CN101819028A - Machine vision detection system for unchy yarn shape parameters - Google Patents

Abstract

The invention discloses a machine vision detection system for bunchy yarn shape parameters. The system is applied to a method for detecting the quality of slubby yarn in a textile industry. The detection system mainly comprises slubby yarn image acquisition equipment and slubby yarn image processing equipment and a slubby yarn image structure and a slubby yarn image processing algorithm. Due to the adoption of the detection system, a statistical series of slub length, slub space, slub multiplying power and slub cycle can be obtained; the slubby yarn prepared by spinning according to the parameters can have the same vision effect as that of given slubby yarn; and the efficiency and production reliability can be greatly improved and the parameters and spinning effect of the slubby yarn can be visually displayed during the design process of processing materials provided by clients, the spinning process of the slubby yarn, and the quality detection process of the slubby yarn.

Description

Machine vision detection system for unchy yarn shape parameters

Technical field

The present invention relates to the textile industry field, relate in particular to and detect slub method for quality, particularly a kind of machine vision detection system for unchy yarn shape parameters.

Background technology

Slub is a kind of of flame, because its unique surface structure makes its fabric stereoscopic sensation strong, style is well-pressed various.For example: the jeans that the slub denim that is woven into slub is made, distinctive rough, free and easy style of denim and wear-resisting, good water absorption, characteristics that gas penetration potential is strong had both been kept, also has " raindrop, rain wire " special effects simultaneously, the cloth cover bamboo joint part is different with normal part decolouring behind the washing stone mill, stereoscopic sensation and imitative old sense are arranged very much, " rain line " style of random variation has more fine and smooth personal charm, is subjected to the young man deeply, and the elderly's favor is more arranged.

The slubbing of slub is called ring, and its length is called slub length.The basic yarn length of adjacent two rings is called slub space, and the ratio of ring diameter and basic yarn diameter is called slub multiple, generally in 1.5～6 scopes.Slub length, slub space, slub multiple are three basic parameters of slub, and they and basic yarn number number determine the various aspects of slub together in process of production.The circulation law of slub length, slub space, slub multiple and slub is the slub main design contents, is prerequisite and resultant yarn and the style of fabric and the demand whether finished product satisfies the client basic, the decision slub of accurate spinning bunchy yarn.

Present slub detection method mainly contains manual detection method and instrument detecting method, the weakness that ubiquity efficient is low, degree of accuracy is not high, and method of operating loaded down with trivial detailsly, time-consuming can not satisfy fast, the short run market demands.

Manual detection is mainly measured by ruler, and is aided with blackboard and calculates.This method is intuitively simple, but degree of accuracy is low.Rely on gauger's experience level fully, because adopting ruler measures slub length and slub space, determines the ring mean diameter, uses the circulation law of seeking slub around the mode of blackboard with the mode of weighing, can only be adapted to small sample (length of yarn is little), measurement that the ring multiplying power is single, just time-consuming, loaded down with trivial details for large sample (length of yarn long), irregular slub, and precision is not high.

The instrument detecting method mainly adopts Sweden USTER company and homemade imitated bar evenness detector thereof to detect, and utilization obtains its manual analysis being obtained indirectly the parameter of bunchy yarn after wave spectrogram and the outward appearance curve map after detecting again.

The principle of measurement by capacitance yarn evenness and method are to allow yarn pass through between the long detection groove (plane-parallel capacitor pole plate) of 8mm with certain speed, cause the media variations of electric capacity owing to quality (line density) variation of yarn unit length, and then causing the variation of electric capacity, the variation of electric capacity promptly reflects the irregular degree of yarn evenness.In order to detect different yarns, constitute five or four by seven or five pole plates and detect grooves, the pole plate area is different with spacing, adapts to different yarns kind and number number respectively, and this method mainly contains following shortcoming:

1, this method obtains wave spectrogram and outward appearance curve map, only is the indirect analog result of bunchy yarn, and there is very big difference in the bunchy yarn visual effect of directly seeing with human eye, and the quality correlativity final with fabric is not strong.The fineness of supposing every fiber of yarn equates, condenser type detects so is whether the number of fiber of yarn cross section is consistent, and is indifferent to closely dredging with pine of fiber obvolvent, promptly is indifferent to the yarn appearance shape.

2, this method is not suitable for the yarn that contains conductive fiber (tinsel, carbon fiber etc.).

3, the length of this method effective detection zone when measuring yarn is 8mm, is 8mm so the dried instrument of condenser type bar can detect the irregular minimum actual detected length (equivalent Cutting Length) of yarn evenness, and the evenness fault within the 8mm can't reflect.

4, electric capacity is very strong to the susceptibility of yarn water capacity, and the homogeneity of yarn moisture absorption has a strong impact on testing result.

5, the cleaning of capacitor plate, the spot that the flyings between pole plate, pole plate itself are infected with has a strong impact on test result.

6, the unevenness of electric field between capacitor plate, sliver has a strong impact on test result by the position difference of pole plate groove, need keep former form of yarn and the position between the pole plate groove as much as possible.

7, the wave spectrogram confidence level needs the sample of yarn length of necessary length to ensure.

8, do not have supporting slub special measurement software, can't measure ring circle statistics sequence (rule), and wave spectrogram and the manual analysis of outward appearance curve map are obtained the parameter of bunchy yarn indirectly, can not directly apply to the control of spinning equipment.

In addition, the CT1000 type yarn appearance analyser that Chang Ling, Shaanxi company develops is the photo-electric yarn evenness tester, still can't be equal to mutually at present for various reasons with capacitive, also can only indirect application detect, almost be not applied to production practices because the slub that does not match detects software in some outward appearances of slub.

Summary of the invention

The invention provides a kind of machine vision detection system for unchy yarn shape parameters, this detection system will address the above problem.

This machine vision detection system for unchy yarn shape parameters of the present invention mainly adopts slub image capture device and slub image processing equipment, and forms by following method:

A, slub picture structure:

Acquisition resolution, slub be radially: every mm20 pixel, and slub is axial: every mm2～10 pixel;

The background gray levels scope: 0～10, Yarn filoplume gray-scale value scope: 50～80, yarn gray-scale value scope: 120～230;

The picture size of the every mm10 of the slub pixel that 500m is long is (collection) 1024 * 5000000 (demonstration) 256 * 5000000;

B, slub image processing algorithm:

Slub diameter extracting method: threshold setting is 120～230, and width is 7 medium filtering and mean filter; Obtain the diameter sequence;

Slub length and slub space extracting method: adopt ordered sequence cluster statistical analysis algorithms identification calculating rudimentary algorithm to be:

The cluster statistic algorithm of ordered sequence is realized the identification calculating of slub length and slub space,

If use x ₁, x ₂, x _nRepresent one group of ordered sample, then the sample of each group must be

{ x _i, x _I+1, L, x _j(form of i＜j),

All possibility point-scores that the individual sample in order of N is divided into the k class have C _N-1 ^K-1Kind, under certain loss function meaning, having provided a kind of method of asking optimum solution, basic thought is:

1, the diameter of definition class: with D (i, j) expression { x _i, x _I+1, K, x _jDiameter, it is defined as

$D(i,j)=\underset{j=i}{\overset{j}{\mathrm{\Σ}}}{(x_l-\stackrel{\‾}{x_{\mathrm{ij}}})}^2$

Wherein

Calculate the diameter of all possibility classes,

2, use b _{N, k}Represent that n sample is divided into a kind of method of k class:

b _n，k：{i ₁＝1，i ₁+1，L，i ₂-1}，{i ₂，i ₂+1，L，i ₃-1}，L，{i _k＝1，i _k+1+1，L，i _k+1+1＝n}，

The loss of this method is

$L\left(b_{n,k}\right)=\underset{j=1}{\overset{k}{\mathrm{\Σ}}}D(i_j,i_{j+1}-1)$

b _{N, k} ^*Be to make L (b _{N, k}) reach minimum separating,

3, the entire loss that is divided into the optimum solution correspondence of j class of a preceding i sample can obtain with following two recursion formula:

$L\left(b_{n,2}^{*}\right)=\underset{2\≤j\≤n}{\min }\{D(1,j-1)+D(j,n)\}$

$L\left(b_{n,k}^{*}\right)=\underset{k\≤j\≤n}{\min }\{L(b_{j-1}^{*},j-1)+D(j,n)\}$

Concrete steps are as follows:

1, calculate the diameter D that all may classes (i, j) (i＜j),

2, calculate least disadvantage L (b _{N, k} ^*) (2≤j≤i≤n),

3, determine the number of class, and provide optimum solution.

This machine vision detection system for unchy yarn shape parameters provided by the invention has following advantage:

1, uses high-speed high precision line array CCD industrial camera collection (the slub image of long per 10 lines of＞500m/mm); The ultrahigh resolution slub image acquisition of user-defined format involves the consistance of three parameters of design, the yarn travelling speed of film speed, the illumination spectrum of CCD and sets, and need could determine by Theoretical Calculation and repetition test.The distribution of control slub gradation of image value is roughly: background gray levels scope: 0～10; Yarn filoplume gray-scale value scope: 50～80; Yarn gray-scale value scope: 120～230; On this basis, can realize image processing algorithms such as rapid extraction slub diameter, filoplume;

2, design ultrahigh resolution 1024 * 5000000 Flame Image Process fast algorithms come denoising, extract slub border etc.;

3, design ultrahigh resolution 1024 * 5000000 method for displaying image are realized the easy to operate slub display mode of user: left and right sides translation, location positioning, the demonstration of slub parameter etc.;

4, design slub form parameter recognition technology realizes Boundary Extraction, diameter identification, slub length, slub space, slub multiple identification;

5, the slub form parameter statistical analysis technique of ordered sequence comes the circulation law of statistical study slub length, slub space, slub multiple (pseudorandom);

6, the form parameter of slub need be transferred to the slub that twist yarn equipment is controlled the appointment of weaving out in real time; Convert the form parameter of slub to control twist yarn equipment machine order, control the qualified slub of weaving out in real time;

7, use slub length, slub space, slub multiple and occurrence law emulation slub and fabric effect thereof.

Embodiment

This machine vision detection system for unchy yarn shape parameters of the present invention adopts following hardware device:

The slub image capture device:

Line array CCD industrial camera 10bit 2048pixel 20KHz

PCI image pick-up card Camera Link interface

The highlighted full spectrum loop configuration of LED lighting source

Yarn movement control device PLC, stepper motor, rubber tire are right

The slub image processing equipment:

PC, Intel CPU 2.66 GHz, 2GB internal memory

WindowXP operating system VC++6.0 design language

The described algorithm of operation technique scheme

In ring diameter assorting process, be divided into 3 classes: ring, basic yarn, cotton defect knot;

This algorithm can make a distinction cotton defect knot with ring, obtain the random sequence of ring, spun yarn, cotton defect knot;

Add up slub length, slub space, slub multiple on this basis;

The finding method of ring circle statistics sequence (rule):, select the search foundation of the sequence of values of slub space for use as ring circle statistics sequence (rule) because the length variations of ring is significantly smaller than the length variations of slub space.

Concrete grammar: (1) maximal value is sought;

(2) sequence between each maximal value relatively;

(3) collating sequence sequence inequality is an accurate cyclic sequence;

(4) carry out (2) and (3) repeatedly until to finding identical sequence.

For bluring cyclic sequence at random, write down its sequence with big data form, determine the length of cyclic sequence according to concrete application requirements.

By the slub image processing algorithm, design software can obtain slub length, slub space, ring multiplying power, ring circle statistics sequence (rule), and the slub of producing according to these parameter weavings can have and the same visual effect of appointment slub.In the spinning process of the design process of processing with foreign materials slub, slub, slub quality testing process, can raise the efficiency greatly and the production reliability; And intuitively show slub parameter and the weaving effect.

Use method of the present invention, adopt light mechanical and electrical integration and computer image processing technology, adopt high precision driving, intelligent control, high reliability technology, can develop the instrument or the system of advanced fast detecting slub form parameter, promote the update of weaving conventional equipment and production line, realize the digitizing textile machine, will improve slub product quality, product category and production efficiency greatly, produce both at home and abroad and the needs of the high speed development in market to satisfy.

Claims (1)

1. a machine vision detection system for unchy yarn shape parameters mainly adopts slub image capture device and slub image processing equipment, it is characterized in that following method:

A, slub picture structure:

Acquisition resolution, slub be radially: every mm20 pixel, and slub is axial: every mm2～10 pixel;

The background gray levels scope: 0～10, Yarn filoplume gray-scale value scope: 50～80, yarn gray-scale value scope: 120～230;

The picture size of the every mm10 of the slub pixel that 500m is long is (collection) 1024 * 5000000 (demonstration) 256 * 5000000;

B, slub image processing algorithm:

Slub diameter extracting method: threshold setting is 120～230, and width is 7 medium filtering and mean filter, obtains the diameter sequence;

Slub length and slub space extracting method: adopt ordered sequence cluster statistical analysis algorithms identification calculating rudimentary algorithm to be:

The cluster statistic algorithm of ordered sequence is realized the identification calculating of slub length and slub space,

Use x ₁, x ₂, x _nRepresent one group of ordered sample, the sample of each group must be

{ x _i, x _I+1, L, x _j(form of i＜j),

All possibility point-scores that the individual sample in order of N is divided into the k class have C _N-1 ^K-1Kind, under certain loss function meaning, provided a kind of method of asking optimum solution:

The diameter of a, definition class: with D (i, j) expression { x _i, x _I+1, K, x _jDiameter, it is defined as

$D(i,j)=\underset{j=i}{\overset{j}{\mathrm{\Σ}}}{(x_l-\stackrel{\‾}{x_{\mathrm{ij}}})}^2$

Wherein Calculate the diameter of all possibility classes,

B, use b _{N, k}Represent that n sample is divided into a kind of method of k class:

b _n，k：{i ₁＝1，i ₁+1，L，i ₂-1}，{i ₂，i ₂+1，L，i ₃-1}，L，{i _k＝1，i _k+1+1，L，i _k+1+1＝n}，

The loss of this method is

$L\left(b_{n,k}\right)=\underset{j=1}{\overset{k}{\mathrm{\Σ}}}D(i_j,i_{j+1}-1)$

b _{N, k} ^*Be to make L (b _{N, k}) reach minimum separating,

The entire loss that c, a preceding i sample are divided into the optimum solution correspondence of j class can obtain with following two recursion formula:

$L\left(b_{n,2}^{*}\right)=\underset{2\≤j\≤n}{\min }\{D(i,j-1)+D(j,n)\}$

$L\left(b_{n,k}^{*}\right)=\underset{k\≤j\≤n}{\min }\{L(b_{j-1}^{*},j-1)+D(j,n)\}$

Concrete steps are as follows:

A, calculate the diameter D that all may classes (i, j) (i＜j),

B, calculating least disadvantage L (b _{N, k} ^*) (2≤j≤i≤n),

C, determine the number of class, and provide optimum solution.