- Authority
- CN
- China
- Prior art keywords
- defect
- candidate portion
- defect candidate
- region
- failure detector
- 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
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
- G01N21/91—Investigating the presence of flaws or contamination using penetration of dyes, e.g. fluorescent ink
-
- 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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N27/00—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means
- G01N27/72—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating magnetic variables
- G01N27/82—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating magnetic variables for investigating the presence of flaws
- G01N27/83—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating magnetic variables for investigating the presence of flaws by investigating stray magnetic fields
- G01N27/84—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating magnetic variables for investigating the presence of flaws by investigating stray magnetic fields by applying magnetic powder or magnetic ink
Abstract
In the camera for having shooting checked property, in the magnetic powder inspection device of the detection device of the defect from the original image detection checked property obtained by camera, detection device has determining defects portion, the determining defects portion, it is calculated in region (Lc) from the width on long axis direction with predetermined length in the defect candidate portion (43) that original image extracts, defect candidate portion (43) converges on the 2 straight line (s1s parallel with the long axis direction in defect candidate portion (43), s2 between), and 2 straight line (s1, the distance between) s2 (width w) within a predetermined range in the case where, defect candidate portion (43) is determined as defect.
Description
Technical field
This disclosure relates to which failure detector and the defect inspection method using failure detector, more specifically relate to machine
Detect to tool the failure detector of the defect of checked property and the defect inspection method using failure detector.
Background technique
There is the defect detecting test on the surface of ferromagnetic checked property for steel etc., by as non-destructive detection method
A kind of magnetic particle testing, penetrant inspection inspection etc. carry out.Wherein, magnetic powder inspection is also as detection as examined
One of the method the most powerful method of the defect (crackle) of the noxious insult on the surface of object.
If applying magnetic field to steel and magnetizing steel, the disorder of the magnetic flux because of caused by the damage of steel is generated, it should
A part of magnetic flux is revealed as leakage magnetic flux into air.At this point, if there are magnetic powder and/or containing magnetic powder on the surface of steel
Magnetic powder fluid, then magnetic powder is attracted by the leakage magnetic flux and forms the instruction pattern of magnetic powder.Magnetic powder inspection is referred to by observing the magnetic powder
Diagram case checks defect.
On the other hand, in penetrant inspection, by making crack of the penetrating fluid infiltration in the surface opening in checked property, needle
The defects of hole, and it is coated with developer powder on the surface for being removed remaining penetrating fluid, thus because of defect, by showing because of capillary
As and be sucked into surface penetrating fluid formed infiltration instruction pattern.Penetrant inspection is checked by observing infiltration instruction pattern
Defect.
It is automated by these defect detecting tests, thus it can be expected that the raising of productivity, the reduction of expense, the promotion of quality
Deng.Trial of making great efforts is taken into above-mentioned instruction pattern with camera and detects the failure detector of defect by image procossing
Automation.
Patent Document 1 discloses a kind of magnetic powder inspection device, which has：In the table of checked property
Layer portion nearby generates the magnetized portion of rotating excitation field；Yin is carried out with the image that camera takes with to during the generation of rotating excitation field
The damage check portion of shadow correction and the pars affecta on all directions of detection.
Existing technical literature
Patent document
Patent document 1:Japanese Unexamined Patent Publication 2011-038796 bulletin
Summary of the invention
Problems to be solved by the invention
According to the composition of patent document 1, the magnetic powder instruction pattern in rotating excitation field can be observed, previous over sight (OS) is connected
The damage of depth as shallow also can be detected steadily, therefore the detection accuracy of pars affecta improves, and can play the life for improving checked property
The effect of yield.However, magnetic powder instruction pattern is also formed in checked property and non-defective magnetic discontinuous portion and/or surface shape
Therefore the change section of shape will differentiate that these positions and defect are not easy.
Therefore, the failure detector for being designed to provide the recall rate for improving defect of the disclosure and utilization flaw detection dress
The defect inspection method set.
Technical teaching for solving the problem was
To solve the above-mentioned problems, the disclosure is a kind of failure detector, is had：The input dress that checked property is shot
It sets；With the detection device for the defect for detecting the checked property from the original image obtained by the input unit, the flaw detection dress
The feature set is here, the detection device has determining defects portion, the determining defects portion, what is extracted from the original image
The width on long axis direction with predetermined length in defect candidate portion calculates in region, defect candidate portion restrains
Between 2 straight lines parallel with the long axis direction in defect candidate portion and the distance between 2 straight lines are that width exists
In the case where in predetermined range, defect candidate portion is determined as the defect.
And, which is characterized in that the width calculates the center of gravity that region includes defect candidate portion.
It is characterized in that, the detection device also has：Binaryzation is carried out to the original image with first threshold to extract
First extraction unit in the first defect candidate portion；The test zone of inspection area is generated in a manner of comprising the first defect candidate portion
Domain generating unit；And binaryzation is carried out to the inspection area to extract second and lack with the second threshold smaller than the first threshold
Second extraction unit in candidate portion is fallen into, defect candidate portion is region obtained by the inflated processing in the second defect candidate portion.
And, which is characterized in that the inspection area is by from the smallest rectangular containing the first defect candidate portion
The region that shape is surrounded by the rectangle that predetermined amplitude expands.
And, which is characterized in that the expansion process is to make the second defect candidate portion to respective at least long axis side
To the processing of expansion.
It is characterized in that, the failure detector is to be also equipped with：The magnetic powder dissemination apparatus of magnetic powder is applicable in the checked property；
The magnetic powder inspection that the magnetic powder generated because reveal magnetic flux indicates the magnetizing assembly of pattern is formed with the checked property is magnetized
Device.
And, which is characterized in that the length that the width calculates region is 0.3mm or more and 10mm or less.
Moreover, the disclosure is a kind of defect inspection method, and it is the defect inspection method using following failure detectors, it is described
In failure detector, input unit shoots checked property, and detection device is from described in the original image detection obtained as the input unit
The defect of checked property, the defect inspection method be characterized in that, determining defects portion possessed by the detection device, from
In the width calculating region on long axis direction with predetermined length in the defect candidate portion that the original image extracts,
Defect candidate portion converges between 2 straight lines parallel with the long axis direction in defect candidate portion and 2 straight lines
The distance between i.e. width within a predetermined range in the case where, defect candidate portion is determined as the defect.
And, which is characterized in that the width calculates the center of gravity that region includes defect candidate portion.
It is characterized in that, the first extraction unit that the detection device also has, carries out the original image with first threshold
Binaryzation extracts the first defect candidate portion, and inspection area generating unit generates inspection in a manner of comprising the first defect candidate portion
Look into region, the second extraction unit with the second threshold smaller than the first threshold carries out binaryzation to the inspection area to extract the
Two defect candidate portions, defect candidate portion are regions obtained by the inflated processing in the second defect candidate portion.
And, which is characterized in that the inspection area is by from the smallest rectangular containing the first defect candidate portion
The region that shape is surrounded by the rectangle that predetermined amplitude expands.
And, which is characterized in that the expansion process is to make the second defect candidate portion to respective at least long axis side
To the processing of expansion.
It is characterized in that, the failure detector is to be also equipped with：The magnetic powder dissemination apparatus of magnetic powder is applicable in the checked property；
With the magnetizing assembly for making the checked property magnetization and the magnetic powder instruction pattern that formation is generated because revealing magnetic flux.
And, which is characterized in that the length that the width calculates region is 0.3mm or more and 10mm or less.
The effect of invention
According to the disclosure, a kind of failure detector has the input unit shot to checked property and fills from by input
The detection device of the defect of the original image detection checked property of acquirement is set, detection device has determining defects portion, the determining defects
Portion, the width on long axis direction with predetermined length in the defect candidate portion extracted from original image calculate region
In, defect candidate portion converge between 2 straight lines parallel with the long axis direction in defect candidate portion and between 2 straight lines away from
From i.e. width within a predetermined range in the case where, defect candidate portion is determined as defect, therefore, the defect of checked property
Shape quantitatively extracted as characteristic quantity, can with high accuracy come identify defect and it is non-defective with reduce excessively inspection
It surveys.Therefore, it is capable of providing the failure detector that the recall rate of defect improves.
Moreover, width calculates the center of gravity that region includes defect candidate portion, so that the shape of the defect of checked property is more reliable
Ground is quantitatively extracted as characteristic quantity, can with higher accuracy come identify defect and it is non-defective be further reduced leakage
Detection and excessively detection.Therefore, it is capable of providing the failure detector that the recall rate of defect further improves.
Detection device also has：Binaryzation is carried out to original image with first threshold to extract the first of the first defect candidate portion
Extraction unit；The inspection area generating unit of inspection area is generated in a manner of comprising the first defect candidate portion；And than the first threshold
It is worth small second threshold and binaryzation is carried out to inspection area to extract second extraction unit in the second defect candidate portion, defect candidate portion
Region obtained by the inflated processing in the second defect candidate portion, as a result, in the leak detection for the defect for preventing checked property and
Excessively on the basis of detection, the shape of defect is more accurately used as characteristic quantity and is quantitatively extracted, can be with higher accurate
Degree identifies defect and non-defective to be further reduced excessive detection.Therefore, the recall rate for being capable of providing defect further increases
Failure detector.
Moreover, inspection area is by expanding from the smallest rectangle containing the first defect candidate portion by predetermined amplitude
The range check region that big rectangle is surrounded, therefore, more accurately prevent the leak detection of the defect of checked property with
And excessively on the basis of detection, the shape of defect is more accurately used as characteristic quantity and is quantitatively extracted, can be with higher standard
Exactness identifies defect and non-defective to be further reduced excessive detection.Therefore, the recall rate for being capable of providing defect further mentions
High failure detector.
Moreover, expansion process is the processing for expanding the second defect candidate portion to respective at least long axis direction, therefore,
It more accurately prevents the leak detection of the defect of checked property and excessively on the basis of detection, the shape of defect is more accurately made
The amount of being characterized and quantitatively extracted, can identify defect and non-defective to be further reduced excessive inspection with higher accuracy
It surveys.Therefore, it is capable of providing the failure detector that the recall rate of defect further improves.
Failure detector is to be also equipped with：The magnetic powder dissemination apparatus of magnetic powder is applicable in checked property；With make checked property magnetize and
Therefore the magnetic powder inspection device for forming the magnetizing assembly of the magnetic powder instruction pattern generated because revealing magnetic flux has ferromagnetic
The shape of the defect of checked property is more reliably used as characteristic quantity and quantitatively extracted, and can be identified with higher accuracy
Defect and non-defective to be further reduced excessive detection.Therefore, it is capable of providing the flaw detection that the recall rate of defect further improves
Device.
Moreover, it is 0.3mm or more and 10mm hereinafter, therefore that width, which calculates the length in region, the shape of the defect of checked property
Shape is more reliably used as characteristic quantity and is quantitatively extracted, and can identify defect and non-defective with into one with higher accuracy
Step reduces excessively detection.Therefore, it is capable of providing the failure detector that the recall rate of defect further improves.
Moreover, the disclosure is the defect inspection method using following failure detectors, in the failure detector, input unit is clapped
Checked property is taken the photograph, detection device detects the defect of checked property from the original image obtained by input unit, in the defects detection side
In method, determining defects portion possessed by detection device, which exists in the defect candidate portion that extracts from original image
Width on long axis direction with predetermined length calculates in region, defect candidate portion converges on the length with defect candidate portion
Between 2 parallel straight lines of axis direction and the distance between 2 straight lines be width within a predetermined range in the case where,
Defect candidate portion is determined as defect, therefore, the shape of the defect of checked property is quantitatively extracted as characteristic quantity, can
Defect is identified and non-defective to be further reduced excessive detection with high accuracy.Therefore, it is capable of providing the inspection using defect
The defect inspection method for the failure detector that extracting rate improves.
Moreover, width calculates the center of gravity that region includes defect candidate portion, therefore, the shape of the defect of checked property is more reliable
Ground is quantitatively extracted as characteristic quantity, can with higher accuracy come identify defect and it is non-defective be further reduced leakage
Detection and excessively detection.Therefore, it is capable of providing the defects detection of the failure detector further improved using the recall rate of defect
Method.
The first extraction unit that detection device also has carries out binaryzation to original image with first threshold to extract the first defect
Candidate portion, inspection area generating unit generate inspection area in a manner of comprising the first defect candidate portion, and the second extraction unit is than
The small second threshold of one threshold value carries out binaryzation to inspection area to extract the second defect candidate portion, and defect candidate portion is second scarce
Region obtained by falling into that candidate portion is inflated and handling, therefore, leak detection and the excessive base that detects in the defect for preventing checked property
On plinth, the shape of defect is more reliably used as characteristic quantity and is quantitatively extracted, and can identify defect with higher accuracy
With non-defective to be further reduced excessive detection.Therefore, it is capable of providing the flaw detection further improved using the recall rate of defect
The defect inspection method of device.
Moreover, inspection area is by expanding from the smallest rectangle containing the first defect candidate portion by predetermined amplitude
Region that big rectangle is surrounded, therefore, the defect for being more reliably prevented from checked property leak detection and excessively detect
On the basis of, the shape of defect is more reliably used as characteristic quantity and is quantitatively extracted, can be scarce to identify with higher accuracy
It falls into and non-defective to be further reduced excessive detection.Therefore, it is capable of providing the spy further improved using the recall rate of defect
Hurt the defect inspection method of device.
Moreover, expansion process is the processing for expanding the second defect candidate portion to respective at least long axis direction, therefore,
It is more reliably prevented from the leak detection of the defect of checked property and excessively on the basis of detection, the shape of defect is more reliably used as spy
Sign amount and quantitatively extracted, can identify defect and non-defective to be further reduced excessive detection with higher accuracy.
Therefore, it is capable of providing the defect inspection method of the failure detector further improved using the recall rate of defect.
Failure detector is to be also equipped with：The magnetic powder dissemination apparatus of magnetic powder is applicable in checked property；With make checked property magnetize and
Therefore the magnetic powder inspection device for forming the magnetizing assembly of the magnetic powder instruction pattern generated because revealing magnetic flux has ferromagnetic
The shape of the defect of checked property is more reliably used as characteristic quantity and quantitatively extracted, and can be identified with higher accuracy
Defect and non-defective to be further reduced excessive detection.Therefore, it is capable of providing and is further improved using the recall rate of defect
The defect inspection method of failure detector.
Moreover, it is 0.3mm or more and 10mm hereinafter, therefore that width, which calculates the degree in region, the defect shape of detected material
It is more reliably quantitatively extracted as characteristic quantity, can identify defect and non-defective with further with higher accuracy
Reduce leak detection and excessively detection.Therefore, it is capable of providing lacking for the failure detector further improved using the recall rate of defect
Fall into detection method.
Detailed description of the invention
Fig. 1 is to show the schematic diagram of the magnetic powder inspection device of an example as failure detector of the present embodiment.
Fig. 2 is to show the block diagram of an example of control system of magnetic powder inspection device.
Fig. 3 is the flow chart for an example for illustrating the detection work of detection device.
Fig. 4 is to show the skeleton diagram of an example that the image after binaryzation has been carried out by the first extraction unit.
Fig. 5 is the general of the image of an example for showing the first defect candidate portion extracted and the inspection area generated
Sketch map.
Fig. 6 is to show the skeleton diagram of an example that the image after binaryzation has been carried out by the second extraction unit.
Fig. 7 is to show the skeleton diagram of the image of an example in the second defect candidate portion extracted.
Fig. 8 is the skeleton diagram of the image of inflated treated an example in the second defect candidate portion that shows.
Fig. 9 is the skeleton diagram for an example for the method for illustrating to extract the characteristic quantity in defect candidate portion.
Figure 10 is to show the skeleton diagram of the image of an example of the defect extracted.
Specific embodiment
Hereinafter, being illustrated referring to details of the attached drawing to embodiment of the present disclosure.Firstly, being related to present embodiment
And failure detector be described in detail.Fig. 1 is to show the magnetic of an example as failure detector of the present embodiment
The schematic diagram of powder failure detector 1.It should be noted that the checked property 10 of the check object as magnetic powder inspection device 1 is defeated
Send direction as shown with arrows in Fig. 1 from right to left.
Magnetic powder inspection device 1 be use magnetic powder as mark, mechanically to the surface with ferromagnetic checked property 10
And then the device that the defect at the lower section on surface is detected automatically.The magnetic powder inspection device 1 illustrated in Fig. 1 is used for size
Bigger check object, checked property 10 are the steel of the corner post shape of long size.Also, magnetic powder inspection device 1 has roll-type
Conveying device 11, magnetic powder dissemination apparatus 12, magnetizing assembly 13, blower device 14, ultraviolet crack inspection lamp 15, camera 16, Yi Jibiao
Remember device 17.Moreover, magnetic powder inspection device 1 also has controller (not shown), detection device etc. herein.
Roll-type transporting device 11 as conveying device conveys checked property 10.Roll-type transporting device 11 has multiple rollers,
And it is configured to convey checked property 10 with desired speed.Also, roll-type transporting device 11 is set to spread from magnetic powder and fill
The position for setting 12 reaches labelling apparatus 17 by the setting position of magnetizing assembly 13, ultraviolet crack inspection lamp 15, camera 16 etc.
The transport path of checked property 10 until position.It should be noted that conveying device is as long as it can convey checked property 10
It is not limited to roll-type transporting device 11, such as the conveyer belt being made of cricoid band etc. can also be made.
Roll-type transporting device 11 is provided with conveying distance measuring device (not shown) herein.Conveying distance measuring device meter
Survey the conveying distance of checked property 10.As conveying distance measuring device, such as it is able to use by pair roller type conveying device 11
The device that the rotary encoder that the rotation displacement of roller is measured is constituted as sensor.It should be noted that conveying away from
From measuring device, it is not limited to rotary encoder, non-contacting measurement device such as laser surface speedometer also can be used, it can also
These to be applied in combination.
Magnetic powder dissemination apparatus 12 is applicable in the magnetic powder of (distribution) as mark to the surface of the check object of checked property 10, matches
It is placed in the upstream side on the conveying direction of checked property 10.As magnetic powder, such as it can be used the surface of magnetic powder (iron powder) with glimmering
The fluorescentmagnetic particle(powder) that body of light covers, and use the Magnetic testing liquid of the water and/or oil that are dispersed with the fluorescentmagnetic particle(powder).Magnetic powder spreads dress
It sets 12 to have such as tank (not shown), pump, nozzle, and is configured to carry out pressure to the magnetic particle inspection liquid for being accommodated in tank by pumping
It conveys and is sprayed from nozzle.Magnetic powder dissemination apparatus 12 is configured to, can continuously be spread to the surface of checked property 10 desired by
Amount magnetic particle inspection liquid.
Magnetizing assembly 13 applies magnetic field to checked property 10 and makes its magnetization, and is adjacent to and is configured at magnetic powder dissemination apparatus 12
Downstream side.Magnetizing assembly 13 has：It is relatively configured 2 in upstream side and downstream side on the conveying direction of checked property 10
A perforation coil 19,20；And 2 interpolar coil (days that arow configures in the conveying direction between the perforation coil 19,20
Text：Very Inter U イ Le) 21,22.Perforation coil 19,20 is formed as circular, is configured in the mode penetrated through in the annulus tested
Look into the transport path of object 10.On the other hand, interpolar coil 21,22 is formed as U-shaped, be configured in a manner of through the gap by
Check the transport path of object 10.Also, magnetizing assembly 13 is configured to, by having 2 interpolar coils 21,22 at 2
The same rotating excitation field is generated between perforation coil 19,20.
More specifically, magnetic field is generated on the conveying direction of checked property 10 in perforation coil 19,20, in interpolar line
In circle 21,22, magnetic field is generated on the direction orthogonal with the conveying direction of checked property 10 as its gap direction.Also,
If 90 degree of phase deviation of alternating current is flowed in perforation coil 19,20 and interpolar coil 21,22, by conveying direction
The magnetic field in magnetic field and the direction orthogonal with conveying direction is formed by the rotation for generating in plane and rotating with certain magnetic field strength
Magnetic field.Due to the rotating excitation field, in the surface section of checked property 10, leakage magnetic flux is independently generated with the direction of damage, is formed
Magnetic powder indicates pattern.
It should be noted that constituting the perforation coil 19 of magnetizing assembly 13,20, quantity of interpolar coil 21,22 etc. can fit
Locality design.For example, magnetizing assembly 13 can be the composition also other than interpolar coil 21,22 with multiple interpolar coils,
On the other hand, it is also possible to penetrate through the composition that coil 19 and 1 interpolar coil 21 is constituted by 1.
Blower device 14, which has, to be blown up air in the side opposite with gravity towards checked property 10 and adjusts tested
Look into the function of the flow velocity of the Magnetic testing liquid of the surface flow of object 10.Also, by the blower device 14, the magnetic powder of formation is indicated
Pattern is apparent from.Blower device 14 be respectively arranged at perforation coil 19 and interpolar coil 21 between, 2 interpolar coils 21,22
Between and interpolar coil 22 and perforation coil 20 between.
The magnetic that ultraviolet crack inspection lamp 15 as light source penetrates through between coil 19,20 to the surface of checked property 10 at 2
Powder checks that liquid irradiates ultraviolet light.It, can also be in ultraviolet crack inspection in order to keep the uitraviolet intensity of the coverage of camera 16 uniform
Parabolic reflecting plate is arranged in lamp 15.It is purple in order to avoid being influenced by the strong rotating excitation field generated by magnetizing assembly 13
Outside line flaw detection lamp 15 is preferably arranged to leave the degree of suitable distance, such as 600mm~2000mm from checked property 10.It needs
Illustrate, magnetic screen can also be carried out to ultraviolet crack inspection lamp 15.
It should be noted that using the surface of such as magnetic powder (iron powder) with white, black, red such coloring face as magnetic powder
In the case where expecting the visible magnetic particle covered, light source also can be set to visible light.However, if being irradiated to fluorescentmagnetic particle(powder) ultraviolet
Line, then magnetic powder indicates pattern in visible light region bright flash (luminous), it is difficult to the dirt on the surface by checked property 10, rule
The influence of mould etc. can detect damage with high accuracy.Therefore, ultraviolet crack inspection lamp 15 and fluorescent magnetic are more preferably used
Powder.
As input unit camera 16 to by ultraviolet crack inspection lamp 15 irradiate ultraviolet light checked property 10 surface into
Row shooting.Camera 16 is in order to detect defect with higher accuracy, it is preferred, therefore, that be set as relative to checked property 10
Surface from the configuration shoot of vertical direction.Moreover, camera 16 is in order to avoid the strong rotation that is generated by magnetizing assembly 13
The influence in magnetic field, it is preferred that it is configured to leave the degree of suitable distance, such as 600mm~2000mm from checked property 10,
And by magnetic screen.
As camera 16, line camera can be used, face formula camera also can be used.Moreover, about camera 16 is equipped on
Element, CCD (Charge Coupled Device, charge-coupled device) image sensor can be used, also can be used
CMOS (Complementary Metal-Oxide Semiconductor, complementary metal oxide semiconductor) image sensor.
Labelling apparatus 17 to the defect on the surface of the checked property 10 detected by detection device carry out can be visual label.
As labelling apparatus 17, such as can be with using with air pressure jet ink the label rifle that is marked.It should be noted that
The check object smaller for size magnetic powder inspection device 1, for example can be by defective joint product together to not conforming to
In the case where system as the discharge of lattice tablet tray, labelling apparatus 17 also can be omitted.
Then, the control system of magnetic powder inspection device 1 of the present embodiment is described in detail.Fig. 2 is to show
The block diagram of an example of the control system of magnetic powder inspection device 1 is gone out.
Magnetic powder inspection device 1 has controller C.Roll-type transporting device 11, conveying distance measurement are electrically connected in controller C
Device 18, magnetic powder dissemination apparatus 12, magnetizing assembly 13, blower device 14, ultraviolet crack inspection lamp 15, camera 16, labelling apparatus 17,
And detection device 30.It should be noted that various other than controller C is also electrically connected to constitute illustrated by Fig. 2
Sensor class.Magnetic powder inspection device 1 is configured to, and controller C controls above-mentioned various devices to automatically detect checked property
The defect on 10 surface.
Controller C is configured to, simultaneously by input signals such as the detected values that reads in various setting values, various sensors
And output control signal, to control the work for the various devices that magnetic powder inspection device 1 has.Controller C is by carry out operation
The processing unit and the main storage means for saving data etc. of processing and control processing are constituted.Controller C is, for example, to have as place
Manage CPU (Central Processing Unit, central processing unit), ROM (the Read Only as main storage means of device
Memory, read-only memory), RAM (Random Access Memory, random-access memory), timer, input circuit,
The microcomputer of output circuit and power circuit etc..It preserves in main storage means and is related to for executing present embodiment
And the control program of work, various data etc..It should be noted that these various procedure, datas etc. can also be stored in
The storage device of controller C split settings, and the form that can be read for controller C.
Detection device 30 is read in the picture signal (original image) obtained by camera 16 and is carried out to original image defined
Reason, to detect the defect on the surface of checked property 10.It should be noted that controller C is configured to, progress is obtained by camera 16
Original image the input to detection device 30.Detection device 30 in the same manner as controller C, by carry out calculation process and control
The processing unit of reason and the main storage means for saving data etc. are constituted, and for example, have CPU, main storage means, timer, input
The microcomputer of circuit, output circuit, power circuit etc..
Detection device 30 at least have determine the defect candidate portion that is extracted from original image whether be defect determining defects
Portion.The shape of the defect of checked property 10 is quantitatively extracted as characteristic quantity as a result, can be scarce to identify with high accuracy
It falls into and non-defective to reduce excessively detection.Therefore, it is capable of providing the failure detector that the recall rate of defect improves.
Detection device 30 illustrated by Fig. 2 have the first extraction unit 31, inspection area generating unit 32, the second extraction unit 33,
And determining defects portion 34.These each sections are for example made of program.It should be noted that being also configured to, each section etc.
It is realized with hardware.
First extraction unit 31 is configured to, and inputs the original image that is obtained by camera 16, and with preset first threshold into
Row binaryzation extracts the first defect candidate portion.
Inspection area generating unit 32 is configured to, with the side containing the first defect candidate portion extracted by the first extraction unit 31
Formula generates inspection area.
Second extraction unit 33 is configured to, and inputs the inspection area generated by inspection area generating unit 32, and to preset
Second threshold carry out binaryzation and extract the second defect candidate portion.
Determining defects portion 34 is configured to, and inputs the second defect candidate portion being extracted by the second extraction unit 33, and calculate swollen
The width in swollen treated defect candidate portion detects defect.Also, detection device 30 is configured to, and will be examined by determining defects portion 34
The position data of defect out is exported to controller C.
Then, the work of magnetic powder inspection device 1 is described in detail.Magnetic powder inspection device 1 is configured to, checked property
10 from roll-type transporting device 11 successively to conveying under each device, and detect the defect on the surface of checked property 10.
Firstly, being applicable in magnetic particle inspection liquid by surface of the magnetic powder dissemination apparatus 12 to checked property 10.Surface has been applicable in magnetic
Powder checks that the checked property 10 of liquid is conveyed into the rotating excitation field region formed by magnetizing assembly 13.At this point, in checked property
10 surface is deposited in the case of a fracture, the leakage magnetic field from generating due to the damage, and magnetic powder contained by magnetic particle inspection liquid is by this
Reveal magnetic field suction.At this point, being rotated by the magnetic field that magnetizing assembly 13 is formed, therefore the extension in injury region, with the damage
Direction and/or shape independently generate leakage magnetic field, which damage also all attracts magnetic powder.Also, magnetic powder collection is together in damage, therefore
Magnetic powder instruction pattern corresponding with damage is formed in the surface of checked property 10.
It should be noted that the flow velocity in the magnetic particle inspection liquid of the surface flow of checked property 10 is suitable by blower device 14
Ground is adjusted.If the flow velocity of magnetic particle inspection liquid is too low, many times are needed until forming magnetic powder instruction pattern and stablizing, are checked
Effect decline.On the other hand, if the flow velocity of magnetic particle inspection liquid is excessively high, magnetic powder is become difficult to by the leakage magnetic field from due to damage
Attract, magnetic powder instruction pattern becomes unintelligible.Therefore, from the viewpoint of being reliably formed magnetic powder instruction pattern, magnetic powder inspection
The flow velocity for looking into liquid is preferably 5mm/s or more and 100mm/s or less.
The magnetic powder instruction pattern formed in this way, the fluorophor on the surface as covering magnetic powder are absorbed by ultraviolet crack inspection lamp
The energy of the ultraviolet lights of 15 irradiations and the visible light of high luminance for exciting, being discharged when the fluorophor returns to ground state, and by
Camera 16 is shot.Detection device 30 is input by the original image that the camera 16 obtains.Detection device 30 carries out the original image
Defined processing exports the position data of the defect to controller C to detect the defects of original image.Controller C is based on
The position data of the defect and the measurement data of conveying distance measuring device 18 control the work of labelling apparatus 17.Also, it is right
The defect on the surface of checked property 10 is marked by labelling apparatus 17.
Then, the detection method about the defect using detection device 30 is described in detail.Utilizing this embodiment party
In the defect inspection method of the failure detector of formula, firstly, determining defects portion 34 possessed by detection device 30 is cut out from original image
The width on long axis direction with predetermined length in the defect candidate portion being extracted into calculates region.Also, defect is sentenced
Determine portion 34, width calculate region in defect candidate portion converge on 2 straight lines parallel with the long axis direction in defect candidate portion it
Between and the distance between 2 straight lines (width) within a predetermined range in the case where, defect candidate portion is judged to lacking
It falls into.In this way, using present embodiment failure detector defect inspection method in, can by be entered defect candidate portion from
And quantitatively extracted using the shape of the defect of checked property 10 as characteristic quantity, and with high accuracy come identify defect and it is non-lack
It falls into reduce excessively detection.Therefore, it is capable of providing the defect inspection method of the failure detector improved using the recall rate of defect.
It also can include the step for extracting defect candidate portion in the defect inspection method using the failure detector of present embodiment
Suddenly, it is detected with the leak detection for being more reliably prevented from the defect of checked property 10 with excessive.Then, failure detector is utilized to such
Defect inspection method in detail to being illustrated.Fig. 3 is the flow chart for an example for illustrating the detection work of detection device 30.
Firstly, the progress image of detection device 30 is taken into (step S1).Detection device 30 is carried out via controller C by camera
16 picture signal (original images) obtained are taken into.Original image is made of horizontal × vertical such as 1280 × 960 pixels.Also,
The information of shooting time is saved in original image.
In the defect inspection method using the failure detector of present embodiment, substantially carry out lease making by 3 steps from original image
As extracting defect.1st process (i) is completed by the first extraction unit 31.
Firstly, the first extraction unit 31 is emphasized processing (step S2).First extraction unit 31 uses respectively as pre-treatment
Kind filter is emphasized processing to original image.As long as emphasizing that the processing of the damage of original image is emphasized in processing, e.g.
LUT (Look Up Table, inquiry table) conversion process, amplification shrink process, masking processing etc., are also possible to combine these
It is handled obtained by various processing.It emphasizes to handle it should be noted that, although also can be omitted, but in order to improve the detection of defect
Rate is preferably emphasized processing.
Then, the first extraction unit 31 carries out binaryzation (step S3) with first threshold.First extraction unit 31 uses first threshold
To image of emphasizing that treated (be omitted emphasize processing in the case where for original image) carry out binaryzation.The first threshold is to make
For in order to detect defect and the preset threshold value of accurate value adjusted with high accuracy, and it is stored in detection device
30 main storage means (not shown).Detection device 30 is also configured to that first threshold can be changed.
Fig. 4 is to show the skeleton diagram of an example that the image after binaryzation has been carried out by the first extraction unit 31.It needs to illustrate
, Fig. 4 extracted explanation needed for part pixel.In Fig. 4, magnetic powder instruction pattern is shown with black.To this with black
Each magnetic powder instruction pattern that color is shown carries out labeling.Here, so-calledization label is in binary image to connection
Pixel is by the identical label (number) of 1 closed area marking of contour line and to carry out the processing of region division.
Also, the first extraction unit 31 extracts the region (step S4) being consistent with the first defect candidate portion.So far is the 1st
Process.First extraction unit 31 presses the region of each label, calculated as decision content such as area, length, longitudinal direction width
Degree, lateral width.Also, the decision content that the first extraction unit 31 calculates these and the first determinating reference value for pre-saving into
Row relatively determines whether the first defect candidate portion, and extracts the region being consistent with the first defect candidate portion.
Here, so-called first defect candidate portion is the region for being assumed to defect.Also, the first extraction unit 31 is to label
All areas be carried out similarly whether be the first defect candidate portion judgement.
Fig. 5 is to show the image of an example in the first defect candidate portion 40 extracted and the inspection area 41 generated
Skeleton diagram.It should be noted that Fig. 5 is to have extracted the first defect candidate portion 40 using the length in region as decision content
State, for the state for having extracted 8 first defect candidate portions 40a, 40b, 40c, 40d, 40e, 40f, 40g and 40h.
It should be noted that as long as the value that decision content can be compared the feature in region can, can be suitably set.
And, if it is the judgement in the first defect candidate portion 40, can be carried out based on a kind of decision content, a variety of judgements can also be based on
Value carries out.Moreover, in the case where determining whether the first defect candidate portion 40 based on a variety of decision contents, as long as at these
It completes to determine based at least one kind of decision content in a variety of decision contents.For example, it is also possible to using at least two kinds of in 3 kinds of decision contents
Decision content meet benchmark in the case where, by the regional determination be the first defect candidate portion 40 method.
Moreover, it can be, first extraction unit 31 be not repeated step S4 by each region, but formerly calculated institute
After having the decision content in region, the decision content calculated by each region compared with the first determinating reference value and whether
For the judgement in the first defect candidate portion 40.
Then, advance to the 2nd process (ii).2nd process is mentioned by the inspection area generating unit 32 of detection device 30 and second
Portion 33 is taken to complete.
The generation (step S5) of the progress of inspection area generating unit 32 inspection area 41.Inspection area generating unit 32 to contain respectively
There is the mode for each of extracting the first defect candidate portion 40 by the first extraction unit 31 to generate inspection area 41.In this way, test zone
Domain 41 is collapsed near the first defect candidate portion 40, and thus preventing will simultaneously in low region the existing defects a possibility that
Damage, dust of non-harmful etc. detect the excessive detection at defect.Moreover, the operand of detection device 30 can be reduced.
It is shown in FIG. 5 and 8 first defect candidate portion 40a, 40b, 40c, 40d, 40e, 40f, 40g and 40h phases
Inspection area 41a, 41b, 41c, 41d, 41e, 41f, 41g and the 41h accordingly generated respectively.The test zone illustrated in Fig. 5
Domain 41 is by the region that the longitudinal and horizontal rectangle upwardly extended is surrounded.Also, inspection area 41 be by from its area at
The region that rectangle after expanding respectively by predetermined amplitude for the smallest rectangle to longitudinal direction and laterally is surrounded.
It should be noted that the enlarging amplitude of the longitudinal direction and transverse direction may be the same or different.Here, test zone
As long as region of the domain 41 containing the first defect candidate portion 40, such as be also possible to be wrapped by diamond shape, trapezoidal, round, ellipse etc.
The region enclosed.Moreover, inspection area 41 can be inclined region, such as it is also possible to the long axis in the first defect candidate portion 40
The rectangle etc. just upwardly extended.
Then, the second extraction unit 33 is emphasized processing (step S6).Inspection area of second extraction unit 33 to original image
41 are carried out similarly with the step S2 carried out by the first extraction unit 31 and emphasize to handle.It should be noted that being also possible to detection dress
Setting 30 in step s 2 will emphasize treated image is stored in main storage portion, and the second extraction unit 33 will be by that will save
The inspection area 41 of image cut out to omitting step S6.By using such method, so as to reduce detection device
30 operand.
Moreover, the second extraction unit 33 carries out binaryzation (step S7) with second threshold.Second extraction unit 33 is used as than
The small second threshold of one threshold value to the image that is emphasised that treated (be omitted emphasize processing in the case where for original image) carry out
Binaryzation.The second threshold is as in order to detect defect and suitable value adjusted institute is preset with high accuracy
Value, and it is stored in the main storage means (not shown) of detection device 30.Detection device 30 is also configured to that can be changed
Two threshold values.
Fig. 6 is to show the skeleton diagram of an example that the image after binaryzation has been carried out by the second extraction unit 33.Here, reference
Fig. 6 extracts many smaller regions compared with Fig. 4.Also, for example, the first defect candidate portion 40c in Fig. 5 and with
The corresponding region one defect candidate portion 40d is concatenated and is formed as 1 region.This is because：Second threshold compares first threshold
It is small, the width of damage is narrow or shallow etc. and indicate not become apparent from for pattern as magnetic powder, the smaller region of luminance is also detected
Out.
About damage, even a continuous damage, also to there is luminance at random according to its depth and/or width big
Region and the small region of luminance, be taken with this state.However, by with the second threshold smaller than first threshold into
For row binaryzation to can also detect the smaller region of luminance, enabling to will not be at shallow position etc. with the comparable region of defect
Place midway disconnects.Also, the second extraction unit 33 refers to each magnetic powder of the image after binaryzation in the same manner as the first extraction unit 31
Diagram case carries out labeling.
Then, the second extraction unit 33 extracts the region (step S8) being consistent with the second defect candidate portion.It is the 2nd so far
Process.Second extraction unit 33 by each region for having label, calculated as decision content for example area, length, longitudinal width,
Transverse width.Also, the decision content that the second extraction unit 33 calculates these compares with the second determinating reference value pre-saved
It relatively determines whether the second defect candidate portion, and extracts the region being consistent with the second defect candidate portion.
Here, so-called second defect candidate portion, is the region for being assumed defect in the same manner as the first defect candidate portion.And
And second extraction unit 33 for label all areas be carried out similarly whether be the second defect candidate portion judgement.
The second determinating reference value is difficult to compared with being the first determinating reference value used in the first extraction unit 31 by region
The value extracted as defect.It is so-called be difficult to be extracted mean for example if not bigger region and/or longer region then
It will not be extracted as defect.That is, the range of the second determinating reference value is contained within the range of the first determinating reference value and opposite
Narrow range.
Fig. 7 is to show the skeleton diagram of the image of an example in the second defect candidate portion 42 being extracted.It needs to illustrate
It is that Fig. 7 is that the length in region has extracted the state in the second defect candidate portion 42 as decision content, to have extracted 6 the
The state of two defect candidate portion 42a, 42b, 42c, 42e, 42f and 42g.Also, do not have to propose and the in Fig. 5 in Fig. 7
The corresponding region one defect candidate portion 40h.This is because：Second determinating reference value is to be more difficult than the first determinating reference value by area
The value that domain is extracted as defect, therefore, it is determined that not being to be assumed to lack for region corresponding with the first defect candidate portion 40h
Sunken region.
It should be noted that as long as the feature in region can be compared, can suitably set by decision content.And
And, if it is the judgement in the second defect candidate portion 42, can be carried out based on a kind of decision content, a variety of decision contents can also be based on
Come carry out.Moreover, in the case where determining whether the second defect candidate portion 42 based on a variety of decision contents, a variety of sentence above-mentioned
It completes to determine based at least one kind of decision content in definite value.For example, it is also possible to use at least two kinds of decision contents in 3 kinds of decision contents
Meet in the case where benchmark be by the regional determination the second defect candidate portion 42 method.
It should be noted that the second determinating reference value is difficult to if not than the first determinating reference value using region as defect
It can also be with come the value extracted.For example, the second determinating reference value can also be identical as the first determinating reference value, in such situation
Under, the second extraction unit 33 is configured to, and is made whether it is that the second defect is waited based on a variety of decision contents than the first extraction unit more than 31
Select the judgement in portion 42.
Moreover, the second extraction unit 33 is also possible to step not be repeated by each region in the same manner as the first extraction unit 31
Rapid S8, but after the decision content for calculating all areas in advance, the decision content calculated by each region and second determines
The comparison of a reference value and whether be the second defect candidate portion 42 judgement.
Then, advance to the 3rd process (iii).3rd process is completed by the determining defects portion 34 of detection device 30.
Determining defects portion 34 carries out expansion process (step S9) respectively to the second defect candidate portion 42.Here, so-called expansion
Processing is the processing for making the region in the second defect candidate portion 42 amplify (expansion).Determining defects portion 34 is for the second defect candidate portion
Each of 42 are allowed to amplify on it is longitudinal and lateral by predetermined quantities.That is, the expansion process in determining defects portion 34 is
The processing for amplifying the second defect candidate portion 42 to its peripheral direction.At this point, as long as expansion process is for the second defect candidate portion
42 it is each be allowed at least on long axis direction amplify.By the expansion process, 1 continuous damage midway can be prevented disconnected
It opens to disconnect again and is but judged as non-defective situation, the leak detection of defect can be prevented.So, the second defect candidate portion 42
It is inflated that treated that region becomes defect candidate portion.
Fig. 8 is the skeleton diagram of the image of the inflated an example handled in the second defect candidate portion 42 that shows.If referring to figure
8, then 3 for example in Fig. 7 the second defect candidate portion 42a, 42b and 42c is formed 1 and is lacked by inflated processing connection
Fall into candidate portion 43a.Also, determining defects portion 34, in the same manner as the first extraction unit 31 and the second extraction unit 33, to the second defect
Candidate portion 42 passes through each defect candidate portion 43 and performance labeling obtained by inflated processing.
Then, defect candidate portion 43 is considered as elliptic arc and carries out approximation with ellipse by determining defects portion 34, is based on the ellipse
To carry out the center of gravity in defect candidate portion 43 and the calculating (step S10) of long axis.Fig. 9 is for extraction defect candidate portion 43a
Characteristic quantity method the skeleton diagram that is illustrated of an example.It is shown in FIG. 9 and is calculated in lacking in oval bowing
Fall into the center of gravity c (center of fiqure) of the position of the inside of candidate portion 43a.
Then, carry out width calculating region Lc cuts out (step S11).Determining defects portion 34 cuts out defect candidate portion 43a
On long axis direction with predetermined length width calculate region Lc.It is according to examined that the width, which calculates region Lc,
Process segment, processing method of object 10 etc. and predetermined value.
It should be noted that if width calculating region Lc is too small, though not being then defect, it is likely to show to make
Ratio for the feature of defect, error detection is got higher.On the other hand, it if width calculating region Lc is excessive, calculates and is waited as defect
It selects the whole width B of portion 43a and the line width of non-defective candidate portion 43a, becomes difficult to extraction as defect candidate portion 43a's
The ratio of feature, the leak detection of defect is got higher.Because of such situation, own, width calculates the length of region Lc, preferably
0.3mm or more and 10mm are hereinafter, more preferably 1mm or more and 6mm or less.Width calculates region Lc and is set as such range, from
And the defect shape of checked property 10 is more accurately used as characteristic quantity and quantitatively extracted, and can be known with higher accuracy
Other defect and it is non-defective be further reduced leak detection and excessively detection.
Defect will not become magnetic discontinuous portion, change section of surface shape of defect etc., differentiate near its end and become
It must be not easy.Moreover, defect, the magnetic discontinuous portion of defect, change section of surface shape etc. will not be become, it is attached far from center of gravity
Closely, then it is bigger relative to the inclination of long axis direction, become more to be difficult to detect correct line width.It is preferred, therefore, that width
Calculate the center of gravity c that region Lc includes defect candidate portion 43a.The shape of the defect of checked property 10 is more accurately used as spy as a result,
Sign amount and quantitatively extracted, can identify defect and non-defective to be further reduced excessive detection with higher accuracy.
Here, so-called width, which calculates the center of gravity c that region Lc includes defect candidate portion 43a, to be meaned：Line is being calculated into region Lc from width
Every one end at both ends when extending to the direction vertical with long axis direction, center of gravity c is located between this 2 lines.
Then, determining defects portion 34 carries out the calculating (step S12) of width w.Width w is so that defect candidate portion 43a is received
It holds back between 2 straight lines parallel with the long axis direction in defect candidate portion 43 and the distance between 2 straight lines becomes the smallest width
Degree.
Firstly, determining defects portion 34 is described and defect candidate portion 43a in a manner of clipping defect candidate portion 43a in two outsides
Parallel 2 straight lines s1, the s2 of long axis direction.In the illustration of Fig. 9, describing in the curved inside of defect candidate portion 43a has
Straight line s1, describe has straight line s2 on the outside.Moreover, determining defects portion 34 is parallel with long axis direction not to keep 2 straight lines s1, s2
The mode of change makes 2 straight lines s1, s2 respectively close to defect candidate portion 43a.That is, straight line s1 connects along the direction shown in arrow d1
Closely, straight line s2 is along the direction class shown in arrow d2.Also, straight line s1 and straight line s2 are respectively with defect candidate portion 43a's
Position, the i.e. intersection point v1 and intersection point v2 that profile initially intersects are determined.Also, between 2 straight lines s1, the s2 so determined
Distance be defined as the width w of defect candidate portion 43a.
Determining defects portion 34 by width w within a predetermined range in the case where defect candidate portion 43a sentenced
It is set to defect, to carry out the detection (step S13) of defect.Also, for defect candidate shown in Fig. 8 portion 43e and 43f
It is carried out similarly the detection of defect.
Figure 10 is to show the skeleton diagram of the image of an example for the defect 50 being extracted.3 defect candidates in fig. 8
In portion 43a, 43e and 43f, remove except defect candidate portion 43e and 43f, only defect candidate portion 43a is judged as defect
50.Also, the position data for the defect 50 that detection device 30 is detected to controller C output.
Be formed in checked property 10 magnetic powder instruction pattern in, because its formation will thus show feature.Such as it is harmful
The magnetic powder instruction pattern of damage has clear-cut, wire shaped, without the such feature of interruption, on the other hand, harmless damage
Magnetic powder indicates that pattern has feature as unintelligible profile, zigzag, multiple interrupt etc..If carrying out figure in magnetic-particle test
As processing, then the feature of such magnetic powder instruction pattern especially significantly shows on width w.Also, present embodiment passes through
The value of width w is reflected in determinating reference, thus other than the magnetic powder instruction pattern generated because carrying out harmful damage is with this
Magnetic powder instruction pattern identification.Therefore, the defects detection side realized using magnetic powder inspection device 1 according to the present embodiment
Method can capture the feature of new magnetic powder instruction pattern form as the value on image procossing, be able to carry out because of harmful damage
The higher identification of accuracy of wound and the magnetic powder instruction pattern generated and the magnetic powder instruction pattern other than this.
It should be noted that defect candidate portion 43 whether be defect 50 judgement, can also be also based on other than width w
Other decision contents carry out.At this point, such as area, length can also can be used used as decision content in the same manner as step S4
Degree, longitudinal width, transverse width, and based on the method compared with determining defects a reference value to determine whether defect.Also,
At this point, the judgement carried out based on width w can also be made preferential.It should be noted that determining defects a reference value can be and the
The second determinating reference value in two extraction units 33 is compared, region is difficult to the value being extracted as defect.
It should be noted that detection device 30 is also configured to, original image, is calculated the image extracted in each step
The data such as width w out, decision content are suitably stored in main storage portion.
Moreover, detection device 30 is configured to suitably change each threshold value, each determinating reference value, width calculating area
The value in domain etc..Thus.The size for the defect being detected as defect can be suitably changed, defect can be suitably adjusted
Check precision, ease of use is good.
As described above, in the present embodiment, checked property 10 is shot using camera 16, detection device 30
The defect inspection method realized from the magnetic powder inspection device 1 of the defect of the original image detection checked property 10 obtained by camera 16
In, determining defects portion 34 possessed by detection device 30, in the defect candidate portion 43 extracted from original image in long axis direction
Width with predetermined length calculates in the Lc of region, and defect candidate portion 43 converges on the long axis side with defect candidate portion 43
To between parallel 2 straight lines s1, s2 and the distance between 2 straight lines s1, s2 become the smallest width w predetermined
In the case where in range, defect candidate portion 43 is determined as defect 50.
The shape of the defect of checked property 10 is quantitatively extracted as characteristic quantity as a result, can be come with high accuracy
Identify defect and non-defective to reduce excessively detection.Therefore, according to the present embodiment, it is capable of providing and is mentioned using the recall rate of defect
The defect inspection method that high magnetic powder inspection device 1 is realized.
Moreover, the magnetic powder inspection device 1 of present embodiment have the camera 16 that checked property 10 is shot and from by
The detection device 30 of the defect for the original image detection checked property 10 that camera 16 obtains, detection device 30 have determining defects portion
34, which has on long axis direction predetermined in the defect candidate portion 43 extracted from original image
The width of length calculates in the Lc of region, and it is straight that defect candidate portion 43 converges on 2 parallel with the long axis direction in defect candidate portion 43
Between line s1, s2 and the case where the distance between 2 straight lines s1, s2 become the smallest width w within a predetermined range
Under, defect candidate portion 43 is determined as defect 50.
The shape of the defect of checked property 10 is quantitatively extracted as characteristic quantity as a result, can be come with high accuracy
Identify defect and non-defective to reduce excessively detection.Therefore, according to the present embodiment, the recall rate for being capable of providing defect improves
Magnetic powder inspection device 1.
It should be noted that the magnetic powder inspection device 1 as failure detector is not limited to above-mentioned composition.For example, can also
To be, controller C and detection device 30 are integrally constituted.That is, controller C is also possible to have the first extraction unit 31, test zone
The composition of domain generating unit 32, the second extraction unit 33 and determining defects portion 34.Due to being set as such composition, magnetic powder is visited
Hurting device 1 can simplify, miniaturization can be achieved.Moreover, detection device 30 is also possible to have the monitor for showing following data
It constitutes：Extract the image in the first defect candidate portion 40, the inspection area 41 of generation, the figure for extracting the second defect candidate portion 42
Picture, the image in defect candidate portion 43 obtained by carry out expansion process of the second defect candidate portion 42, the defect 50 of detection image,
The data of the decision content of calculating etc..By such composition, the testing result of defect, ease of use can suitably be determined
It is good.
Moreover, magnetic powder inspection device 1 is also possible to be also equipped with the composition for another controller for being linked to controller C.It is another
Controller, in the same manner as controller C, by the main memory saving of carry out calculation process and the processing unit and preservation data of control processing
It sets etc. and to constitute, for example, have the microcomputer of CPU, main storage means, timer, input circuit, output circuit, power circuit etc.
Calculation machine.Controller C will extract the image in the first defect candidate portion 40, the inspection area 41 of generation, extract the second defect time
The image in portion 42, the second defect candidate portion 42 is selected to carry out the defect of the image in defect candidate portion 43 obtained by expansion process, detection
50 image, calculating the data of decision content etc. sent to another controller.On the other hand, another controller will be from controller C
The data such as these data sent and size, the material of the checked property 10 that pre-save are stored in its primary storage in association
Device.
Due to being set as such composition, the mapping of the defect 50 of checked property 10 can be made by another controller
The production management of data, checked property 10 can easily be done.It should be noted that controller C is also configured to, from another
One controller receives the data such as size, the material of checked property 10, and makes the mapping data of the defect 50 of checked property 10.
Industrial availability
The failure detector of the disclosure is not limited to magnetic powder inspection device 1, for example, it can be use penetrating fluid to be examined
The penetrant inspection unit that the defect on the surface of object 10 is detected a flaw can be applied to have and clap the surface of checked property 10
The input unit taken the photograph and the original image obtained by input unit is handled to detect the detection device of the defect on surface
, all failure detectors.
Description of symbols
1 magnetic powder inspection device (failure detector)
10 checked properties
16 cameras (input unit)
30 detection devices
31 first extraction units
32 inspection area generating units
33 second extraction units
34 determining defects portions
40 first defect candidate portions
41 inspection areas
42 second defect candidate portions
43 defect candidate portions
50 defects
C center of gravity (center of fiqure)
Lc width calculates region
S1, s2 straight line
W width
Claims (14)
1. a kind of failure detector, has：
The input unit that checked property is shot；With
Detection device detects the defect of the checked property from the original image obtained by the input unit,
The failure detector is characterized in that,
The detection device has determining defects portion,
The determining defects portion has on long axis direction in the defect candidate portion extracted from the original image and predefines
Length width calculate region in, defect candidate portion converge on it is parallel with the long axis direction in defect candidate portion
2 straight lines between and the distance between 2 straight lines be width within a predetermined range in the case where, by the defect
Candidate portion is determined as the defect.
2. failure detector according to claim 1, which is characterized in that
The width calculates the center of gravity that region includes defect candidate portion.
3. failure detector according to claim 1 or 2, which is characterized in that
The detection device also has：
First extraction unit carries out binaryzation to the original image with first threshold to extract the first defect candidate portion；
Inspection area generating unit generates inspection area in a manner of comprising the first defect candidate portion；And
Second extraction unit carries out binaryzation to the inspection area with the second threshold smaller than the first threshold to extract the
Two defect candidate portions,
Defect candidate portion is region obtained by the inflated processing in the second defect candidate portion.
4. failure detector according to claim 3, which is characterized in that
The inspection area is by expanding from the smallest rectangle containing the first defect candidate portion by predetermined amplitude
The region that big rectangle is surrounded.
5. failure detector according to claim 3 or 4, which is characterized in that
The expansion process is the processing for expanding the second defect candidate portion to respective at least long axis direction.
6. failure detector according to any one of claim 1 to 5, which is characterized in that
The failure detector is also equipped with：
The magnetic powder dissemination apparatus of magnetic powder is applicable in the checked property；With
Magnetizing assembly magnetizes the checked property and is formed because the magnetic powder that leakage magnetic flux generates indicates pattern.
7. failure detector according to any one of claim 1 to 6, which is characterized in that
The length that the width calculates region is 0.3mm or more and 10mm or less.
It is the defect inspection method using failure detector 8. a kind of defect inspection method, in the failure detector,
Input unit shoots checked property,
Detection device detects the defect of the checked property from the original image obtained by the input unit,
The defect inspection method is characterized in that,
Determining defects portion possessed by the detection device, in the defect candidate portion extracted from the original image in long axis side
The width with predetermined length calculates in region upwards, defect candidate portion converges on and defect candidate portion
Between 2 parallel straight lines of the long axis direction and the distance between 2 straight lines are the feelings of width within a predetermined range
Under condition, defect candidate portion is determined as the defect.
9. defect inspection method according to claim 8, which is characterized in that
The width calculates the center of gravity that region includes defect candidate portion.
10. defect inspection method according to claim 8 or claim 9, which is characterized in that
The first extraction unit that the detection device also has carries out binaryzation to the original image with first threshold to extract first
Defect candidate portion,
Inspection area generating unit generates inspection area in a manner of comprising the first defect candidate portion,
Second extraction unit carries out binaryzation to the inspection area to extract second with the second threshold smaller than the first threshold
Defect candidate portion,
Defect candidate portion is region obtained by the inflated processing in the second defect candidate portion.
11. defect inspection method according to claim 10, which is characterized in that
The inspection area is by expanding from the smallest rectangle containing the first defect candidate portion by predetermined amplitude
The region that big rectangle is surrounded.
12. defect inspection method described in 0 or 11 according to claim 1, which is characterized in that
The expansion process is the processing for expanding the second defect candidate portion to respective at least long axis direction.
13. the defect inspection method according to any one of claim 8 to 12, which is characterized in that
The failure detector is also equipped with：
The magnetic powder dissemination apparatus of magnetic powder is applicable in the checked property；With
Magnetizing assembly magnetizes the checked property and is formed because the magnetic powder that leakage magnetic flux generates indicates pattern.
14. the defect inspection method according to any one of claim 8 to 13, which is characterized in that
The length that the width calculates region is 0.3mm or more and 10mm or less.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2016-062233 | 2016-03-25 | ||
| JP2016062233A JP6697302B2 (en) | 2016-03-25 | 2016-03-25 | Flaw detection device and defect detection method using flaw detection device |
| PCT/JP2017/011857 WO2017164338A1 (en) | 2016-03-25 | 2017-03-23 | Flaw detection apparatus and flaw detection method using flaw detection apparatus |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108885193A true CN108885193A (en) | 2018-11-23 |
Family
ID=59899620
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201780019559.0A Pending CN108885193A (en) | 2016-03-25 | 2017-03-23 | Failure detector and the defect inspection method for utilizing failure detector |
Country Status (4)
| Country | Link |
|---|---|
| JP (1) | JP6697302B2 (en) |
| KR (1) | KR102342791B1 (en) |
| CN (1) | CN108885193A (en) |
| WO (1) | WO2017164338A1 (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020133954A1 (en) * | 2018-12-27 | 2020-07-02 | 帝沃检测技术(上海)有限公司 | Digital performance analysis method and device |
| CN113167567A (en) * | 2019-01-25 | 2021-07-23 | 东丽株式会社 | Structure inspection method and manufacturing method, structure inspection device and manufacturing device |
| CN114072668A (en) * | 2019-06-28 | 2022-02-18 | 杰富意钢铁株式会社 | Method and apparatus for inspecting surface defect of round bar |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110892255B (en) * | 2017-07-07 | 2024-03-01 | 佳能株式会社 | Image processing apparatus, method and storage medium for detecting defect from image |
| CN109557172B (en) * | 2018-12-29 | 2024-04-02 | 东方电气集团东方锅炉股份有限公司 | Test block and test method for magnetic powder detection depth test |
| JP2021043095A (en) | 2019-09-12 | 2021-03-18 | 株式会社ジェイテクト | Magnetic particle inspection device |
| CN115115612B (en) * | 2022-07-26 | 2022-11-04 | 山东安德机械科技有限公司 | Surface defect detection method and system for mechanical parts |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0882616A (en) * | 1994-09-13 | 1996-03-26 | Sumitomo Metal Ind Ltd | Method and apparatus for inspecting flaw with magnetic particle |
| JPH09210969A (en) * | 1996-01-31 | 1997-08-15 | Nippon Steel Corp | Automatic magnetic particle inspection device |
| JP2000002523A (en) * | 1998-06-18 | 2000-01-07 | Ohbayashi Corp | Crack measuring method for structure surface |
| JP2003168114A (en) * | 2001-12-04 | 2003-06-13 | Olympus Optical Co Ltd | Defect sorting device |
| JP2005069887A (en) * | 2003-08-25 | 2005-03-17 | Nippon Steel Corp | Defect inspection method and apparatus |
| CN101082592A (en) * | 2006-06-01 | 2007-12-05 | 大日本网目版制造株式会社 | Uneven checking method and device and recording medium |
| CN101449149A (en) * | 2006-05-23 | 2009-06-03 | 麒麟工程技术系统公司 | Surface examining device |
| CN101484818A (en) * | 2006-06-14 | 2009-07-15 | 浜松光子学株式会社 | Semiconductor defect analysis device, defect analysis method, and defect analysis program |
| KR20090113489A (en) * | 2008-04-28 | 2009-11-02 | 선문대학교 산학협력단 | Method for Detecting Defect in Glass Panel |
| CN102027364A (en) * | 2008-05-15 | 2011-04-20 | 住友金属工业株式会社 | Magnetic flaw detecting method and magnetic flaw detection device |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3077793B2 (en) * | 1996-02-27 | 2000-08-14 | 住友金属工業株式会社 | Automatic magnetic particle inspection equipment |
| WO2000060344A1 (en) | 1999-03-31 | 2000-10-12 | Hitachi, Ltd. | Method and apparatus for non destructive testing |
| JP2001281226A (en) | 2000-03-29 | 2001-10-10 | Daido Steel Co Ltd | Method and apparatus for fluorescent magnetic-particle flaw detection |
| JP4970101B2 (en) | 2007-03-26 | 2012-07-04 | 大日本スクリーン製造株式会社 | Defect detection method |
| JP5086970B2 (en) | 2008-11-05 | 2012-11-28 | パナソニック株式会社 | Wood appearance inspection device, wood appearance inspection method |
| KR20110038796A (en) | 2009-10-09 | 2011-04-15 | 김정인 | Twist thread |
| US8575923B1 (en) * | 2011-01-07 | 2013-11-05 | OilPatch Technology | Method and apparatus for special end area inspection |
| JP5702639B2 (en) * | 2011-03-25 | 2015-04-15 | 株式会社Ｓｃｒｅｅｎホールディングス | Image acquisition device for unevenness inspection, unevenness inspection device, and position determination method of irradiation unit |
| KR20140082335A (en) * | 2012-12-24 | 2014-07-02 | 엘지디스플레이 주식회사 | Method and apparatus of inspecting mura of flat display |
| JP6596188B2 (en) * | 2015-04-02 | 2019-10-23 | マークテック株式会社 | Flaw detector and flaw detection method using flaw detector |
-
2016
- 2016-03-25 JP JP2016062233A patent/JP6697302B2/en active Active
-
2017
- 2017-03-23 KR KR1020187029124A patent/KR102342791B1/en active IP Right Grant
- 2017-03-23 CN CN201780019559.0A patent/CN108885193A/en active Pending
- 2017-03-23 WO PCT/JP2017/011857 patent/WO2017164338A1/en active Application Filing
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0882616A (en) * | 1994-09-13 | 1996-03-26 | Sumitomo Metal Ind Ltd | Method and apparatus for inspecting flaw with magnetic particle |
| JPH09210969A (en) * | 1996-01-31 | 1997-08-15 | Nippon Steel Corp | Automatic magnetic particle inspection device |
| JP2000002523A (en) * | 1998-06-18 | 2000-01-07 | Ohbayashi Corp | Crack measuring method for structure surface |
| JP2003168114A (en) * | 2001-12-04 | 2003-06-13 | Olympus Optical Co Ltd | Defect sorting device |
| JP2005069887A (en) * | 2003-08-25 | 2005-03-17 | Nippon Steel Corp | Defect inspection method and apparatus |
| CN101449149A (en) * | 2006-05-23 | 2009-06-03 | 麒麟工程技术系统公司 | Surface examining device |
| CN101082592A (en) * | 2006-06-01 | 2007-12-05 | 大日本网目版制造株式会社 | Uneven checking method and device and recording medium |
| CN101484818A (en) * | 2006-06-14 | 2009-07-15 | 浜松光子学株式会社 | Semiconductor defect analysis device, defect analysis method, and defect analysis program |
| KR20090113489A (en) * | 2008-04-28 | 2009-11-02 | 선문대학교 산학협력단 | Method for Detecting Defect in Glass Panel |
| CN102027364A (en) * | 2008-05-15 | 2011-04-20 | 住友金属工业株式会社 | Magnetic flaw detecting method and magnetic flaw detection device |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020133954A1 (en) * | 2018-12-27 | 2020-07-02 | 帝沃检测技术(上海)有限公司 | Digital performance analysis method and device |
| CN113167567A (en) * | 2019-01-25 | 2021-07-23 | 东丽株式会社 | Structure inspection method and manufacturing method, structure inspection device and manufacturing device |
| CN114072668A (en) * | 2019-06-28 | 2022-02-18 | 杰富意钢铁株式会社 | Method and apparatus for inspecting surface defect of round bar |
Also Published As
| Publication number | Publication date |
|---|---|
| JP2017173251A (en) | 2017-09-28 |
| KR102342791B1 (en) | 2021-12-22 |
| WO2017164338A1 (en) | 2017-09-28 |
| JP6697302B2 (en) | 2020-05-20 |
| KR20180123088A (en) | 2018-11-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108885193A (en) | Failure detector and the defect inspection method for utilizing failure detector | |
| CN106053593B (en) | Flaw detection device and flaw detection method using flaw detection device | |
| CN105849534B (en) | Detection method of surface flaw and surface defect detection apparatus | |
| CN113588692B (en) | Computer vision-based weld defect nondestructive testing method | |
| JP2011013007A (en) | Magnetic particle flaw inspection apparatus | |
| JP2020008501A (en) | Surface defect detection device and surface defect detection method | |
| CN107709977B (en) | Surface defect detection device and surface defect detection method | |
| WO2019193912A1 (en) | Drug determination device and drug determination method | |
| JP2009008659A (en) | Surface flaw detection device | |
| CN115731222A (en) | Billet surface detection method, device, equipment and computer readable storage medium | |
| JP4618501B2 (en) | Fluorescence flaw detector and fluorescent flaw detection method | |
| JP2010025586A (en) | Configuration determination method and configuration determination apparatus | |
| JP2009074952A (en) | Visual inspection method | |
| Ma et al. | A machine vision assisted system for fluorescent magnetic particle inspection of railway wheelsets | |
| JP2009109243A (en) | Device for inspecting resin sealing material | |
| JP2012108026A (en) | Detection method of foreign material-contaminated product | |
| JP3440569B2 (en) | Magnetic particle flaw detection method and apparatus | |
| KR100957130B1 (en) | Surface inspection apparatus for case, and surface inspection method using the same | |
| JP5889426B1 (en) | Web inspection device, web inspection method, web inspection program | |
| JPH09210969A (en) | Automatic magnetic particle inspection device | |
| JP2015059854A (en) | Defect inspection method and defect inspection device | |
| JP2000182114A (en) | Watermark inspecting device | |
| JP2006284308A (en) | Visual examination method of semiconductor device | |
| JP2000233220A (en) | Method for detecting weld zone of steel strip | |
| JP3389692B2 (en) | Magnetic particle flaw detection method and apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20181123 |
|
| WD01 | Invention patent application deemed withdrawn after publication |