CN106290392A - A kind of little micro-bearing surface pitting defects online test method and system thereof - Google Patents

A kind of little micro-bearing surface pitting defects online test method and system thereof Download PDF

Info

Publication number
CN106290392A
CN106290392A CN201610641082.7A CN201610641082A CN106290392A CN 106290392 A CN106290392 A CN 106290392A CN 201610641082 A CN201610641082 A CN 201610641082A CN 106290392 A CN106290392 A CN 106290392A
Authority
CN
China
Prior art keywords
bearing
image
character
processing module
bearing surface
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
Application number
CN201610641082.7A
Other languages
Chinese (zh)
Inventor
陈伟庆
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ningbo Daer Machinery Technology Co Ltd
Original Assignee
Ningbo Daer Machinery Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ningbo Daer Machinery Technology Co Ltd filed Critical Ningbo Daer Machinery Technology Co Ltd
Priority to CN201610641082.7A priority Critical patent/CN106290392A/en
Publication of CN106290392A publication Critical patent/CN106290392A/en
Pending legal-status Critical Current

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
    • G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
    • 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
    • G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
    • 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
    • G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
    • G01N21/952—Inspecting the exterior surface of cylindrical bodies or wires
    • 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
    • G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
    • G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques

Abstract

The invention belongs to bearing surface defect detection field, especially relate to a kind of little micro-bearing surface pitting defects online test method and system thereof.It is an object of the invention to break through the deficiency of existing detection method, it is according to the feature gathering bearing image, realize micro bearing external defects based on ccd image identification technology and automatically detect screening, it is according to the feature of the collection bearing image of the different parts adopted, design one and can detect bearing both ends of the surface by parallelism recognition, bearing top circle surface and little micro-bearing surface pitting defects on-line detecting system of end face image character, bearing different parts can be carried out parallel parsing detection simultaneously, treatment effeciency is high, micro bearing surface defects detection operation is made to realize automatization, with the method substituting the commonly used manual detection of current enterprise, improve precision and efficiency of detecting, make defect recognition rate, the clearance of substandard product reaches higher level, improve product quality, improve the credit worthiness of enterprise, improve the level of modernization of enterprise.

Description

A kind of little micro-bearing surface pitting defects online test method and system thereof
Technical field
The invention belongs to bearing surface defect detection field, especially relate to a kind of little micro-bearing surface pitting defects online Detection method and system thereof.
Background technology
Bearing is a very important part of machinery industry, and its use is the most universal.The machining accuracy of bearing and matter Magnitude relation is to the serviceability of engineering goods and life-span, and therefore the crudy to bearing detects always Production of bearing producer and closes The problem of the heart.In industrial processes, in order to get rid of substandard products and find out the product hidden danger etc. relevant with fault, product to be done Quality examination, in the past many is completed by reviewer, and this method varies with each individual the shortcomings such as poor reliability.
In recent years, along with the development of China's bearing industry, bearing has become as important export-oriented commodity.Respectively produce enterprise Industry passes through technological transformation, makes to have employed automatization's unit or automatic production line in Production of bearing, and strides forward to modernization, but only Having surface defects detection and still use artificial visual method with rejecting, not only efficiency is low, and poor reliability, with automatization's processing line The most unbecoming, the most each enterprise is also badly in need of a kind of automatic inspection line and detects the crudy of bearing.Due to small axle Hold the load mould on surface, scratch its size, the degree of depth and distributing position are all random.Therefore, by contact measurement not only difficulty Greatly, and efficiency is low, and uses ccd image identification technology to do non-contact detecting, is the preferred approach solving this problem.
Existing based on Computer Vision Detection Technique Bearing testing apply in, bearing one side can only be realized and automatically exist Line detects, and needs artificial upset another side to detect again, and is sampling observation, it is impossible to realizes online automatization's complete detection. Simultaneously as micro bearing surface is little, and be imprinted with label, the load mould on surface, scuffing defect the most small, and surface is past Toward there being one layer of oil film, affect incident intensity and the direction of light source, the most how to select light source, how to shield the interference of ambient light with And the most rightly by pattern visual evoked potentials, and be not distorted, it is by the key issue of defect recognition.
Meanwhile, bearing character is the important symbol that all bearing products are the most indispensable, the most all indicates On the non-referenced end face of lasso, it is also possible to require to be marked on reference face of ring, retainer, sealing ring and dust cap according to technology Deng.The information that can obtain from bearing character includes bearing designation and manufacturer's title, the most also includes factory Other additional information that family formulates jointly with user.Current domestic Production of bearing producer still uses this workload of artificial range estimation Greatly, the detection of inefficient method complete matched bearings character, with bearing enterprise production scaleization extremely contradiction.Bearing character defect The weak link of detection always bearing industry detection technique, consults each big open bibliographic data base, lacks bearing character at present Fall into detection research almost without, there is no the correlation-detection system detection for bearing character defect yet.
Therefore, it is necessary to according to the feature of micro bearing, design a kind of little micro-bearing surface pitting defects on-line checking Method and system thereof, it is possible to the defects such as the most quick scuffing to product surface, deformation detect, it is ensured that product quality, Improve precision and efficiency of detecting, thoroughly solve the hard nut to crack of this puzzlement enterprise, be necessary.
Summary of the invention
For above technical problem, it is an object of the invention to break through the deficiency of existing detection method, it is provided that Yi Zhong little Micro-bearing surface pitting defects online test method and system thereof, realize micro bearing outer surface based on ccd image identification technology Defect detects automatically, makes micro bearing surface defects detection operation realize automatization, commonly used manually to substitute current enterprise The method of detection, improves precision and efficiency of detecting, makes defect recognition rate, the clearance of substandard product reaches higher level, carry Height dispatches from the factory the oeverall quality of product, avoids actual loss as far as possible, improves the credit worthiness of enterprise, improves the modernization water of enterprise Flat.
For reaching above-mentioned purpose, the present invention provides following technical scheme:
A kind of little micro-bearing surface pitting defects online test method, its concrete detecting step is:
1) image data acquiring
By micro bearing automatic transmission system, bearing to be detected being positioned over on-line detecting system, on-line detecting system is adjusted Whole LED lamplight, and utilize CCD camera to carry out synchronizing to capture to both ends of the surface and the outer round surface image of bearing to be detected, and will The image collected is transferred in the computer processing module of correspondence by image pick-up card；
2) image information pretreatment
A) to the end face image information collected at end face image processing module and end view drawing as character recognition module respectively Carrying out pretreatment, wherein, the pretreatment of end face image processing module includes that image filtering denoising and image binaryzation process, end face The pretreatment of image character identification module includes that image filtering denoising, image binaryzation, character normalization, thinning process；
B) the outer round surface image information collected is carried out image filtering denoising, figure at outer round surface image processing module As binary conversion treatment；
3) bearing template matching defect recognition
End face image processing module carries out edge extracting and template matching to pretreated end face image information, is divided Analysis result；End face image character processing module carries out bearing character recognition and template matching to end face image information, is analyzed Result；Outer round surface image processing module surface image information to foreign round carries out edge extracting and template matching, obtains analyzing knot Really；
4) feedback screening
Analysis result is carried out feeding back to computer control module by each computer processing module, and computer control module controls Memory module carries out result storage, and controls to screen switch and screen, to realize the online of little micro-bearing surface pitting defects The automatic sorting of detection and substandard product processes.
As preferably, in step 2) in, the pretreatment of end face image processing module includes image filtering denoising and image two Value processes, and its image filtering uses BLPF algorithm that bearing image carries out denoising, and image binaryzation processes and uses iteration Threshold method realizes the binaryzation of bearing image.
As preferably, in step 2) in, end view drawing is as the concretely comprising the following steps of pretreatment of character recognition module: select 5 × 5 Median filtering method carries out denoising, and uses sciagraphy extract single bearing character and carry out realizing its local binary with Otsu method Change, the single binaryzation character obtained is normalized and micronization processes.
As preferably, in step 2) in, outer round surface image processing module carries out image filtering denoising with medium filtering, and Image binaryzation process is carried out with minimum error method.
As preferably, in step 3) in, first spcial character is processed by end face image character processing module, then adopts Realize the extraction to character feature by grid and intersection feature, and use template matching method based on improvement to carry out coupling point Analysis, obtains testing result.
A kind of little micro-bearing surface pitting defects on-line detecting system, described on-line detecting system mainly includes that CCD takes the photograph Camera, image pick-up card, computer, control unit, propulsive mechanism, conveyer belt, screening switch, light source, wherein, on-line checking system The conveyer belt of system is connected with production line, and ccd video camera and light source are arranged on above conveyer belt, and ccd video camera passes through image acquisition Card is connected with computer, and computer is connected with propulsive mechanism and screening switch by control unit.
As preferably, described computer includes that end face image processing module, end view drawing are as character recognition module, cylindrical table Face image processing module, memory module and computer control module, end face image processing module, end view drawing are as character recognition mould Block, outer round surface image processing module concurrent working, and analysis result is transmitted to memory module and computer control module.
As preferably, described CCD selects Basler digital camera A102f.
Beneficial effects of the present invention:
The present invention substitutes the method for the commonly used manual detection of current enterprise, improves precision and efficiency of detecting, makes defect know Not rate, the clearance of substandard product reaches higher level, improves the oeverall quality of the product that dispatches from the factory, avoids actual damage as far as possible Losing, improve the credit worthiness of enterprise, improve the level of modernization of enterprise, it is in particular in:
1, the present invention realizes micro bearing external defects based on ccd image identification technology and automatically detects, and it is according to adopting Different parts collection bearing image feature, design one can parallelism recognition detection bearing both ends of the surface, bearing top circle surface With little micro-bearing surface pitting defects on-line detecting system of end face image character, can bearing different parts be carried out parallel simultaneously Analyzing detection, treatment effeciency is high.
2, the present invention is directed to bearing different parts carries out parallel analysis detection, its analysis detection algorithm selected is root It is optimized according to the feature at each position of bearing, as the image of bearing both ends of the surface and outer round surface be have employed different filtering Denoising and image binaryzation process, and specific aim is higher, and treatment effeciency and accuracy are the highest.
3, for the problem that existing system middle (center) bearing character machining accuracy is the highest, the present invention uses 5 × 5 median filtering methods Remove the noise in image, use a kind of quickly Otsu method to carry out image binaryzation, utilize the way of fitting circle to determine bearing The center of circle and then position bearing, have studied normalization and the thinning algorithm of character, uses grid and intersection feature to extract Going out character feature, finally select the template matching method identification bearing character of a kind of improvement, these algorithms make the detection efficiency of system Being greatly improved with precision, its discrimination has reached 99.375%, analyzes that bearing character is normal and the situation of defect, to product Raw error is analyzed, and makes micro bearing surface defects detection operation realize automatization.
Accompanying drawing explanation
Fig. 1 is the structural representation of the present invention.
Detailed description of the invention
Below in conjunction with the embodiment of the present invention and accompanying drawing, the technical scheme in the embodiment of the present invention is carried out clear, complete Ground describes, it is clear that described embodiment is only a part of embodiment of the present invention rather than whole embodiments.Based on this Embodiment in invention, the every other reality that those of ordinary skill in the art are obtained under not making creative work premise Execute example, broadly fall into the scope of protection of the invention.
As it is shown in figure 1, a kind of little micro-bearing surface pitting defects on-line detecting system, described on-line detecting system is main Including ccd video camera, image pick-up card, computer, control unit, propulsive mechanism, conveyer belt, screening switchs, light source, wherein, The conveyer belt of on-line detecting system is connected with production line, and ccd video camera and light source are arranged on above conveyer belt, and ccd video camera leads to Crossing image pick-up card to be connected with computer, computer is connected with propulsive mechanism and screening switch by control unit, wherein, described Computer include end face image processing module, end view drawing as character recognition module, outer round surface image processing module, storage mould Block and computer control module, end face image processing module, end view drawing are as character recognition module, outer round surface image processing module Concurrent working, and analysis result is transmitted to memory module and computer control module.
CCD selects: CCD selects Basler digital camera A102f, and its technical characteristics is: super-resolution degree 1392 (level) * 1040 (vertically) pixel, every Pixel Dimensions 6.45m*6.45m, electronic type full frame shutter, ISO, high noise Ratio, shutter control time 20s～1.31s, RS-232 programming Control can be passed through, the requirement of image recognition can be met completely.Figure TEC-M55 telecentric lens is used: focal length 55mm, maximum diameter of hole be the image full-size of imaging than 1 2.8, on CCD as amplifying It is infinity～0.5* for 8.8mm × 6.6mm or Ф 11mm, amplification.
Computer selects: be configured that more than CPU 2.8G, internal memory 512M, video card 128M, hard disk 80G needed for computer.Meter Calculation machine include end face image processing module, end view drawing as character recognition module, outer round surface image processing module, memory module and Computer control module, end face image processing module, end view drawing are parallel as character recognition module, outer round surface image processing module Work, and analysis result is transmitted to memory module and computer control module.The system of the present invention applies image acquisition, divides Analysis, the real-time processing technique differentiated and light mechanical and electrical integration, on-line automaticization that can realize bearing detects and defective The automatic sorting of product processes, and preferably meets real-time and the accuracy requirement of detection.This system is easy and simple to handle, interface is friendly Good, reliable, may also be combined with database technology, by testing result data real-time storage, to carry out quality analysis process simultaneously. Meanwhile, the present invention can detect bearing both ends of the surface, bearing top circle surface and little micro-bearing surface of end face image character by parallelism recognition Pitting defects on-line detecting system, can carry out parallel parsing detection simultaneously to bearing different parts, and treatment effeciency is high.
A kind of little micro-bearing surface pitting defects online test method, its concrete detecting step is:
1) image data acquiring
By micro bearing automatic transmission system, bearing to be detected being positioned over on-line detecting system, on-line detecting system is adjusted Whole LED lamplight, and utilize CCD camera to carry out synchronizing to capture to both ends of the surface and the outer round surface image of bearing to be detected, and will The image collected is transferred in the computer processing module of correspondence by image pick-up card；
2) image information pretreatment
A) to the end face image information collected at end face image processing module and end view drawing as character recognition module respectively Carry out pretreatment.
The pretreatment of end face image processing module includes that image filtering denoising and image binaryzation process, and its image filtering is adopted With BLPF, bearing image being carried out denoising, Butterworth LPF BLPF is maximum flat filter, its passband And not having obvious discontinuity between stopband, therefore its spatial domain response does not has " ring " phenomenon to occur, and fog-level reduces, The present invention is directed to end face image information uses BLPF that bearing image is carried out denoising, and its picture quality is carried greatly High.Due to bearing defect and character gray value closely, it is difficult to from image, directly identify defect characteristic；And defect characteristic The gray value of peripheral region is bigger with the grey value difference of background.In order to improve detection efficiency and precision, system is with defect characteristic Peripheral region is detection target, image carries out binary conversion treatment, bearing face image is divided into highlight bar around defect characteristic Territory and background area two parts.Therefore, image binaryzation processes the binaryzation using iteration method to realize bearing image, its tool Body step is: selects an approximate threshold as the initial value of estimated value, then splits, and produces subimage, and according to son The characteristic of image chooses new threshold value, then with new Threshold segmentation image, circulates through several times, makes the image slices of erroneous segmentation Vegetarian refreshments drops to minimum.The gray scale of the bearing image after pretreatment, the gray scale of its bearing face and defect peripheral region is all There is contrast clearly.Therefore, through above pretreatment, for realizing the identification of computer, detection provides reliable foundation.
End view drawing includes image filtering denoising, image binaryzation, character normalization, word as the pretreatment of character recognition module Symbol micronization processes；End view drawing is as the concretely comprising the following steps of pretreatment of character recognition module:
I. selecting 5 × 5 median filtering methods to carry out denoising, median filtering method does not results in image detail and thickens, it is ensured that The character feature effect that subsequent extracted goes out, has the accurate identification utilizing realization to character.The selection of window size has 3 × 3,5 × 5,7 × 7 etc., the size of window selects accordingly according to the size of noise.By repetition test, the present invention selects window to be in 5 × 5 Value filtering method；
Ii. sciagraphy is used to extract single bearing character and carry out realizing its local binarization, due to axle with Otsu method Holding image and can not cover all gray scales from 0 to 255, therefore carried out algorithm optimization on the basis of Otsu method, the method is permissible Greatly reduce the number of times that required variance calculates, thus improve program efficiency, implement process as follows:
If the gray value of piece image has 0 to arrive m-1 level, m is the number of greyscale levels that image comprises；niIt is the picture of i for gray value Prime number, then sum of all pixels is:
N = Σ i = 0 m - 1 n i - - - ( 1 - 1 )
The probability of each gray value:
p i = n i N - - - ( 1 - 2 )
Use a k value that 0 to m-1 level is divided into two groups: C0=[0,1,2 ..., k] and C1=[k+1 ..., m-1], then Calculate C0And C1Improvement inter-class variance formula be:
σ2(k)=ω0(k)×ω1(k)×[μ0ω(k)-μ1(k)]2 (1-3)
In formula, ω0(k) and ω1K () is respectively C0And C1Total pixel number included in group；μ0(k) and μ1K () is respectively C0And C1Organize the average gray value of whole pixel.
By k from 0 to m-1 value successively, select to make σ (k)2Reach the k of maximum as threshold value.Then (1-3) formula can be derived One-tenth following formula:
σ 2 ( k ) = [ μ k ω ( k ) - μ ( k ) ] 2 ω ( k ) [ 1 - ω ( k ) ] - - - ( 1 - 4 )
Wherein:
μ K = Σ i = 0 m - 1 p i × i - - - ( 1 - 5 )
μ ( k ) = Σ i = 0 k p i × i - - - ( 1 - 6 )
ω ( k ) = Σ i = 0 k p i - - - ( 1 - 7 )
In formula, i and PiIt is respectively each gray level and its probability of occurrence in image.
It is assumed here that in image gray level be the number of pixels of k ' be zero, i.e. Pk′=0, if selecting k '-1 is threshold value Time:
μ ( k ′ - 1 ) = Σ i = 0 k ′ - 1 p i × i - - - ( 1 - 8 )
ω ( k ′ - 1 ) = Σ i = 0 k ′ - 1 p i - - - ( 1 - 9 )
μ K - 1 = Σ i = 0 m - 1 p i × i - - - ( 1 - 10 )
If gray level k ' is as threshold value in selection image:
μ ( k ′ ) = Σ i = 0 k ′ p i × i = Σ i = 0 k ′ - 1 p i × i + k ′ p k ′ = Σ i = 0 k ′ - 1 p i × i = μ ( k ′ - 1 ) - - - ( 1 - 11 )
ω ( k ′ ) = Σ i = 0 k ′ p i = Σ i = 0 k ′ - 1 p i + p k ′ = Σ i = 0 k ′ - 1 p i = ω ( k ′ - 1 ) - - - ( 1 - 12 )
μ K = Σ i = 0 m - 1 p i × i - - - ( 1 - 13 )
As the above analysis:
σ2(k '-1)=σ2(k′) (1-14)
Assume again that image has continuous gray-scales, K1,K2,K3…Kn... pixel count be zero, with reference to said method permissible Obtain:
σ2(k1-1)=σ2(k1)=σ2(k2-1)=σ2(k2) ...=σ2(kn-1)=σ2(kn) (1-15)
Owing to being affected by Production of bearing operating mode, during detection, all can in tested bearing position in camera coverage every time Having small variations, follow-up analysis, before testing it needs to be determined that the accurate location of bearing.Spy from bearing image Levying and understand, bearing character typically indicates the side at dust cap or lasso, owing to bearing is axisymmetric, and normally detects institute Shooting image center all can in bearing inner race, at this use multi-point fitting circle method determine the bearing center of circle coordinate thus Location image, after must positioning image, carries out character band expansion and Character segmentation to location image, obtains single character.
The single binaryzation character obtained is normalized and micronization processes, due to factors such as illumination, hardware and algorithms Impact, the single character boundary obtained through above-mentioned process is different, and character identifying method used herein is template Joining method, the character of same type only stores the template of a kind of size in a computer, and this just requires the single word extracted Symbol is normalized.Character normalization is also known as character normalization, and its effect is to make various sizes of character pass through After scaling processes, size is unified, during being normalized character, uses neighbor interpolation method.
B) the outer round surface image information collected is carried out image filtering denoising, figure at outer round surface image processing module As binary conversion treatment, outer round surface image processing module carries out image filtering denoising with medium filtering, and enters with minimum error method Row image binaryzation processes；
3) bearing template matching defect recognition
A. end face image processing module carries out edge extracting and template matching to pretreated end face image information, obtains Analysis result；
B. end face image character processing module carries out bearing character recognition and template matching to end face image information, is divided Analysis result；Character feature extraction is the previous the most crucial stage of character recognition, it is therefore an objective to extract most generation from character The feature of table.Feature Selection is the most suitable, is related to the accuracy of character recognition, the speed of detecting system and precision.Choose Feature must be fulfilled for following condition: one is principle of similarity, and selected feature should be that to have certain indeformable identical Type；Two is uniqueness principle, and selected feature should be mapped；Three is stability principle, selected feature Should be in each image converts with image, keep consistent.In step 3) in, end face image character processing module is first to spy Different character processes, and then uses grid and intersection feature to realize the extraction to character feature, utilizes grid and cross point Feature obtains the 29 dimensional feature data vectors about character, is stored in character standard feature database by each character feature vector, Follow-up identification is carried out afterwards according to these representative features；
Template matching method belongs to the recognition mode of a kind of discrete input pattern classification, and essence is to realize input data and sample Between similarity contrast, according to similarity size decision-making input data generic.The present invention uses a kind of improvement Template matching method carry out character recognition, the method is to carry out twice coupling realization on the basis of template matching method.Concrete side Method is as follows:
(1) rough matching process.In artwork and template, all interlacing is mated every row extraction data, the artwork of i.e. 1/4 And template data.This matching process decreases a large amount of information to be dealt with, is effectively improved matching efficiency.
(2) accurate matching process.Error smallest point process in rough matching occur carries out second time in its neighborhood Search coupling, and then it is identified result.
Template matching method ultimate principle according to above-mentioned improvement, by normalization and refinement after 25 × 25 character to be identified and Template character binaryzation, extracts grid and cross point two category feature of character after both binaryzations, obtains the characteristic vector of 29 dimensions, Character to be identified and each template are carried out twice characteristic matching, and what last similarity was maximum is the character identified.To knowledge Character and standard character the most out are compared, if inconsistent, judge character existing defects.Show through experiment, improve The more traditional template matching method of template matching method realize simple, discrimination is high, and detection speed is fast, little by external interference.This is System uses and makes bearing character defects detection reach higher level based on the template matching method improved, and its discrimination reaches 99.375%.
Surface image processing module surface image information to foreign round carries out edge extracting and template matching to foreign round, is divided Analysis result；Showing through experiment, the template matching method that the template matching method of improvement is more traditional realizes simple, and discrimination is high, detection Speed is fast, little by external interference, and native system uses the template matching method based on improving to make bearing character defects detection reach higher Level.
4) feedback screening
Analysis result is carried out feeding back to computer control module by each computer processing module, and computer control module controls Memory module carries out result storage, and controls to screen switch and screen, to realize little micro-bearing surface pitting defects The automatic sorting of on-line checking and substandard product processes.
Utilize native system that bearing carries out character recognition, each 200 times of the bearing of test Multiple Type kinds of characters situation, inspection The rate that tests the speed is overlapped less than 1s/, and discrimination reaches 99.375%, therefrom chooses 10 experimental datas (view data of other numbering is slightly) As shown in table 1.Testing result statistical table is as shown in table 2.
Table 1
Table 2
The present invention uses 5 × 5 median filtering methods to remove the noise in image, uses a kind of quickly Otsu method to carry out image two Value, utilizes the way of fitting circle determine the center of circle of bearing and then position bearing, have studied the normalization of character with thin Change algorithm, use grid and intersection feature to extract character feature, finally select the template matching method identification axle of a kind of improvement Holding character, these algorithms make the detection efficiency of system and precision be greatly improved, and its discrimination has reached 99.375%, carries High measurement accuracy and efficiency.
The present invention realizes micro bearing external defects based on ccd image identification technology and automatically detects, and its basis is adopted Different parts collection bearing image feature, design one can parallelism recognition detection bearing both ends of the surface, bearing top circle surface and Little micro-bearing surface pitting defects on-line detecting system of end face image character, can divide bearing different parts simultaneously parallel Analysis detection, treatment effeciency is high；The present invention is directed to bearing different parts carries out parallel analysis detection, the analysis detection that it is selected Algorithm is that the feature at each position according to bearing is optimized, as have employed the image of bearing both ends of the surface and outer round surface not Same filtering and noise reduction and image binaryzation process, and specific aim is higher, and treatment effeciency and accuracy are the highest, meanwhile, for existing The problem that system middle (center) bearing character machining accuracy is the highest, the present invention uses 5 × 5 median filtering methods to remove the noise in image, Use a kind of quickly Otsu method to carry out image binaryzation, utilize the way of fitting circle determine the center of circle of bearing and then bearing is carried out Location, have studied normalization and the thinning algorithm of character, uses grid and intersection feature to extract character feature, finally selects The template matching method identification bearing character of a kind of improvement, these algorithms make the detection efficiency of system and precision have the biggest proposing High.
Finally, above example and accompanying drawing are only in order to illustrate technical scheme and unrestricted, although by above-mentioned The present invention is described in detail by embodiment, it is to be understood by those skilled in the art that can in form and carefully On Jie, it is made various change, without departing from claims of the present invention limited range.

Claims (8)

1. one kind little micro-bearing surface pitting defects online test method, it is characterised in that its concrete detecting step is:
1) image data acquiring
By micro bearing automatic transmission system, bearing to be detected being positioned over on-line detecting system, on-line detecting system adjusts LED lamplight, and utilize CCD camera to carry out synchronizing to capture to both ends of the surface and the outer round surface image of bearing to be detected, and will adopt Collect to image be transferred in corresponding computer processing module by image pick-up card；
2) image information pretreatment
A) the end face image information collected is carried out as character recognition module respectively at end face image processing module and end view drawing Pretreatment, wherein, the pretreatment of end face image processing module includes that image filtering denoising and image binaryzation process, end view drawing picture The pretreatment of character recognition module includes that image filtering denoising, image binaryzation, character normalization, thinning process；
B) the outer round surface image information collected is carried out image filtering denoising, image two at outer round surface image processing module Value processes；
3) bearing template matching defect recognition
End face image processing module carries out edge extracting and template matching to pretreated end face image information, obtains analyzing knot Really；End face image character processing module carries out bearing character recognition and template matching to end face image information, obtains analysis result； Outer round surface image processing module surface image information to foreign round carries out edge extracting and template matching, obtains analysis result；
4) feedback screening
Analysis result is carried out feeding back to computer control module by each computer processing module, and computer control module controls storage Module carries out result storage, and controls to screen switch and screen, to realize the on-line checking of little micro-bearing surface pitting defects And the automatic sorting of substandard product processes.
The little micro-bearing surface pitting defects online test method of one the most according to claim 1, it is characterised in that: in step Rapid 2), in, the pretreatment of end face image processing module includes that image filtering denoising and image binaryzation process, and its image filtering is adopted With BLPF, bearing image being carried out denoising, image binaryzation processes the two-value using iteration method to realize bearing image Change.
The little micro-bearing surface pitting defects online test method of one the most according to claim 1, it is characterised in that: in step Rapid 2) in, end view drawing is as the concretely comprising the following steps of pretreatment of character recognition module: select 5 × 5 median filtering methods to carry out denoising, And use sciagraphy extract single bearing character and carry out realizing its local binarization with Otsu method, single two will obtained Value character is normalized and micronization processes.
The little micro-bearing surface pitting defects online test method of one the most according to claim 1, it is characterised in that: in step Rapid 2), in, outer round surface image processing module carries out image filtering denoising with medium filtering, and carries out image with minimum error method Binary conversion treatment.
The little micro-bearing surface pitting defects online test method of one the most according to claim 1, it is characterised in that: in step Rapid 3) in, first spcial character is processed by end face image character processing module, then uses grid and intersection feature real The now extraction to character feature, and use the template matching method based on improving to carry out the matching analysis, obtain testing result.
6. one kind little micro-bearing surface pitting defects on-line detecting system, it is characterised in that: described on-line detecting system is main Including ccd video camera, image pick-up card, computer, control unit, propulsive mechanism, conveyer belt, screening switchs, light source, wherein, The conveyer belt of on-line detecting system is connected with production line, and ccd video camera and light source are arranged on above conveyer belt, and ccd video camera leads to Crossing image pick-up card to be connected with computer, computer is connected with propulsive mechanism and screening switch by control unit.
The little micro-bearing surface pitting defects on-line detecting system of one the most according to claim 6, it is characterised in that: described Computer include end face image processing module, end view drawing as character recognition module, outer round surface image processing module, storage mould Block and computer control module, end face image processing module, end view drawing are as character recognition module, outer round surface image processing module Concurrent working, and analysis result is transmitted to memory module and computer control module.
The little micro-bearing surface pitting defects on-line detecting system of one the most according to claim 6, it is characterised in that: described CCD select Basler digital camera A102f.
CN201610641082.7A 2016-08-05 2016-08-05 A kind of little micro-bearing surface pitting defects online test method and system thereof Pending CN106290392A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610641082.7A CN106290392A (en) 2016-08-05 2016-08-05 A kind of little micro-bearing surface pitting defects online test method and system thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610641082.7A CN106290392A (en) 2016-08-05 2016-08-05 A kind of little micro-bearing surface pitting defects online test method and system thereof

Publications (1)

Publication Number Publication Date
CN106290392A true CN106290392A (en) 2017-01-04

Family

ID=57665946

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610641082.7A Pending CN106290392A (en) 2016-08-05 2016-08-05 A kind of little micro-bearing surface pitting defects online test method and system thereof

Country Status (1)

Country Link
CN (1) CN106290392A (en)

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106908444A (en) * 2017-03-20 2017-06-30 辽宁科技大学 A kind of taper roll bearing end face identifying system and method based on image procossing
CN106910191A (en) * 2017-03-06 2017-06-30 中国科学院合肥物质科学研究院 A kind of bearing channel recognition methods based on depth image
CN107146704A (en) * 2017-07-10 2017-09-08 河南远中电力设备有限公司 A kind of transformer processing unit (plant) and production technology
CN107328793A (en) * 2017-06-30 2017-11-07 航天新长征大道科技有限公司 A kind of ornaments surface word print flaw detection method and device based on machine vision
CN107478654A (en) * 2017-09-21 2017-12-15 桐城市闲产网络服务有限公司 A kind of computer system motherboard intelligent checking system
CN108109137A (en) * 2017-12-13 2018-06-01 重庆越畅汽车科技有限公司 The Machine Vision Inspecting System and method of vehicle part
CN108965561A (en) * 2018-07-18 2018-12-07 惠州市盈泰欣精密金属科技有限公司 Plate detection device and its detection system in a kind of mobile phone
CN109001215A (en) * 2018-08-03 2018-12-14 深圳大图科创技术开发有限公司 A kind of power terminals defect detecting system based on machine vision
CN109029993A (en) * 2018-06-20 2018-12-18 中国计量大学 In conjunction with the bearing fault detection algorithm of genetic algorithm optimization parameter and machine vision
CN109975307A (en) * 2019-03-11 2019-07-05 中国科学院上海技术物理研究所 Bearing surface defect detection system and detection method based on statistics projection training
CN112444490A (en) * 2021-02-01 2021-03-05 深圳宜美智科技股份有限公司 Hole plugging defect detection method based on image detection and hole plugging defect detection equipment
CN116124799A (en) * 2023-04-04 2023-05-16 苏州赛肯智能科技有限公司 Cylinder defect detection mechanism

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH1183756A (en) * 1997-09-08 1999-03-26 Ishikawajima Harima Heavy Ind Co Ltd Method and device for automatically inspecting bearing
JP2002098643A (en) * 2000-09-22 2002-04-05 Koyo Seiko Co Ltd Appearance examination method and apparatus
CN200993170Y (en) * 2006-07-25 2007-12-19 河南科技大学 Bearing rolling body and rivet missing on-line detecting device
CN101221134A (en) * 2007-01-10 2008-07-16 宁波工程学院 Method and device for detecting tiny bearing surface defect by computer vision technology
CN103049750A (en) * 2013-01-11 2013-04-17 广州广电运通金融电子股份有限公司 Character recognition method
CN104360501A (en) * 2014-10-15 2015-02-18 西安交通大学 Visual detection method and device for defects of liquid crystal display screen
CN103226106B (en) * 2013-03-15 2016-02-24 江南大学 Based on the bearing dustproof cover quality monitoring method of machine vision

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH1183756A (en) * 1997-09-08 1999-03-26 Ishikawajima Harima Heavy Ind Co Ltd Method and device for automatically inspecting bearing
JP2002098643A (en) * 2000-09-22 2002-04-05 Koyo Seiko Co Ltd Appearance examination method and apparatus
CN200993170Y (en) * 2006-07-25 2007-12-19 河南科技大学 Bearing rolling body and rivet missing on-line detecting device
CN101221134A (en) * 2007-01-10 2008-07-16 宁波工程学院 Method and device for detecting tiny bearing surface defect by computer vision technology
CN103049750A (en) * 2013-01-11 2013-04-17 广州广电运通金融电子股份有限公司 Character recognition method
CN103226106B (en) * 2013-03-15 2016-02-24 江南大学 Based on the bearing dustproof cover quality monitoring method of machine vision
CN104360501A (en) * 2014-10-15 2015-02-18 西安交通大学 Visual detection method and device for defects of liquid crystal display screen

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
石炜等: "机器视觉在圆锥滚子轴承内圈外表面缺陷检测中的应用", <机械设计与制造> *

Cited By (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106910191B (en) * 2017-03-06 2019-10-18 中国科学院合肥物质科学研究院 A kind of bearing channel recognition methods based on depth image
CN106910191A (en) * 2017-03-06 2017-06-30 中国科学院合肥物质科学研究院 A kind of bearing channel recognition methods based on depth image
CN106908444A (en) * 2017-03-20 2017-06-30 辽宁科技大学 A kind of taper roll bearing end face identifying system and method based on image procossing
CN106908444B (en) * 2017-03-20 2023-10-10 辽宁科技大学 Tapered roller bearing end face recognition system and method based on image processing
CN107328793A (en) * 2017-06-30 2017-11-07 航天新长征大道科技有限公司 A kind of ornaments surface word print flaw detection method and device based on machine vision
CN107146704A (en) * 2017-07-10 2017-09-08 河南远中电力设备有限公司 A kind of transformer processing unit (plant) and production technology
CN107478654A (en) * 2017-09-21 2017-12-15 桐城市闲产网络服务有限公司 A kind of computer system motherboard intelligent checking system
CN108109137A (en) * 2017-12-13 2018-06-01 重庆越畅汽车科技有限公司 The Machine Vision Inspecting System and method of vehicle part
CN109029993A (en) * 2018-06-20 2018-12-18 中国计量大学 In conjunction with the bearing fault detection algorithm of genetic algorithm optimization parameter and machine vision
CN108965561A (en) * 2018-07-18 2018-12-07 惠州市盈泰欣精密金属科技有限公司 Plate detection device and its detection system in a kind of mobile phone
CN109001215A (en) * 2018-08-03 2018-12-14 深圳大图科创技术开发有限公司 A kind of power terminals defect detecting system based on machine vision
CN109975307A (en) * 2019-03-11 2019-07-05 中国科学院上海技术物理研究所 Bearing surface defect detection system and detection method based on statistics projection training
CN112444490A (en) * 2021-02-01 2021-03-05 深圳宜美智科技股份有限公司 Hole plugging defect detection method based on image detection and hole plugging defect detection equipment
CN112444490B (en) * 2021-02-01 2021-04-16 深圳宜美智科技股份有限公司 Hole plugging defect detection method based on image detection and hole plugging defect detection equipment
CN116124799A (en) * 2023-04-04 2023-05-16 苏州赛肯智能科技有限公司 Cylinder defect detection mechanism

Similar Documents

Publication Publication Date Title
CN106290392A (en) A kind of little micro-bearing surface pitting defects online test method and system thereof
CN111179251B (en) Defect detection system and method based on twin neural network and by utilizing template comparison
CN108520274B (en) High-reflectivity surface defect detection method based on image processing and neural network classification
CN108074231B (en) Magnetic sheet surface defect detection method based on convolutional neural network
CN108562589B (en) Method for detecting surface defects of magnetic circuit material
CN111693534B (en) Surface defect detection method, model training method, device, equipment and medium
CN107389701A (en) A kind of PCB visual defects automatic checkout system and method based on image
CN105069790A (en) Rapid imaging detection method for gear appearance defect
CN107966454A (en) A kind of end plug defect detecting device and detection method based on FPGA
CN109087286A (en) A kind of detection method and application based on Computer Image Processing and pattern-recognition
CN108665458A (en) Transparent body surface defect is extracted and recognition methods
CN106501272B (en) Machine vision soldering tin positioning detection system
Deng et al. Defect detection of bearing surfaces based on machine vision technique
Battiato et al. Pollen13k: a large scale microscope pollen grain image dataset
KR101782363B1 (en) Vision inspection method based on learning data
CN113177924A (en) Industrial production line product flaw detection method
Fu et al. Medicine glass bottle defect detection based on machine vision
Dominguez-Nicolas et al. Indentation image analysis for Vickers hardness testing
CN109241948A (en) A kind of NC cutting tool visual identity method and device
CN116434206A (en) Cotton quality character detection method based on machine vision technology
Abdellah et al. Defect detection and identification in textile fabric by SVM method
CN111487192A (en) Machine vision surface defect detection device and method based on artificial intelligence
CN109685756A (en) Image feature automatic identifier, system and method
CN112396580B (en) Method for detecting defects of round part
Lin et al. Defect detection system for optical element surface based on machine vision

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20170104