CN105205828B - Knitted fabric flaw detection method based on Optimal Gabor Filters - Google Patents
Knitted fabric flaw detection method based on Optimal Gabor Filters Download PDFInfo
- Publication number
- CN105205828B CN105205828B CN201510686033.0A CN201510686033A CN105205828B CN 105205828 B CN105205828 B CN 105205828B CN 201510686033 A CN201510686033 A CN 201510686033A CN 105205828 B CN105205828 B CN 105205828B
- Authority
- CN
- China
- Prior art keywords
- gabor
- particle
- image
- function
- knitted fabric
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- 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
-
- 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
- G06T2207/20024—Filtering details
-
- 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
Abstract
The present invention relates to a kind of knitted fabric flaw detection method based on Optimal Gabor Filters, is characterized in, comprising the following steps: step 1, construction Gabor filter extract Gabor filtering parameter;Step 2 carries out Gabor process of convolution to indefectible knitted fabric image, fitness function is constructed using Fisher criterion, utilization sub-line is that particle group optimizing (QPSO) algorithm carries out optimization processing to the Gabor filtering parameter that step 1 is extracted, and obtains the optimized parameter of Gabor filter;Step 3, the Gabor filter optimized parameter obtained by step 2 carry out Gabor process of convolution to knitted fabric image to be detected;Step 4, progress binary conversion treatment obtain the Defect Detection result of knitted fabric.Flaw detection method of the present invention can improve knitted fabric Defect Detection efficiency and Detection accuracy.
Description
Technical field
The present invention relates to a kind of knitted fabric flaw detection method based on Optimal Gabor Filters, belongs to image procossing
Technical field.
Background technique
Domestic warp knit industry has obtained quick development, while becoming the major production base of world's warp knit product,
External Countries warp knit industry also gradually emerges, such as India, some countries in Southeast Asia, Vietnam, and competition is promoted to increase.It faces
Competition, improving product quality and reducing production cost becomes the key of Business survival.Fabric defect is the master for influencing cloth quality
Want factor.According to investigations, due to the presence of fault, make fabric price reduction 45-65%, enterprise income is by very big loss.
Most of current textile enterprise still takes the mode manually estimated to detect fabric defects.The shortcomings that artificial detection, is: (1) vision
The limitation of precision and easy fatigue cause to detect unstable;(2) manually range estimation Detection accuracy is low, and the detection of experienced operator is accurate
Rate is difficult to more than 85%;(3) limited view of people can not detect very wide region simultaneously, and the energy of people is limited, often sends out
The case where raw missing inspection;(4) testing cost is relatively high, the expenditure such as training cost, wage including a large amount of workers.Therefore, it is based on machine
Use of the defect detection system of device vision in enterprise practical production process can greatly improve the quality of tricot machine cloth outputting, together
When, can reduce artificial, the production cost of enterprise be saved, to increase Competitive Products.
In general, after acquiring knitted fabric image, preprocessing process need to be carried out, then feature extraction is carried out and flaw is sentenced
Not.Pretreatment primarily to enhancing image contrast, compare flaw information and image background information obvious, in addition need to be into
Row image denoising removes the noise in image.Feature extraction is exactly analyzed and processed image, therefrom extracts suitable special
Levy vector.Flaw differentiates, exactly detected defect areas by some classifier.Wherein feature extraction is Defect Detection side
The core link of method, in general, the method for feature extraction mainly have statistic law, modelling and frequency domain method etc..Based on statistic law
Auto-correlation function, gray level co-occurrence matrixes, the methods of Mathematical Morphology method and fractal theory can be divided by extracting feature.It is mentioned based on modelling
Feature is taken to have Wold model and Markov random field model etc..Fourier transformation, small can be divided by extracting feature based on frequency domain method
Wave conversion and Gabor transformation etc..Gabor function is one group of similar measure by rotation and flexible processing, is one group of narrowband
Bandpass filter has preferable resolution capability in spatial domain and frequency domain, there is apparent direction selection and frequency selective characteristic,
It can be realized the best alignment by union of airspace and frequency domain, therefore Gabor transformation is suitble to the analysis of texture image.
Summary of the invention
The purpose of the present invention is overcoming the deficiencies in the prior art, provide a kind of based on Optimal Gabor Filters
Knitted fabric flaw detection method can be improved the accuracy rate and real-time of detection.
According to technical solution provided by the invention,
In a specific embodiment, a kind of knitted fabric flaw detection method based on Optimal Gabor Filters, it is special
Sign is, comprising the following steps:
Step 1, construction Gabor filter, extract Gabor filtering parameter;
Step 2 carries out Gabor process of convolution to indefectible knitted fabric image, constructs fitness using Fisher criterion
Function, utilization sub-line are that particle group optimizing (QPSO) algorithm carries out at optimization the Gabor filtering parameter that step 1 is extracted
Reason, obtains the optimized parameter of Gabor filter;
Step 3, the Gabor filter optimized parameter obtained by step 2 carry out knitted fabric image to be detected
Gabor process of convolution;
Step 4, progress binary conversion treatment obtain the Defect Detection result of knitted fabric.
Further, the step 1 follows the steps below to implement:
Step 1.1 establishes Gabor filter function, specifically implements in accordance with the following methods:
The two-dimentional Gaussian kernel function g (x, y) that Gabor filter function G (x, y) is tuned by the multiple SIN function of directionality
It modulates, in the spatial domain, Gaussian kernel function is related to three parameter δx, δy, θ, wherein δxFor Gaussian kernel function
Scale parameter in x-axis direction, δyFor the scale parameter on Gaussian kernel function y-axis direction, θ is Gaussian kernel function
Angle is rotated, (x ', y ') is the coordinate after (x, y) rotation θ angle;In frequency domain, Fourier transformation is related to frequency
Three parameters are F0, u0, v0, wherein F0It is the centre frequency of oval Gabor filter function, u0It is that Gabor is filtered in x-axis direction
The centre frequency of function, v0It is the centre frequency of Gabor filter function on the direction y;
Gabor filter function in two-dimensional space is indicated are as follows:
Wherein,
Gabor filter function in spatial domain is fourier transformed to obtain the Gabor filter function of frequency domain by step 1.2
Wherein,
Step 1.3, the Gabor filter function constructed from step 1.2It is middle extract 6 Gabor filtering parameters (α,
γ, u0, v0, F0, θ).
Further, coefficient F in the Gabor filter function G (x, y) that the step 1.1 constructsoWhen=0, then 2- is modulated to
DGabor filter function G (x, y);u0=0, v0When=0, then the Gabor filter function G (x, y) of ellipse is modulated to.
Further, the step 2 follows the steps below to implement:
Step 2.1, initialization population, including determine maximum number of iterations, search space, particle number, it is random just
The position of beginningization particle;
Step 2.2, in first time iteration, the initial position of each particle is current individual desired positions;To indefectible
Knitted fabric image carries out Gabor process of convolution, constructs fitness function using Fisher criterion, it is corresponding to calculate each particle
Functional value;The fitness function value of all particles is compared finds the particle with minimum fitness function value, the particle more afterwards
Position be global desired positions;
Step 2.3 is updated the position of each particle, finds out each particle using with step 2.2 same procedure
Fitness function value, more new individual desired positions and global desired positions;
Step 2.4, when reaching iteration termination condition, training terminate, global desired positions are Gabor to be determined
The optimal value of filtering parameter;Otherwise, the number of iterations adds 1, goes to step 2.3.
Further, the step 2.1 follows the steps below to implement: the number of iterations n=0 when setting initial, greatest iteration time
Number is max_n;Gabor filtering parameter has (alpha, gamma, u0, v0F0, θ), then search space is 6 dimensions;The number of particle is M, each grain
Son initial position beWherein i=1,2 ..., M/
Further, image R (x, y) of the image after Gabor convolution may be expressed as: in the step 2.2
Wherein, T (x, y) is indefectible knitted fabric image, and R (x, y) is the image after Gabor filter convolution, and * is
The convolution operation of image,It is the Fourier transformation of image T (x, y), IDFT is inverse discrete fourier transform;
The energy of image R (x, y) after the Gabor convolution indicates are as follows:
Wherein,
It is G respectivelye(x, y) and GoThe discrete Fourier transform of (x, y);
The objective function according to Fisher criterion construction indicates are as follows:
Wherein, Gabor filtering parameter Φ=(alpha, gamma, u0, v0, F0θ), it is X × Y that μ (Φ) and σ (Φ), which is size respectively,
Image passes through the average energy value and standard deviation after Gabor convolution;
There are 6 decision variables as a result, the nonlinear programming problem of 5 constraint conditions can be described as:
s.t.
0≤θ≤π;
In first time iteration, the initial position of each particle is current individual desired positions, i.e.,By
The objective function of Fisher criterion construction calculates the corresponding fitness function value of each particle;
The fitness function value of all particles is compared finds the particle with minimum fitness function value more afterwards, the particle
Position is global desired positions.
Further, in the step 2.3 particle location updating equation are as follows:
Taking "+" in formula or taking the probability of "-" is all 0.5, and wherein β is known as converging diverging coefficient,For section (0,1)
On uniform random number, the convergence process of particle i is with pointFor attractor, coordinate are as follows:
WhereinIt is equally distributed random number on a section (0,1);
Following formula is used in the step 2.3 when more new individual desired positions:
After the individual desired positions of each particle determine, according toUpdate global desired positions.
It further, is that S (x, y) carries out Gabor process of convolution to knitted fabric image to be detected in the step 3,
Image Q (x, y) after obtaining convolution:
Further, binary conversion treatment is carried out using following formula in the step 4:
Wherein B (x, y) is bianry image, is the final result of Defect Detection, if the value of B (x, y) is 1, mapping to be checked
As corresponding location of pixels has flaw;If the value of B (x, y) is 0, the corresponding location of pixels of image to be detected is indefectible;μ
It is the average energy value of image after convolution, σ is that energy scale is poor, and c is experimental constant, is obtained by experiment.
The invention has the following advantages:
(1) present invention can be modulated into 2-D-Gabor filter using the Gabor filter of any modulation, can also adjust
Oval Gabor filter is made, so that the Gabor constructed more effectively detects different types of flaw;
(2) present invention uses the single optimal of construction using QPSO algorithm training Gabor filter parameter, when detection
Gabor filter, can efficient, accurate detection knitted fabric flaw, be more advantageously used for industrial production;
(3) present invention constructs objective function, the Gabor filter of obtained Gabor filtering parameter construction by Fisher criterion
Wave device more agrees with flawless textile image texture, so that the Gabor filter constructed more effectively detects warp knit
Fabric defects.
Detailed description of the invention
Fig. 1 is the flow chart of knitted fabric flaw detection method of the present invention.
Specific embodiment
Below with reference to specific attached drawing, the invention will be further described.
Knitted fabric flaw detection method of the present invention based on Optimal Gabor Filters, as shown in Figure 1, specific packet
Include following steps:
The Gabor filter that step 1, construction can be modulated arbitrarily obtains the Gabor filtering ginseng it needs to be determined that optimal value
Number;
Step 1.1, the Gabor filter function for establishing any modulation are specifically implemented in accordance with the following methods:
Gabor filter function G (x, y) is established, is the two-dimentional Gaussian tuned by a kind of multiple SIN function of directionality
Made of kernel function g (x, y) modulation.The time-frequency combination of Gabor filter positions, multiple dimensioned, multidirectional characteristic, so that
Gabor filter function passes through dilation appropriate or rotation, the Gabor of the different directions different scale of available self similarity
Filter function.
Gabor filter function indicates in two-dimensional space are as follows:
The Gabor filter function G (x, y) of 2-D or ellipse, coefficient F in formula (1) can be arbitrarily modulated into according to formula (1)o
When=0, it is modulated to 2-D Gabor filter function G (x, y);u0=0, v0When=0, it is modulated to the Gabor filter function G of ellipse
(x, y);
Wherein,
In the spatial domain, Gaussian kernel function is related to three parameter δx, δy, θ, wherein δxFor Gaussian kernel function x
Scale parameter in axis direction, δyFor the scale parameter on Gaussian kernel function y-axis direction, θ is Gaussian kernel function
Angle is rotated, (x ', y ') is the coordinate after (x, y) rotation θ angle.In frequency domain, frequency involved by Fourier transformation
Three parameters be F0, u0, v0, wherein F0It is the centre frequency of oval Gabor filter function, u0It is that Gabor is filtered in x-axis direction
The centre frequency of wave function, v0It is the centre frequency of Gabor filter function on the direction y.
This any modulation Gabor filter is to be obtained by Gaussian kernel function multiplied by multiple SIN function, Ke Yigai
It is written as:
G (x, y)=Ge(x, y)+jGo(x, y) (5);
Wherein, Ge(x, y) is the real part of Gabor filter, Go(x, y) is the imaginary part of Gabor filter, can be respectively indicated
It is as follows:
Step 1.2, Gabor filter show powerful image characteristics extraction ability, but calculation amount is larger.For simplification
It calculates, meets requirement of real-time, Gabor filter function is fourier transformed to obtain the Gabor filtering letter of frequency domain in spatial domain
Number
Wherein,
There are 6 Gabor filtering parameters (alpha, gamma, u in step 1.3, formula (6)0, v0, F0, θ) it needs to be determined that optimal value.
Step 2 carries out Gabor process of convolution to indefectible knitted fabric image, constructs fitness using Fisher criterion
Function, utilization sub-line are that particle group optimizing (QPSO) algorithm carries out optimization processing to the Gabor filtering parameter extracted,
Obtain the optimized parameter of Gabor filter;
Step 2.1, initialization population, including determine maximum number of iterations, search space, particle number, it is random just
The position (an as class value of Gabor filtering parameter) of beginningization particle.
If the number of iterations n=0, maximum number of iterations max_n when initial.It needs to be determined that the parameter of optimal value have (alpha, gamma,
u0, v0, F0, θ), then search space is 6 dimensions.The number of particle is M, and the initial position of each particle isWherein i=1,2 ..., M.
Step 2.2, in first time iteration, the initial position of each particle is current individual desired positions.To indefectible
Knitted fabric image carries out Gabor process of convolution, constructs fitness function using Fisher criterion, it is corresponding to calculate each particle
Functional value.The fitness function value of all particles is compared finds a particle with minimum fitness function value more afterwards, should
The position of particle is global desired positions.
To extract knitted fabric feature, fitness function is constructed, indefectible knitted fabric image is carried out at Gabor convolution
Reason.In the spatial domain, convolution will be carried out to real and imaginary parts respectively by calculating convolution, then be merged again.By formula (6), space
Gabor filter function is fourier transformed to obtain the Gabor filter function of frequency domain in domainIn Fu of convolution of functions
Leaf transformation is the product of function Fourier transformation, i.e., the product in the convolution respective frequencies domain in spatial domain.Image passes through Gabor volumes
Image R (x, y) after product may be expressed as:
Wherein, T (x, y) is indefectible knitted fabric image, and R (x, y) is the image after Gabor filter convolution, and * is
The convolution operation of image,It is the Fourier transformation of image T (x, y), IDFT is inverse discrete fourier transform.
In general, the image after the Gabor convolution defined by formula (7) is the image of a plural form, energy
It can indicate are as follows:
Wherein,
It is G respectivelye(x, y) and GoThe discrete Fourier transform of (x, y).
Fitness function can be constructed according to Fisher criterion, constructed optimization according to the cost function of Fisher criterion and asked
The objective function of topic:
Wherein, Gabor filtering parameter Φ=(alpha, gamma, u0, v0, F0, θ), μ (Φ) and σ (Φ) are that size is x × Y respectively
Image passes through the average energy value and standard deviation after Gabor convolution.
There are 6 decision variables as a result, the nonlinear programming problem of 5 constraint conditions can be described as:
s.t.
0≤θ≤π (12e);
In first time iteration, the initial position of each particle is current individual desired positions, i.e.,By formula
(9) the corresponding fitness function value of each particle is calculated.
The fitness function value of all particles is compared finds a particle with minimum fitness function value, the grain more afterwards
The position of son is global desired positions.If the global desired positions of entire populationWherein,
Step 2.3 is updated the position of each particle, finds out each particle using with step 2.2 same procedure
Fitness function value, more new individual desired positions and global desired positions.
By QPSO algorithm, the location updating equation of particle are as follows:
Taking "+" in formula or taking the probability of "-" is all 0.5.Wherein β is known as converging diverging coefficient, under normal circumstances, parameter beta
It can be used and controlled with the mode that the number of iterations linearly reduces.For the uniform random number on section (0,1).The receipts of particle i
Process is held back with pointFor attractor, coordinate are as follows:
WhereinIt is equally distributed random number on a section (0,1).
In formula (14)Referred to as be averaged desired positions, is defined as all particle individual desired positions
It is average, i.e.,
After being updated to the position of each particle, using the fitness for finding out each particle with step 2.2 same procedure
Functional value, then by following formula more new individual desired positions:
The individual desired positions of each particle obtained by above formula save: so far, having minimum fitness
The position of functional value.
After the individual desired positions of each particle determine, so that it may update global desired positions according to formula (13).
Step 2.4, when reaching iteration termination condition, training terminate, global desired positions are Gabor to be determined
One group of optimal value of filtering parameter;Otherwise, the number of iterations adds 1, goes to step 2.3.
Iteration termination condition is usually that the number of iterations n is equal to max_n.
Step 3 constructs Optimal Gabor Filters by optimized parameter, carries out Gabor volumes to knitted fabric image to be detected
Product processing;
Optimized parameter construction Optimal Gabor Filters G* (x, y) obtained by step 2, to knitted fabric figure to be detected
As being that S (x, y) carries out Gabor process of convolution, the image Q (x, y) after obtaining convolution:
Step 4, progress binary conversion treatment obtain Defect Detection result.
In image Q (x, y) after convolution, flawless region and area image defective have different energy and ring
It should be worth, the energy value E of each location of pixels can be obtained by formula (8)r(x, y).Binary conversion treatment is carried out by following formula again:
Wherein B (x, y) is bianry image, and according to formula (10) (11), μ is the average energy value of image after convolution, and σ is energy
Standard deviation is measured, c is an experimental constant, is obtained by experiment.
B (x, y) is exactly the final result of Defect Detection, by B (x, y) to determine whether containing flaw.If the value of B (x, y)
It is 1, then the corresponding location of pixels of image to be detected has flaw;If the value of B (x, y) is 0, the corresponding picture of image to be detected
Plain position is indefectible.
Claims (1)
1. a kind of knitted fabric flaw detection method based on Optimal Gabor Filters, characterized in that the following steps are included:
Step 1, construction Gabor filter, extract Gabor filtering parameter;
The step 1 follows the steps below to implement:
Step 1.1 establishes Gabor filter function, specifically implements in accordance with the following methods:
Gabor filter function G (x, y) is modulated by the two-dimentional Gaussian kernel function g (x, y) that the multiple SIN function of directionality tunes
It forms, in the spatial domain, Gaussian kernel function is related to three parameter δx, δy, θ, wherein δxFor Gaussian kernel function x-axis
Scale parameter on direction, δyFor the scale parameter on Gaussian kernel function y-axis direction, θ is the rotation of Gaussian kernel function
Gyration, (x ', y ') are the coordinates after (x ', y ') rotation θ angle;In frequency domain, Fourier transformation is related to frequency
Three parameters are F0, u0, v0, wherein F0It is the centre frequency of oval Gabor filter function, u0It is that Gabor is filtered in x-axis direction
The centre frequency of function, v0It is the centre frequency of Gabor filter function on the direction y;
Gabor filter function in two-dimensional space is indicated are as follows:
Wherein,
Coefficient F in the Gabor filter function G (x, y) that the step 1.1 constructsoWhen=0, then 2-DGabor filter function is modulated to
G (x, y);u0=0, v0When=0, then the Gabor filter function G (x, y) of ellipse is modulated to;
Gabor filter function in spatial domain is fourier transformed to obtain the Gabor filter function of frequency domain by step 1.2
Wherein,
Step 1.3, the Gabor filter function constructed from step 1.2Middle extraction 6 Gabor filtering parameters (alpha, gamma, u0,
v0, F0, θ);
Step 2 carries out Gabor process of convolution to indefectible knitted fabric image, constructs fitness function using Fisher criterion,
Utilization sub-line is that particle group optimizing (QPSO) algorithm carries out optimization processing to the Gabor filtering parameter that step 1 is extracted, and is obtained
The optimized parameter of Gabor filter;
The step 2 follows the steps below to implement:
Step 2.1, initialization population, including determining maximum number of iterations, search space, the number of particle, random initializtion
The position of particle;
The step 2.1 follows the steps below to implement: the number of iterations n=0 when setting initial, maximum number of iterations max_n;
Gabor filtering parameter has (alpha, gamma, u0, v0, F0, θ), then search space is 6 dimensions;The number of particle be M, each particle it is initial
Position isWherein i=1,2 ..., M/;
Step 2.2, in first time iteration, the initial position of each particle is current individual desired positions;To indefectible warp knit
Textile image carries out Gabor process of convolution, constructs fitness function using Fisher criterion, calculates the corresponding letter of each particle
Numerical value;The fitness function value of all particles is compared finds the particle with minimum fitness function value, the position of the particle more afterwards
It sets as global desired positions;
Image R (x, y) of the image after Gabor convolution may be expressed as: in the step 2.2
Wherein, T (x, y) is indefectible knitted fabric image, and R (x, y) is the image after Gabor filter convolution, and * is image
Convolution operation,It is the Fourier transformation of image T (x, y), IDFT is inverse discrete fourier transform;
The energy of image R (x, y) after the Gabor convolution indicates are as follows:
Wherein, WithIt is G respectivelye(x, y) and GoThe discrete Fourier transform of (x, y);
The objective function according to Fisher criterion construction indicates are as follows:
Wherein, Gabor filtering parameter Φ=(alpha, gamma, u0, v0, F0, θ), μ (Φ) and σ (Φ) are the image that size is X × Y respectively
Average energy value and standard deviation after Gabor convolution;
There are 6 decision variables as a result, the nonlinear programming problem of 5 constraint conditions can be described as:
s.t.
0≤θ≤π;
In first time iteration, the initial position of each particle is current individual desired positions, i.e.,By Fisher standard
The objective function then constructed calculates the corresponding fitness function value of each particle;
The fitness function value of all particles is compared finds the particle with minimum fitness function value, the position of the particle more afterwards
As global desired positions;
Step 2.3 is updated the position of each particle, using the adaptation for finding out each particle with step 2.2 same procedure
Spend functional value, more new individual desired positions and global desired positions;
The location updating equation of particle in the step 2.3 are as follows:
Taking "+" in formula or taking the probability of "-" is all 0.5, and wherein β is known as converging diverging coefficient,It is equal on section (0,1)
Even distribution random numbers, the convergence process of particle i is with pointFor attractor, coordinate are as follows:
WhereinIt is equally distributed random number on a section (0,1);Indicate global desired positions GnD dimension value;
Following formula is used in the step 2.3 when more new individual desired positions:
After the individual desired positions of each particle determine, according toUpdate global desired positions;
Step 2.4, when reaching iteration termination condition, training terminates, and global desired positions are Gabor to be determined filtering
The optimal value of parameter;Otherwise, the number of iterations adds 1, goes to step 2.3;
Step 3, the Gabor filter optimized parameter obtained by step 2 carry out Gabor volumes to knitted fabric image to be detected
Product processing;
It is that S (x, y) carries out Gabor process of convolution to knitted fabric image to be detected in the step 3, the figure after obtaining convolution
As Q (x, y):
Step 4, progress binary conversion treatment obtain the Defect Detection result of knitted fabric;
Binary conversion treatment is carried out using following formula in the step 4:
Wherein B (x, y) is bianry image, is the final result of Defect Detection, if the value of B (x, y) is 1, image to be detected phase
Corresponding location of pixels has flaw;If the value of B (x, y) is 0, the corresponding location of pixels of image to be detected is indefectible;μ is volume
The average energy value of image after product, σ are that energy scale is poor, and c is experimental constant, is obtained by experiment.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510686033.0A CN105205828B (en) | 2015-10-20 | 2015-10-20 | Knitted fabric flaw detection method based on Optimal Gabor Filters |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510686033.0A CN105205828B (en) | 2015-10-20 | 2015-10-20 | Knitted fabric flaw detection method based on Optimal Gabor Filters |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105205828A CN105205828A (en) | 2015-12-30 |
CN105205828B true CN105205828B (en) | 2019-03-19 |
Family
ID=54953489
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510686033.0A Active CN105205828B (en) | 2015-10-20 | 2015-10-20 | Knitted fabric flaw detection method based on Optimal Gabor Filters |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105205828B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105678767B (en) * | 2016-01-07 | 2019-02-19 | 无锡信捷电气股份有限公司 | A kind of cloth surface flaw detection method based on SoC Hardware/Software Collaborative Design |
CN106204543B (en) * | 2016-06-29 | 2019-08-06 | 江南大学 | Fabric defect detection method based on single category support vector machines |
CN107256545B (en) * | 2017-05-09 | 2019-11-15 | 华侨大学 | A kind of broken hole flaw detection method of large circle machine |
CN108010029B (en) * | 2017-12-27 | 2020-11-03 | 江南大学 | Fabric defect detection method based on deep learning and support vector data description |
CN108986065B (en) * | 2018-04-19 | 2021-11-23 | 三明学院 | Warp knitting fabric flaw fusion filtering detection method, device, equipment and storage medium |
CN109816631A (en) * | 2018-12-25 | 2019-05-28 | 河海大学 | A kind of image partition method based on new cost function |
CN112036541B (en) * | 2020-10-16 | 2023-11-17 | 西安工程大学 | Fabric defect detection method based on genetic algorithm optimization neural network |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5371808A (en) * | 1992-05-14 | 1994-12-06 | The United States Of America As Represented By The Secretary Of Commerce | Automated recognition of characters using optical filtering with maximum uncertainty - minimum variance (MUMV) functions |
US5684894A (en) * | 1994-11-28 | 1997-11-04 | Eastman Kodak Company | Scale specific and robust line/edge encoding of images |
CN101493520A (en) * | 2009-01-16 | 2009-07-29 | 北京航空航天大学 | SAR image variation detecting method based on two-dimension gamma distribution |
CN101882304A (en) * | 2010-06-24 | 2010-11-10 | 西北工业大学 | Self-adaptive de-noising and characteristic enhancing method of SAR (Synthetic Aperture Radar) image |
CN102609904A (en) * | 2012-01-11 | 2012-07-25 | 云南电力试验研究院(集团)有限公司电力研究院 | Bivariate nonlocal average filtering de-noising method for X-ray image |
CN103020654A (en) * | 2012-12-12 | 2013-04-03 | 北京航空航天大学 | Synthetic aperture radar (SAR) image bionic recognition method based on sample generation and nuclear local feature fusion |
CN103955922A (en) * | 2014-04-17 | 2014-07-30 | 西安工程大学 | Method for detecting flaws of printed fabric based on Gabor filter |
CN103971329A (en) * | 2014-05-26 | 2014-08-06 | 电子科技大学 | Cellular nerve network with genetic algorithm (GACNN)-based multisource image fusion method |
CN104198498A (en) * | 2014-09-12 | 2014-12-10 | 河海大学常州校区 | Method and device for detecting cloth flaws based on adaptive orthogonal wavelet transform |
CN104239903A (en) * | 2014-10-10 | 2014-12-24 | 江南大学 | QPSO (quantum-behaved particle swarm optimization) algorithm based image edge detection method |
-
2015
- 2015-10-20 CN CN201510686033.0A patent/CN105205828B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5371808A (en) * | 1992-05-14 | 1994-12-06 | The United States Of America As Represented By The Secretary Of Commerce | Automated recognition of characters using optical filtering with maximum uncertainty - minimum variance (MUMV) functions |
US5684894A (en) * | 1994-11-28 | 1997-11-04 | Eastman Kodak Company | Scale specific and robust line/edge encoding of images |
CN101493520A (en) * | 2009-01-16 | 2009-07-29 | 北京航空航天大学 | SAR image variation detecting method based on two-dimension gamma distribution |
CN101882304A (en) * | 2010-06-24 | 2010-11-10 | 西北工业大学 | Self-adaptive de-noising and characteristic enhancing method of SAR (Synthetic Aperture Radar) image |
CN102609904A (en) * | 2012-01-11 | 2012-07-25 | 云南电力试验研究院(集团)有限公司电力研究院 | Bivariate nonlocal average filtering de-noising method for X-ray image |
CN103020654A (en) * | 2012-12-12 | 2013-04-03 | 北京航空航天大学 | Synthetic aperture radar (SAR) image bionic recognition method based on sample generation and nuclear local feature fusion |
CN103955922A (en) * | 2014-04-17 | 2014-07-30 | 西安工程大学 | Method for detecting flaws of printed fabric based on Gabor filter |
CN103971329A (en) * | 2014-05-26 | 2014-08-06 | 电子科技大学 | Cellular nerve network with genetic algorithm (GACNN)-based multisource image fusion method |
CN104198498A (en) * | 2014-09-12 | 2014-12-10 | 河海大学常州校区 | Method and device for detecting cloth flaws based on adaptive orthogonal wavelet transform |
CN104239903A (en) * | 2014-10-10 | 2014-12-24 | 江南大学 | QPSO (quantum-behaved particle swarm optimization) algorithm based image edge detection method |
Non-Patent Citations (4)
Title |
---|
Optimal Gabor filters for textile flaw detection;BODNAROVA等;《Pattern Recognition》;20021231;第35卷(第12期);第2977-2978页 |
Texture segmentation using filters with optimized energy separation;T.Randen等;《IEEE Trans.Image Process》;19991231;第8卷(第4期);第571-582页 |
图像分割中关键技术的研究;王瑶;《中国优秀硕士学位论文全文数据库 信息科技辑》;20090315(第03期);第33-37页 |
基于Gabor滤波器组的织物疵点检测方法;韩润萍等;《计算机工程与应用》;20071231;第43卷(第7期);第211-214页 |
Also Published As
Publication number | Publication date |
---|---|
CN105205828A (en) | 2015-12-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105205828B (en) | Knitted fabric flaw detection method based on Optimal Gabor Filters | |
CN110927706B (en) | Convolutional neural network-based radar interference detection and identification method | |
CN107194919B (en) | Mobile phone screen defect detection method based on regular texture background reconstruction | |
Pixia et al. | Recognition of greenhouse cucumber disease based on image processing technology | |
CN110837768B (en) | Online detection and identification method for rare animal protection | |
CN107451999A (en) | foreign matter detecting method and device based on image recognition | |
Nurhaida et al. | Performance comparison analysis features extraction methods for batik recognition | |
CN104732536A (en) | Sub-pixel edge detection method based on improved morphology | |
CN106709450A (en) | Recognition method and system for fingerprint images | |
CN111161222B (en) | Printing roller defect detection method based on visual saliency | |
CN109242848A (en) | Based on OTSU and GA-BP neural network wallpaper defects detection and recognition methods | |
CN103955922A (en) | Method for detecting flaws of printed fabric based on Gabor filter | |
CN102096824A (en) | Multi-spectral image ship detection method based on selective visual attention mechanism | |
Rani et al. | Detection of edges using mathematical morphological operators | |
CN106663315A (en) | Method for denoising noisy image | |
Feng et al. | A new edge detection algorithm based on Canny idea | |
CN114972339A (en) | Data enhancement system for bulldozer structural member production abnormity detection | |
Sibi Chakkaravarthy et al. | Automatic leaf vein feature extraction for first degree veins | |
CN104899592B (en) | A kind of road semiautomatic extraction method and system based on circular shuttering | |
CN109636785A (en) | A kind of visual processing method identifying particles of silicon carbide | |
Tang et al. | A novel structure adaptive algorithm for feature-preserving 3d mesh denoising | |
CN110321808B (en) | Method, apparatus and storage medium for detecting carry-over and stolen object | |
CN116778558A (en) | Face wrinkle detection method and system based on adaptive hybrid hessian filter | |
CN110599510A (en) | Picture feature extraction method | |
Sun et al. | Comparison between mean filter and median filter algorithm in image denoising field |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |