CN109166119A - Fabric defect detection method, device, equipment and machine readable media - Google Patents
Fabric defect detection method, device, equipment and machine readable media Download PDFInfo
- Publication number
- CN109166119A CN109166119A CN201811034583.4A CN201811034583A CN109166119A CN 109166119 A CN109166119 A CN 109166119A CN 201811034583 A CN201811034583 A CN 201811034583A CN 109166119 A CN109166119 A CN 109166119A
- Authority
- CN
- China
- Prior art keywords
- image
- network model
- classifier
- fault
- sample
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0008—Industrial image inspection checking presence/absence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30124—Fabrics; Textile; Paper
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
Abstract
This application discloses fabric defect detection method, device, equipment and machine readable medias, this method goes out abnormal image using first nerves network class, it is gone wrong image using nervus opticus network class, is gone wrong the type of fault existing for the fabric surface that image is included using classifier classification.This method, device, equipment and machine readable media can quickly note abnormalities image, and can exclusive PCR finds real problem image well in abnormal image, exact classification finally is carried out to fault, to reduce false detection rate, improves accuracy in detection.
Description
Technical field
This application involves Fabric Detection technical fields, in particular to fabric defect detection method, device, equipment and machine can
Read medium.
Background technique
On the production line of the fabrics such as woven fabric, looped fabric, non-woven cloth, need whether to detect produced fabric
There are faults, for example, whether having spot, broken hole, fluffing etc. on fabric.
Current detection method mainly is found to knit before perching equipment by testing staff station in such a way that naked eyes detect
Object fault is simultaneously marked fault or records.In the case where the yield of fabric is very big, being detected by testing staff will be very
Take manpower, moreover, testing staff is easy fatigue after a period of time that works, to there is a possibility that erroneous detection occurs.Therefore,
The overall fault detection efficiency detected by testing staff is not high and accuracy in detection is not sufficiently stable.
In the related technology, perching is carried out using computer mainly to realize by machine vision and fault classification method,
That is, obtaining cloth image to be detected by photography technology utilizes CNN by presetting multiple classifications
At least one convolutional layer and at least one that (Convolutional Neural Networks, convolutional neural networks) model includes
A pond layer determines the feature of image, and the full articulamentum for including using CNN model determines the image point according to the feature of the image
Do not belong to the probability of each classification in multiple classification, and the classification of maximum probability be determined as classification described in the image,
To obtain the classification of fault described in the cloth image.But due to the presence of fabric surface stamp, pass through single neural network
Model easily causes erroneous detection, causes the accuracy rate of detection very low.
Summary of the invention
In view of problem above, the embodiment of the present invention provides a kind of fabric defect detection method, device and equipment, can solve
The technical issues of certainly above-mentioned background technology part is mentioned.
The fabric defect detection method of embodiment according to the invention, comprising: when to detect that specified image included knits
When object surface whether there is fault, classified using the first nerves network model trained to the specified image, wherein
The first nerves network model is to belong to fabric surface included in it to there may be the different of fault for detection image
Normal image, which still falls within fabric surface included in it, can not have the normal picture of fault;If the first nerves net
The specified image classification is the abnormal image by network model, then utilizes the nervus opticus net with memory capability trained
Network model classifies to the specified image, wherein the nervus opticus network model is to belong to wherein for detection image
The fabric surface for being included has that fault image still falls within the normal picture;If the nervus opticus network mould
The specified image classification is described problem image by type, then is classified using the classifier trained to the specified image
And determine the fault type for the fabric surface that the specified image is included, wherein the classifier is wrapped for detection image
The type of fault existing for the fabric surface contained.
The fabric defects detection device of embodiment according to the invention, comprising: the first categorization module refers to for working as to detect
When determining fabric surface that image is included with the presence or absence of fault, the first nerves network model that utilization has been trained is to the specified figure
As classifying, wherein the first nerves network model is to belong to fabric surface included in it to have for detection image
The normal picture of fault can not be had by still falling within fabric surface included in it there may be the abnormal image of fault;The
Two categorization modules, if being used for the first nerves network model for the specified image classification is the abnormal image, benefit
Classified with the nervus opticus network model with memory capability trained to the specified image, wherein described second
Neural network model is to belong to fabric surface included in it to there are problems that fault image is still fallen within for detection image
The normal picture;Third categorization module, if being used for the nervus opticus network model for the specified image classification is institute
Problem image is stated, then is classified using the classifier trained to the specified image, wherein the classifier is for detecting
The type of fault existing for the fabric surface that image is included.
The fabric defects detection equipment of embodiment according to the invention, including processor;And memory, it is stored thereon with
Executable instruction, wherein the executable instruction makes the processor execute method above-mentioned upon being performed.
The machine readable media of embodiment according to the invention, is stored thereon with executable instruction, wherein described executable
Instruction makes machine execute method above-mentioned upon being performed.
It can be seen from the above that the scheme of the embodiment of the present invention goes out Abnormal Map using first nerves network class
Picture is gone wrong image using nervus opticus network class, is deposited using the classifier classification fabric surface that image is included that goes wrong
Fault type, therefore, compared with prior art, the scheme of the embodiment of the present invention can quickly note abnormalities image,
And can exclusive PCR finds real problem image well in abnormal image, finally to fault carry out exact classification, thus
False detection rate is reduced, accuracy in detection is improved.
Detailed description of the invention
Fig. 1 is the flow chart of the method for model training of one embodiment according to the invention;
Fig. 2 is the overview flow chart of the fabric defect detection method of one embodiment according to the invention;
Fig. 3 is the flow chart of the fabric defect detection method of one embodiment according to the invention;
Fig. 4 is the schematic diagram of the fabric defects detection device of one embodiment according to the invention;
Fig. 5 is the schematic diagram of the fabric defects detection equipment of one embodiment according to the invention.
Specific embodiment
Theme described herein is discussed referring now to example embodiment.It should be understood that discussing these embodiments only
It is in order to enable those skilled in the art can better understand that being not to claim to realize theme described herein
Protection scope, applicability or the exemplary limitation illustrated in book.It can be in the protection scope for not departing from present disclosure
In the case of, the function and arrangement of the element discussed are changed.Each example can according to need, omit, substitute or
Add various processes or component.For example, described method can be executed according to described order in a different order, with
And each step can be added, omits or combine.In addition, feature described in relatively some examples is in other examples
It can be combined.
As used in this article, term " includes " and its modification indicate open term, are meant that " including but not limited to ".
Term "based" indicates " being based at least partially on ".Term " one embodiment " and " embodiment " expression " at least one implementation
Example ".Term " another embodiment " expression " at least one other embodiment ".Term " first ", " second " etc. may refer to not
Same or identical object.Here may include other definition, either specific or implicit.Unless bright in context
It really indicates, otherwise the definition of a term is consistent throughout the specification.
The scheme that the embodiment of the present invention provides goes out abnormal image using first nerves network class, utilizes nervus opticus net
Network classification is gone wrong image, is gone wrong the type of fault existing for the fabric surface that image is included using classifier classification, because
This, compared with prior art, the scheme of the embodiment of the present invention can quickly note abnormalities image, and can be in abnormal image very
Exclusive PCR finds real problem image well, finally carries out exact classification to fault, to reduce false detection rate, improves detection
Accuracy.
The scheme of the embodiment of the present invention includes model training stage and actually detected stage.
Fig. 1 shows the flow chart of the method for model training of one embodiment according to the invention.It is shown in FIG. 1
Method 100 correspond to model training stage, using training data training obtain convolutional neural networks (CNN:
Convolutional Neural Network) model, the Recognition with Recurrent Neural Network (RNN:Recurrent with memory capability
Neural Network) model and SVM (Support Vector Machine, support vector machines) classifier.Wherein, CNN mould
Type belongs to normal picture or abnormal image for detecting the image comprising fabric inputted, and RNN model is for detecting CNN
The abnormal image that model inspection goes out belongs to normal picture or problem image, and SVM classifier is used to press fault to problem image
Type is classified.Wherein, normal picture refers to that the image of fault is not present in fabric surface included in it, and abnormal image is
Refer to that CNN category of model goes out its included in fabric surface there may be the image of fault, problem image refers to wherein institute
There are the images of fault for the fabric surface for including.It will be apparent from the above that it is also likely to be problem figure that abnormal image, which may be normal picture,
Picture.Method 100 shown in FIG. 1 can by computer or other suitably there is the electronic equipment of computing capability to realize.
As shown in Figure 1, receiving the image of multiple original shootings in box 102.Wherein, the image of multiple original shooting
Including multiple normal pictures and multiple problem images.
In box 104, image labeling (Image Annotation) processing is executed to the image of multiple original shooting, with
Obtain first sample image set SP1.Wherein, each of first sample image set SP1 sample image is to multiple original
One of image of the image of shooting executes what image labeling was handled.Image labeling processing is known technology, herein
Omit descriptions thereof.
In box 106, gray processing processing is executed to first sample image set SP1, it will be in first sample image set SP1
Each sample image is converted to gray level image.
In box 108, some or all sample images are chosen from the first sample image set SP1 that gray processing is handled and are made
For drawing of seeds picture.It such as, but not limited to, include more problem images in selected drawing of seeds picture, because of usual situation
Under, first sample image set SP1 includes less problem.
In box 110, one or many angularly rotations, mirror image are executed to each drawing of seeds picture and/or other are suitable
Operation, with from obtaining one or more images derived from each drawing of seeds picture.Wherein, the first sample of gray processing processing
Sample image in image set SP1 and the second sample graph image set is together to form from the image obtained derived from each drawing of seeds picture
SP2。
Under normal conditions, problem sample image is fewer than normal sample image, such as problem sample image and normal sample
The ratio of image may be 1: 10~1: 20, thus problem sample image and normal sample figure in first sample image set SP1
The quantity of picture be it is unbalanced, and unbalanced sample will lead to when carrying out neural metwork training last training result occur it is different
Normal deviation.Therefore, in 108 selected seed image of box, normal sample image of the sample image than selection the problem of selection
It is more, so as to after the angularly operation of rotation, mirror image etc. of process box 110 in obtained the second sample graph image set SP2,
The quantity of problem sample image and normal sample image is balance, abnormal variation occurs to avoid training result.In addition, passing through
The operation of box 106 and 108 can increase the quantity of training sample (for example, can be by 2500 sample images after treatment
Obtain the sample image more than 50000 or even 100000), and with the increase of training samples, the mind that finally training obtains
There is higher accuracy in detection through network model and classifier.
Box 106-110 constitutes the image preprocessing process (Image Preprocessing) of method 100.
In box 112, the property parameters of each gray level image in the second sample graph image set SP2 are obtained, wherein the attribute
Parameter includes but is not limited to the length of image, width etc..
In box 114, from each rule chosen in the second sample graph image set SP2 in its property parameters the first rule set of satisfaction
Multiple images then, as training the third sample graph image set SP3 of CNN model.Wherein, first rule set is for defining
The condition that sample image suitable for training CNN model needs to meet.For example, the first rule set is defined suitable for CNN model
Length limitation, width limitation that sample image needs to meet etc..Third sample graph image set SP3 includes multiple normal pictures and multiple
Problem image.
In box 116, each rule that its property parameters meets Second Rule concentration are chosen from the second sample graph image set SP2
Multiple images then, as training the 4th sample graph image set SP4 of RNN model.Wherein, the Second Rule collection is for defining
The condition that sample image suitable for training RNN model needs to meet.For example, the definition of Second Rule collection is suitable for training RNN mould
Length limitation, width limitation that the sample image of type needs to meet etc..4th sample graph image set SP4 include multiple normal pictures and
Multiple problem images.
In box 118, its property parameters is chosen from the second sample graph image set SP2 and meets each rule in third rule set
Multiple images then, as training the 5th sample graph image set SP5 of SVM classifier.Wherein, the third rule set is for fixed
Justice is suitable for the condition that the sample image of training SVM classifier needs to meet.For example, the definition of third rule set is suitable for training
Length limitation, width limitation that the sample image of SVM classifier needs to meet etc..5th sample graph image set SP5 includes multiple more
The problem of kind fault is classified image.Fault classification is such as, but not limited to spot, yarn defect, float, printing and dyeing fault, side defect, fold, latitude
Tiltedly, broken hole, hook silk, sanding unevenness, blur, fluffing, scratch, roll line, stop Mark.Wherein, spot includes greasy dirt, rust spot, color
Point, spot, mildew, auxiliary agent spot;Yarn defect includes dead cotton, slubbing, flyings, thick young yarn, soiled yarn, the dry unevenness of item;Float include broken yarn,
Knot, dropped stitch, rotten needle, try to stop people from fighting each other it is elastic, disconnected try to stop people from fighting each other, spacing is unstable, cloth cover plays snake, filling is shown up, color fibre, needle path, mistake yarn, yarn
Trace, try to stop people from fighting each other it is show-through, try to stop people from fighting each other and show up;Printing and dyeing fault include bite, stamp displacement, stamp staining, stamp cross that bottom, stamp is bad, contaminates
Flower two tone colour, loses colour, difference;When defect includes pin hole, double needle hole, crimping, rotten side, narrow envelope, wealthy envelope;Fold includes intermediate folder
Trace, cloth cover corrugation, folding line;Skew of weft includes twill, arch.
In box 120, use the image of third sample graph image set SP3 as training data, training obtains CNN model.
In box 122, use the image of the 4th sample graph image set SP4 as training data, training obtains RNN model.
Due to there is memory unit (memory) in RNN model, certain information can be carried out to the state recycled before
Storage, so the partial information for the object predicted before may be by, this is conducive to presence when predicting lower an object every time
The stamp (stamp sequence) of certain rule is identified, interference of the stamp to defects identification is excluded.
In box 124, use the image of the 5th sample graph image set SP5 as training data, training obtains SVM classifier.
Fig. 2 shows the overview flow charts of the fabric defect detection method of one embodiment according to the invention.Shown in Fig. 2
Method 200 correspond to the actually detected stage, use and utilize obtained CNN model, RNN model and SVM points of the training of method 100
Class device, which carrys out the fabric surface that detection image is included, whether there is fault.Method 200 shown in Fig. 2 can for example by computer or
Suitably there is the electronic equipment of computing capability to realize for other.
As shown in Fig. 2, in box 202, when want detection image T belong to normal picture or problem image and if it is
When problem image is image the problem of belonging to which class fault, pretreatment is executed to image T, such as, but not limited to, image T is turned
It is changed to gray level image etc..
In box 204, classified using the CNN model trained to pretreated image T.
In box 206, if image T is classified as normal picture by CNN model, it is determined that the fabric table that image T is included
Fault is not present in face, and then process terminates.
In box 208, if image T is classified as abnormal image by CNN model, using the RNN model trained to warp
Pretreated image T classifies.
In one aspect, when CNN model classifies to image T, the feature vector of image T can be obtained, when image T quilt
When CNN category of model is abnormal image, the feature vector for the image T that CNN model obtains can be inputted in RNN model by RNN mould
Type classifies to the feature vector of image T.
In box 210, if image T is classified as normal picture by RNN model, it is determined that the fabric table that image T is included
Fault is not present in face, and then process terminates.
In box 212, if image T is classified as problem image by RNN model, the SVM classifier pair trained is utilized
Pretreated image T classifies.
It in one aspect, can be by the image T's that CNN model obtains when image T is problem image by RNN category of model
Classified in feature vector input SVM classifier by feature vector of the SVM classifier to image T.
In box 214, determines the fault type for the fabric surface that image T is included and export the information of fault type, so
Process terminates afterwards.
Other modifications
It will be understood by those skilled in the art that although in the above embodiments, method 200 includes executing pre- place to image T
The box 202 of reason, however, the present invention is not limited thereto.In other embodiments of the invention, such as, but not limited to, exist
In the case where being suitable for being classified using model or classifier under the original state of image T, method 200 can not also be wrapped
It includes and pretreated box 202 is executed to image T.
It will be understood by those skilled in the art that although in the above embodiments, method 100 includes holding to the received image of institute
The box 104 of row image labeling processing, however, the present invention is not limited thereto.In other embodiments of the invention, such as but
It is not limited to, in the case where the 102 received image of institute of box has executed image labeling processing, method 100 can not also include
To received image execute the box 104 of image labeling processing.
It will be understood by those skilled in the art that although in the above embodiments, method 100 includes box 106-108 to spread out
Raw more sample images, however, the present invention is not limited thereto.In other embodiments of the invention, such as but do not limit to
In in the case where the quantity of existing sample image is enough, method 100 can not also include box 106-108.
It will be understood by those skilled in the art that although in the above embodiments, method 100 includes box 108-110 with flat
The quantity of weighing apparatus problem sample image and normal sample image and more sample images are obtained, however, the present invention does not limit to
In this.In other embodiments of the invention, such as, but not limited to, the problem sample graph in the 102 received image of institute of box
The quantity of picture and normal sample image be balance and quantity it is enough in the case where, method 100 can not also the side of including
Frame 108-110.
Although method 100 includes box 106 with by sample it will be understood by those skilled in the art that in the above embodiments
Image is converted into gray level image, however, the present invention is not limited thereto.In other embodiments of the invention, such as but not office
It is limited to, in the case where the 102 received image of institute of box has been gray level image, method 100 can not also include box 106.
It will be understood by those skilled in the art that although in the above embodiments, method 100 includes box 112-118 to select
The sample image for being suitable for training CNN model, RNN model and classifier is taken, however, the present invention is not limited thereto.In this hair
In bright other embodiments, method 100 can not also include box 112-118.
Although it will be understood by those skilled in the art that in the above embodiments, being trained to for detection image be to belong to just
The neural network model of normal image or abnormal image is CNN model, however, the present invention is not limited thereto.Of the invention
In other embodiments, the neural network model for being trained to belong to for detection image normal picture or abnormal image can be with
It is BP neural network model or other kinds of neural network model.
Although it will be understood by those skilled in the art that in the above embodiments, being trained to for detecting abnormal image be to belong to
In the neural network model of normal picture or problem image be RNN model, however, the present invention is not limited thereto.In this hair
In bright other embodiments, it is trained to belong to the neural network mould of normal picture or problem image for detecting abnormal image
Type can also be LSTM (Long Short-Term Memory, shot and long term memory) model or other minds with memory capability
Through network model.
Although it will be understood by those skilled in the art that in the above embodiments, being trained to be to belong to for test problems image
The classifier of image is SVM classifier in which class fault the problem of, however, the present invention is not limited thereto.In its of the invention
During he implements, it is trained to be that the classifier of image the problem of belonging to which class fault can also be Bayes for test problems image
Classifier, Nearest Neighbor Classifier, linear classifier or other kinds of classifier.
Fig. 3 shows the flow chart of the fabric defect detection method of one embodiment according to the invention.Side shown in Fig. 3
Method 300 can for example by computer or other suitably there is the electronic equipment of computing capability to realize.
As shown in figure 3, method 300 may include, in box 302, when the fabric surface that detect specified image and included
When with the presence or absence of fault, classified using the first nerves network model trained to the specified image, wherein described
One neural network model is the abnormal image for belonging to fabric surface included in it and there may be fault for detection image
The normal picture of fault can not be had by still falling within fabric surface included in it.
Method 300 can also include: in box 304, if the first nerves network model divides the specified image
Class is the abnormal image, then using the nervus opticus network model with memory capability trained to the specified image into
Row classification, wherein the nervus opticus network model is to belong to fabric surface included in it there are defects for detection image
The problem of putting image still falls within the normal picture.
Method 300 can also include: in box 306, if the nervus opticus network model divides the specified image
Class is described problem image, then classify to the specified image using the classifier trained and determine the specified image
The fault type for the fabric surface for being included, wherein existing for the fabric surface that the classifier is included for detection image
The type of fault.
In one aspect, the first nerves network model is obtained using the training of first group of sample image, described the
Two neural network models be obtained using the training of second group of sample image, and, the classifier is to utilize third group sample
Image training obtains;Wherein, first group of sample image is based on the rule in the first rule set from comprising multiple described
It is chosen in multiple sample images of normal picture and multiple described problem images, second group of sample image is based on second
What the rule in rule set was chosen from the multiple sample image, and, the third group sample image is advised based on third
What the rule then concentrated was chosen from multiple sample images comprising multiple described problem images.
On the other hand, the first nerves network model is convolutional neural networks model and the nervus opticus
Network model is Recognition with Recurrent Neural Network model.
In yet another aspect, the classifier includes Bayes classifier, Nearest Neighbor Classifier, linear classifier, svm classifier
The one of which of device.
Fig. 4 shows the schematic diagram of the fabric defects detection device of one embodiment according to the invention.Dress shown in Fig. 4
Setting 400 can use the mode of software, hardware or software and hardware combining to realize.Device 400 for example may be mounted at computer or
Other suitably have in the electronic equipment of computing capability.
As shown in figure 4, device 400 may include the first categorization module 402, the second categorization module 404 and third classification mould
Block 406.First categorization module 402 is used to utilize when to detect fabric surface that specified image is included with the presence or absence of fault
The first nerves network model trained classifies to the specified image, wherein the first nerves network model is used for
Detection image is to belong to fabric surface included in it and there may be the abnormal image of fault to still fall within included in it
Fabric surface can not have the normal picture of fault.If the second categorization module 404 is used for the first nerves network mould
The specified image classification is the abnormal image by type, then utilizes the nervus opticus network mould with memory capability trained
Type classifies to the specified image, wherein the nervus opticus network model is to belong to wherein to be wrapped for detection image
The fabric surface contained has that fault image still falls within the normal picture.If third categorization module 406 is used for institute
It is described problem image that nervus opticus network model, which is stated, by the specified image classification, then using the classifier trained to described
Specified image carries out the fault type for classifying and determining the fabric surface that the specified image is included, wherein the classifier
The type of fault existing for the fabric surface for being included for detection image.
In one aspect, the first nerves network model is obtained using the training of first group of sample image, described the
Two neural network models be obtained using the training of second group of sample image, and, the classifier is to utilize third group sample
Image training obtains;Wherein, first group of sample image is based on the rule in the first rule set from comprising multiple described
It is chosen in multiple sample images of normal picture and multiple abnormal images, second group of sample image is based on second
What the rule in rule set was chosen from the multiple sample image, and, the third group sample image is advised based on third
What the rule then concentrated was chosen from multiple sample images comprising multiple described problem images.
On the other hand, the first nerves network model is convolutional neural networks model and the nervus opticus
Network model is Recognition with Recurrent Neural Network model.
In yet another aspect, the classifier includes Bayes classifier, Nearest Neighbor Classifier, linear classifier, svm classifier
The one of which of device.
Fig. 5 shows the schematic diagram of the fabric defects detection equipment of one embodiment according to the invention.As shown in figure 5,
Equipment 500 may include processor 502 and memory 504, wherein be stored with executable instruction on memory 504, wherein institute
State executable instruction makes processor 502 execute method 100, method shown in Fig. 2 200 or Fig. 3 shown in FIG. 1 upon being performed
Shown in method 300.
In one aspect, fabric defects detection equipment can also include one or more of the following components: power supply module, more matchmakers
Body component, audio component, input/output (I/O) interface, sensor module and communication component.Wherein, power supply module is to set
Standby 500 various assemblies provide power supply;Multimedia component includes one output interface of offer between equipment 500 and user
Screen;Audio component is configured as output and/or input audio signal;I/O interface be processor 502 and peripheral interface module it
Between interface is provided;Sensor module includes one or more sensors;Communication component is configured to facilitate equipment 500 and other set
The communication of wired or wireless way between standby.
The embodiment of the present invention also provides a kind of machine readable media, is stored thereon with executable instruction, wherein it is described can
It executes instruction and machine is made to execute method 100, method shown in Fig. 2 200 or side shown in Fig. 3 shown in FIG. 1 upon being performed
Method 300.
The specific embodiment illustrated above in conjunction with attached drawing describes exemplary embodiment, it is not intended that may be implemented
Or fall into all embodiments of the protection scope of claims." exemplary " meaning of the term used in entire this specification
Taste " be used as example, example or illustration ", be not meant to than other embodiments " preferably " or " there is advantage ".For offer pair
The purpose of the understanding of described technology, specific embodiment include detail.However, it is possible in these no details
In the case of implement these technologies.In some instances, public in order to avoid the concept to described embodiment causes indigestion
The construction and device known is shown in block diagram form.
The foregoing description of present disclosure is provided so that any those of ordinary skill in this field can be realized or make
Use present disclosure.To those skilled in the art, the various modifications carried out to present disclosure are apparent
, also, can also answer generic principles defined herein in the case where not departing from the protection scope of present disclosure
For other modifications.Therefore, present disclosure is not limited to examples described herein and design, but disclosed herein with meeting
Principle and novel features widest scope it is consistent.
Claims (10)
1. fabric defect detection method, comprising:
When to detect fabric surface that specified image is included with the presence or absence of fault, first nerves network mould that utilization has been trained
Type classifies to the specified image, wherein the first nerves network model is to belong to wherein to be wrapped for detection image
The abnormal image that the fabric surface contained there may be fault, which still falls within fabric surface included in it, can not have defect
The normal picture of point;
If the specified image classification is the abnormal image by the first nerves network model, the tool trained is utilized
There is the nervus opticus network model of memory capability to classify the specified image, wherein the nervus opticus network model
It is to belong to fabric surface included in it to there are problems that fault image still falls within the normal picture for detection image;
If the specified image classification is described problem image by the nervus opticus network model, point trained is utilized
Class device carries out the fault type for classifying and determining the fabric surface that the specified image is included to the specified image, wherein
The type of fault existing for the fabric surface that the classifier is included for detection image.
2. according to the method described in claim 1, wherein,
The first nerves network model is obtained using first group of sample image training, and the nervus opticus network model is
It is obtained using second group of sample image training, and, the classifier is obtained using the training of third group sample image;
Wherein, first group of sample image is based on the rule in the first rule set from comprising multiple normal pictures and more
It is chosen in multiple sample images of a abnormal image, second group of sample image is the rule concentrated based on Second Rule
Then chosen from the multiple sample image, and, the third group sample image is based on the rule in third rule set
It is chosen from multiple sample images comprising multiple described problem images.
3. method according to claim 1 or 2, wherein
The first nerves network model is convolutional neural networks model, and
The nervus opticus network model is Recognition with Recurrent Neural Network model.
4. method according to claim 1 or 2, wherein
The classifier include Bayes classifier, Nearest Neighbor Classifier, linear classifier, SVM classifier one of which.
5. fabric defects detection device, comprising:
First categorization module, for when to detect fabric surface that specified image is included with the presence or absence of fault, using having instructed
Experienced first nerves network model classifies to the specified image, wherein the first nerves network model is for detecting
Image is to belong to fabric surface included in it and there may be the abnormal image of fault to still fall within and knit included in it
Object surface can not have the normal picture of fault;
Second categorization module, if being used for the first nerves network model for the specified image classification is the Abnormal Map
Picture then classifies to the specified image using the nervus opticus network model with memory capability trained, wherein institute
It is to belong to fabric surface included in it to there are problems that fault image also that nervus opticus network model, which is stated, for detection image
It is to belong to the normal picture;
Third categorization module, if being used for the nervus opticus network model for the specified image classification is described problem figure
Picture then carries out the specified image using the classifier trained to classify and determine the fabric table that the specified image is included
The fault type in face, wherein the type of fault existing for the fabric surface that the classifier is included for detection image.
6. device according to claim 5, wherein
The first nerves network model is obtained using first group of sample image training, and the nervus opticus network model is
It is obtained using second group of sample image training, and, the classifier is obtained using the training of third group sample image;
Wherein, first group of sample image is based on the rule in the first rule set from comprising multiple normal pictures and more
It is chosen in multiple sample images of a abnormal image, second group of sample image is the rule concentrated based on Second Rule
Then chosen from the multiple sample image, and, the third group sample image is based on the rule in third rule set
It is chosen from multiple sample images comprising multiple described problem images.
7. device according to claim 5 or 6, wherein
The first nerves network model is convolutional neural networks model, and
The nervus opticus network model is Recognition with Recurrent Neural Network model.
8. device according to claim 5 or 6, wherein
The classifier include Bayes classifier, Nearest Neighbor Classifier, linear classifier, SVM classifier one of which.
9. fabric defects detection equipment, comprising:
Processor;And
Memory is stored thereon with executable instruction, wherein the executable instruction holds the processor
Any one of method of row claim 1-4.
10. machine readable media is stored thereon with executable instruction, wherein the executable instruction makes machine upon being performed
Any one of method of device perform claim requirement 1-4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811034583.4A CN109166119A (en) | 2018-09-05 | 2018-09-05 | Fabric defect detection method, device, equipment and machine readable media |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811034583.4A CN109166119A (en) | 2018-09-05 | 2018-09-05 | Fabric defect detection method, device, equipment and machine readable media |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109166119A true CN109166119A (en) | 2019-01-08 |
Family
ID=64894248
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811034583.4A Pending CN109166119A (en) | 2018-09-05 | 2018-09-05 | Fabric defect detection method, device, equipment and machine readable media |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109166119A (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100009165A1 (en) * | 2008-07-10 | 2010-01-14 | Zyvex Performance Materials, Llc | Multifunctional Nanomaterial-Containing Composites and Methods for the Production Thereof |
US20150381362A1 (en) * | 2014-06-30 | 2015-12-31 | Nicira, Inc. | Encryption System in a Virtualized Environment |
CN106295691A (en) * | 2016-08-04 | 2017-01-04 | 江南大学 | The feature selection of single classification SVM and parameter synchronization optimization method |
CN107369155A (en) * | 2017-07-24 | 2017-11-21 | 广东工业大学 | A kind of cloth surface defect detection method and its system based on machine vision |
CN108038843A (en) * | 2017-11-29 | 2018-05-15 | 英特尔产品(成都)有限公司 | A kind of method, apparatus and equipment for defects detection |
-
2018
- 2018-09-05 CN CN201811034583.4A patent/CN109166119A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100009165A1 (en) * | 2008-07-10 | 2010-01-14 | Zyvex Performance Materials, Llc | Multifunctional Nanomaterial-Containing Composites and Methods for the Production Thereof |
US20150381362A1 (en) * | 2014-06-30 | 2015-12-31 | Nicira, Inc. | Encryption System in a Virtualized Environment |
CN106295691A (en) * | 2016-08-04 | 2017-01-04 | 江南大学 | The feature selection of single classification SVM and parameter synchronization optimization method |
CN107369155A (en) * | 2017-07-24 | 2017-11-21 | 广东工业大学 | A kind of cloth surface defect detection method and its system based on machine vision |
CN108038843A (en) * | 2017-11-29 | 2018-05-15 | 英特尔产品(成都)有限公司 | A kind of method, apparatus and equipment for defects detection |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20220414856A1 (en) | A fabric defect detection method based on multi-modal deep learning | |
CN109035248A (en) | Defect detection method, apparatus, terminal device, server and storage medium | |
CN109187579A (en) | Fabric defect detection method and device, computer equipment and computer-readable medium | |
CN109023839B (en) | Control method for washing machine, device, washing machine and storage medium | |
CN107893309A (en) | Washing methods and device, clothes washing method and device | |
CN108038843A (en) | A kind of method, apparatus and equipment for defects detection | |
KR20170138559A (en) | Washing machine control method and washing machine | |
CN107354662A (en) | Device for clothing processing and its control method, device and computer-readable recording medium | |
CN109509171A (en) | A kind of Fabric Defects Inspection detection method based on GMM and image pyramid | |
CN109145985A (en) | A kind of detection and classification method of Fabric Defects Inspection | |
CN109255390A (en) | Preprocess method and module, discriminator, the readable storage medium storing program for executing of training image | |
CN109242846A (en) | Method, apparatus and equipment for fabric surface defects detection | |
CN109325940A (en) | Textile detecting method and device, computer equipment and computer-readable medium | |
CN110378887A (en) | Screen defect inspection method, apparatus and system, computer equipment and medium | |
Luo et al. | A lightweight detector based on attention mechanism for fabric defect detection | |
CN113936001A (en) | Textile surface flaw detection method based on image processing technology | |
Anami et al. | Comparative analysis of SVM and ANN classifiers for defective and non-defective fabric images classification | |
CN109166119A (en) | Fabric defect detection method, device, equipment and machine readable media | |
Barman et al. | Development of a real-time home textile fabric defect inspection machine system for the textile industry | |
CN115311264B (en) | Fabric flaw line defect identification method for textile production | |
CN116934694A (en) | Cloth cover defect detection method and system | |
CN109146875A (en) | Method, apparatus and equipment for fabric surface defects detection | |
CN112488986A (en) | Cloth surface flaw identification method, device and system based on Yolo convolutional neural network | |
Shanthi et al. | Robust fabric defects inspection system using deep learning architecture | |
CN109543733A (en) | A kind of color-woven fabric Quick production method based on cloud platform |
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: 20190108 |
|
WD01 | Invention patent application deemed withdrawn after publication |