CN110345874A - A kind of new method based on mechanical vision inspection technology measurement pipe tobacco width - Google Patents
A kind of new method based on mechanical vision inspection technology measurement pipe tobacco width Download PDFInfo
- Publication number
- CN110345874A CN110345874A CN201910761617.8A CN201910761617A CN110345874A CN 110345874 A CN110345874 A CN 110345874A CN 201910761617 A CN201910761617 A CN 201910761617A CN 110345874 A CN110345874 A CN 110345874A
- Authority
- CN
- China
- Prior art keywords
- pipe tobacco
- value
- pixel wide
- image
- width
- 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
- 241000208125 Nicotiana Species 0.000 title claims abstract description 215
- 235000002637 Nicotiana tabacum Nutrition 0.000 title claims abstract description 215
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000005259 measurement Methods 0.000 title claims abstract description 19
- 238000005516 engineering process Methods 0.000 title claims abstract description 13
- 238000007689 inspection Methods 0.000 title claims abstract description 10
- 238000001514 detection method Methods 0.000 claims abstract description 19
- 230000011218 segmentation Effects 0.000 claims abstract description 8
- 238000000605 extraction Methods 0.000 claims abstract description 6
- 238000006243 chemical reaction Methods 0.000 claims abstract description 4
- 238000004364 calculation method Methods 0.000 claims description 22
- 238000001914 filtration Methods 0.000 claims description 11
- 230000003287 optical effect Effects 0.000 claims description 8
- 238000005070 sampling Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 4
- 230000000694 effects Effects 0.000 claims description 3
- 230000001174 ascending effect Effects 0.000 claims description 2
- 238000013461 design Methods 0.000 claims description 2
- 238000009499 grossing Methods 0.000 claims description 2
- 238000012163 sequencing technique Methods 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 abstract description 2
- 235000019504 cigarettes Nutrition 0.000 description 10
- 239000000284 extract Substances 0.000 description 3
- 238000012360 testing method Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 210000000988 bone and bone Anatomy 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/02—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- 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
- G06T2207/20032—Median filtering
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Geometry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The title of the invention patent is a kind of new method based on mechanical vision inspection technology measurement pipe tobacco width, and this patent is achieved by the steps of, the acquisition of pipe tobacco sample, and projector detection obtains pipe tobacco rgb image；Image preprocessing is carried out for the pipe tobacco rgb image got, generates the pipe tobacco image progress binary conversion treatment after pretreatment；Center framework extraction is carried out to the pipe tobacco image after binaryzation；Single region pipe tobacco pixel wide is used as to the distance of center framework to both sides；Pipe tobacco after extraction skeleton is segmented and calculates pixel wide in the section of pipe tobacco；The pixel wide of whole pipe tobacco is calculated and calculated carrying out section to pipe tobacco pixel wide in the section after segmentation；Developed width value calculating linear fit equation is detected with projector by pixel wide value and is verified.
Description
Technical field
It is using the pipe tobacco after chopping process section in cigarette process as experiment the invention belongs to tobacco physics detection field
Sample carries out pipe tobacco width measurement, is related to image color information feature extraction and image outline information characteristics extract, be related to
Image information is chosen, image processing algorithm and fit equation；
Background technique
Pipe tobacco is that former cigarette is machined to piece cigarette by multiple working procedures such as redryings and is refined, then passes through volume by the piece cigarette after refining
It is formed after the technology for making tobacco threds section chopping technique processing of cigarette enterprise, the cut width usually required that is 0.7mm-1.3mm；Work now
Under the skill condition of production, although can cause to cut rear pipe tobacco to chopping limit for width, due to the uncertainty of supplied materials in process
Can be jagged phenomena such as；
Based on the capacity limit of finished product cigarette itself, in the case that pipe tobacco width is wide, for finished product cigarette fillibility,
Gas permeability can have significant effect, and have bad experience during inhaling to finished product cigarette product so as to cause consumer；Therefore it cuts
The stability of silk width is of crucial importance for finished product cigarette；
Currently, national standard detection method is detected after being amplified using tobacco optical projector to pipe tobacco, measurement accuracy compared with
It is low, it can only achieve 0.1mm, since usually in 0.7mm-1.3mm, the error of 0.1mm deviates pipe tobacco local width after chopping
Be affected to the detection of single pipe tobacco, and for different testing staff, different detection gimmick in the case where, detection error also can
It is larger；In recent years with the high speed development of machine vision and artificial intelligence, machine is had also been introduced in a variety of physical detections in industry
Vision, and also there are many detection methods based on machine vision for the detection of pipe tobacco width；In these existing methods, have logical
Pipe tobacco center line is crossed to detect, in actual application, more for burr pipe tobacco detection, can exist it is biggish not
Stability；It averages the method for characterizing pipe tobacco width also by multiple spot maximum inscribed circle diameter in pipe tobacco, in practical application
In, as the uncertainty of pipe tobacco burr, leads to the unstability to pipe tobacco width detection；Therefore based on machine vision
The detection of pipe tobacco width stability has become the technological difficulties in the urgent need to address of worker in industry.
Summary of the invention
In view of above-described technical defect at present, thus we are by a set of completely new method come to pipe tobacco width
It is more accurately measured, the purpose of the present invention is to provide a kind of based on the new of mechanical vision inspection technology measurement pipe tobacco width
Method leads to width detection unstability to pipe tobacco burrs on edges in the prior art for solving the problems, such as.
To achieve the above object, the present invention provides a kind of new sides based on mechanical vision inspection technology measurement pipe tobacco width
Method, the new method based on mechanical vision inspection technology measurement pipe tobacco width, this method successively includes: to obtain true coloured picture
Picture；Image preprocessing is carried out for the rgb image got, generates the pipe tobacco binary image after pretreatment；To pipe tobacco figure
As carrying out center framework extraction；Single region pipe tobacco pixel wide is used as to the distance of center framework to both sides；After extracting skeleton
Pipe tobacco is segmented and calculates pixel wide in the section of single pipe tobacco；It is counted carrying out section to pipe tobacco pixel wide in the section after segmentation
It calculates, and calculates the pixel wide of single pipe tobacco；And the pixel wide of pipe tobacco is converted to the practical width of pipe tobacco by linear fit
Degree.
The present invention provides a kind of new methods based on mechanical vision inspection technology measurement pipe tobacco width, which is characterized in that
Include:
Step 1: the pipe tobacco in chopping process section filament cutter exit is sampled, pipe tobacco is carried out by optical projector
The measurement of width carries out multiple spot detection for single pipe tobacco, and obtains developed width of the mean value as pipe tobacco of multiple detected values
Value；
Step 2: image checking is carried out by the pipe tobacco that optical projector detects in step 1, and it is true to get pipe tobacco
Coloured picture picture；
Step 3: carrying out the pretreatment of image to the pipe tobacco rgb image got in step 2, and step includes:
1) median filter process is carried out using nonlinear smoothing filter to the rgb image of pipe tobacco, obtains original pipe tobacco image
Grayscale image, median filtering is a kind of typical nonlinear filtering technique, and basic thought is one kind based on sequencing statistical theory
The nonlinear signal processing technology of noise can effectively be inhibited, the basic principle is that digital picture or the ash of Serial No. pixel
The intermediate value of each point gray value replaces in one neighborhood of the angle value point, the true value for making the pixel value of surrounding close, to disappear
Except isolated noise spot simultaneously again can retain image edge detailss because it independent of in neighborhood those with representative value difference very
Big value.Median filter is similar with the working method of linear filter when handling consecutive image window function, but filtering
It but is no longer ranking operation.This design philosophy is exactly to check the sampling in input signal and judge whether it represents signal,
This functionality is realized using the observation window that odd-numbered samples form, and the gray values in watch window are ranked up, is located at and sees
The gray scale intermediate value among window is examined as output.Then, earliest value is abandoned, new sampling is obtained, repeats calculating process above.
Its calculation method are as follows:
If k1, k2, k3... knFor the sequence gray value in watch window, rear gray value is sorted in ascending order:
km1≤km2≤km3......≤kmn(n is gray value number in window, and m is the serial number of gray value sequence)
Or gray value after sorting in descending order:
km1≥km2≥km3......≥kmn
Then gray scale intermediate value are as follows:
In practical applications, n generally takes odd number, so
kIntermediate value=km(n+1)/2
2) method binary conversion treatment is iterated to the gray level image after obtained median filtering, gets original pipe tobacco image
Binary image, its calculation formula is:
K in formulamaxAnd kminThe maximum gradation value and minimum gradation value of image after median filtering are respectively indicated, gray threshold is then remembered
For T；
The gray value of image is divided into two groups of R1 and R2 according to gray threshold T；
μ in formula1And μ2It is the average gray value of R1 and two groups of R2, T1 is new gray threshold；
3) pixel number is carried out to the pipe tobacco image extracted and calculates i.e. region area calculating, region area is less than
The image of gray threshold T1 range is rejected, and remains larger than the image of gray threshold T1 range as pretreated pipe tobacco figure
Picture；
Step 4: carrying out center framework extraction to pipe tobacco image pretreated in step 3, mainly by being based on chasing after
The method of track judges that the local feature of pipe tobacco obtains the approximate center line trend in current location, and combines single cigarette in pipe tobacco image
The local feature information of silk, obtains the central point of current position object cross section, and by all central points in regional scope
It is concatenated into a center line, the center framework of as single pipe tobacco is implemented as follows:
(1) it marks deleted boundary point；
(2) point being marked is deleted；
(3) continue to mark deleted remaining boundary point；
(4) labeled point is deleted.
(5) until the point not being deleted, algorithm is terminated at this time, generates the skeleton in the region, i.e. pipe tobacco center framework；
Step 5: being used as single region pipe tobacco width to the distance of the pipe tobacco center framework extracted in step 4 to both sides,
Vertical range is mainly carried out by the pipe tobacco edge for selecting number point toward image after pretreatment to the pipe tobacco center framework extracted
It calculates, the pipe tobacco pixel wide by the sum of same point to the pixel distance at pipe tobacco both sides edge as single region；
Step 6: in step 4 extract center framework after pipe tobacco be segmented, mainly by being extracted to pipe tobacco in
Heart skeleton carries out the pipe tobacco pixel wide threshold value setting in single region, and to calculating obtained center framework list region in step 5
Pipe tobacco pixel wide on the width of each point carry out the setting of coefficient of deviation P worst error threshold value, while to pipe tobacco center framework list
The width established standards deviation S threshold range of all the points in the pipe tobacco pixel wide in region, when pipe tobacco pixel wide meets simultaneously
In the case where above 3 points, segmentation calculating is carried out to pipe tobacco, in section mean value be this section of pipe tobacco section between pixel wide value；Its
Calculation formula are as follows:
K is the pipe tobacco pixel wide in pipe tobacco center framework list region, and n is in the pipe tobacco pixel wide in pipe tobacco center framework list region
The number of all the points；
S represents the standard deviation value of the pipe tobacco pixel wide in single region, x in formulaiThe pipe tobacco pixel wide in single region is represented,
All the points mean value in the pipe tobacco pixel wide in single region is represented, n represents the quantity of the pipe tobacco pixel wide in single region；
P is the pipe tobacco pixel wide coefficient of deviation in single region, k in formulaiBe pipe tobacco center framework list region pipe tobacco pixel it is wide
Degree,Mean value in section；
Step 7: to pipe tobacco pixel wide value and the pixel wide of whole pipe tobacco between each section after being segmented in step 6
Value deviates threshold range setting by pixel wide carrying out section to the pipe tobacco after segmentation, while to pixel wide between each section
The setting of standard deviation threshold range is carried out, pixel wide the pipe tobacco section for the threshold range for meeting the above two o'clock setting is carried out equal
Value calculates, pixel wide mean value between the section finally obtained, the pixel wide of as single pipe tobacco, its calculation formula is:
V is pixel wide between each section of pipe tobacco center framework, and m is all the points in pixel wide between each section of pipe tobacco center framework
Number
S in formulaSectionThe standard deviation value of pixel wide, v between representing each sectioniPixel wide between representing each section,It represents each
All the points mean value in pixel wide between a section, the quantity of pixel wide between m represents each section；
Step 8: linear fit equation is calculated by pixel wide value and developed width value, mainly passes through light in step 1
It learns the pipe tobacco pixel wide got in the pipe tobacco width actual value and step 7 that projector calculates and carries out linear relationship fitting, and
The pixel wide of pipe tobacco is converted to the developed width of pipe tobacco；Its calculation formula is:
Wherein y is the practical pipe tobacco width value after calculating,Be obtain section between pixel wide mean value be single pipe tobacco picture
Plain width, k are y pairsRegression coefficient, that is, straight slope, b is that constant is intercept with y-axis；
Step 9: it obtains the pixel wide value of a collection of pipe tobacco image and goes out pipe tobacco with the linear fit prediction equation of step 8
Width value, and detect actual width value with projector and be compared, calculate mean absolute errorIts calculation formula:
tError=| yVery-yIn advance|
Wherein yVeryIt is the practical pipe tobacco width value of projector measurement, yIn advanceIt is the pipe tobacco width value with linear fit equation calculation
Mean absolute errorIts calculation formula:
Wherein t is the practical pipe tobacco width value of projector measurement and pipe tobacco width value error, a out of linear fit equation calculation are
Sample number.
Advantageous effect of the invention:
The present invention provides a kind of new method based on mechanical vision inspection technology measurement pipe tobacco width, is put forward for the first time according to mark
The dual calculation method of quasi- deviation and mean value irrelevance, this method can largely carry out the pipe tobaccos abnormal conditions such as burr
It rejects；
The present invention provides a kind of new method based on mechanical vision inspection technology measurement pipe tobacco width, and first passage is to single
Pipe tobacco carries out segmentation recombination, and is associated calculating to the pipe tobacco width in section and between section, and eventually by the side for seeking mean value
Method characterizes pipe tobacco width, this method can largely compensate for because detection zone it is representational it is not strong caused by detection error
Phenomenon.
Detailed description of the invention
Fig. 1 Figure of abstract
Fig. 2 pipe tobacco original image
Image after Fig. 3 median filtering
Image after Fig. 4 center framework extracts
Image after pipe tobacco is segmented after Fig. 5 center framework
Case study on implementation
A specific embodiment of the invention is described with reference to the drawings below by way of specific example.
Step 1: carrying out the acquisition of standard specimen pipe tobacco to the pipe tobacco in chopping process section filament cutter exit, and quantity of sampling quantity is 5,
Then, the measurement that pipe tobacco width is carried out by optical projector carries out average 5 test points of selection for single pipe tobacco and carries out width
Degree detection, and the width value of 5 detected values is obtained, developed width value of the mean value as the pipe tobacco of each width value is calculated, such as table 1
It is shown；
15 pipe tobacco developed width values of table
| Serial number | Width value 1 | Width value 2 | Width value 3 | Width value 4 | Width value 5 | Width mean value |
| 1 | 1 | 1.1 | 1 | 1 | 1.2 | 1.06 |
| 2 | 1.1 | 1 | 1 | 1.1 | 1.1 | 1.06 |
| 3 | 1.1 | 1.1 | 1 | 1.1 | 1.4 | 1.14 |
| 4 | 1.3 | 1.4 | 1.3 | 1.2 | 1.1 | 1.26 |
| 5 | 1.2 | 1.2 | 1.1 | 1 | 1.1 | 1.12 |
Step 2: shown in Fig. 1, utilize industrial camera in specific light in 5 standard specimen pipe tobaccos by Machine Vision Detection method
According to carrying out Image Acquisition under environment, and get the colour imaging of 5 standard specimen pipe tobaccos；
Step 3: shown in Fig. 2, the pretreatment of image is carried out to the 5 standard specimen pipe tobacco rgb images got in step 2,
Include: with its step
1) rgb image of pipe tobacco is filtered using median filter, obtains the grayscale image of original pipe tobacco image；
2) method binary conversion treatment is iterated to obtained gray level image, gets the binary image of original pipe tobacco image；
3) region area calculating is carried out to the pipe tobacco image extracted, the figure of threshold value T1=500 range is less than for region area
As being rejected, the image of threshold value T1=500 range is remained larger than as pretreated pipe tobacco image；
Step 4: shown in Fig. 3, center bone is carried out to second pipe tobacco image of 5 pipe tobaccos pretreated in step 3
Frame extracts, and mainly judges that the local feature of pipe tobacco obtains the approximate center line in current location and becomes by the method based on tracking
Gesture, and the local feature information of single pipe tobacco in pipe tobacco image is combined, the central point of current position object cross section is obtained, and
All central points in regional scope are concatenated into a center line, the center framework of as single pipe tobacco；
Step 5: being used as single region pipe tobacco width to the distance of the pipe tobacco center framework extracted in step 4 to both sides,
Vertical range is mainly carried out by the pipe tobacco edge for selecting number point toward image after pretreatment to the pipe tobacco center framework extracted
It calculates, the pipe tobacco pixel wide by the sum of same point to the pixel distance at pipe tobacco both sides edge as single region；
Shown in Fig. 5, step 6: pipe tobacco after extracting center framework in step 4 is segmented, mainly by being mentioned to pipe tobacco
The pipe tobacco pixel wide threshold value that the center framework got carries out single region is set as 500, and obtained to calculating in step 5
The width progress coefficient of deviation P worst error threshold value of each point is set as 0.15 in the pipe tobacco pixel wide in center framework list region,
The width established standards deviation S threshold value of all the points in the pipe tobacco pixel wide in pipe tobacco center framework list region is set as simultaneously
0.15, the pipe tobacco that pipe tobacco pixel wide meets the case where above at 3 simultaneously has 7 sections of pipe tobaccos.
Step 7: after being divided into 7 segmentations to pipe tobacco in step 6, pipe tobacco pixel wide deviates threshold range setting between each section
For maximum width 500, minimum widith 0.5, while standard deviation threshold is carried out to pixel wide between each section and is set as SSection=
0.15, mean value computation is carried out between pixel wide the pipe tobacco section for the threshold range for meeting the above two o'clock settingThe pixel wide of as single pipe tobacco, other 4 pipe tobaccos remaining to 5 pipe tobaccos are arrived according to step 4
Pixel wide mean value is 7.8814,8.2695,8.6955,7.6632 between step 7 successively carries out calculating their section；
Step 8: it is got in the 5 pipe tobacco width actual values and step 7 calculated by optical projector in step 1
Pipe tobacco pixel wide carry outLinear relationship fitting, obtain k=0.1148, b=0.1712；
Step 9: obtaining the image pixel width value of 20 pipe tobaccos, by the linear fit equation calculation of step 8 they
Pipe tobacco width value, and detect actual width value with projector and be compared, their mean absolute error isIllustrate that this method can effectively reject pipe tobacco burr, obtain width value, specific data are simultaneously as shown in table 2
2 20 pipe tobacco picture traverse values of table and actual comparison table
Claims (8)
1. a kind of new method based on mechanical vision inspection technology measurement pipe tobacco width, which comprises the following steps:
Step 1: the pipe tobacco in chopping process section filament cutter exit being sampled, pipe tobacco width is carried out by optical projector
Measurement carries out multiple spot detection for single pipe tobacco, and obtains developed width value of the mean value as pipe tobacco of multiple detected values；
Step 2: image checking being carried out by the pipe tobacco that optical projector detects in step 1, and gets the true coloured picture of pipe tobacco
Picture；
Step 3: the pretreatment of image is carried out to the pipe tobacco rgb image got in step 2；
Step 4: center framework extraction is carried out to pipe tobacco image pretreated in step 3；
Step 5: single region pipe tobacco width is used as to the distance of the pipe tobacco center framework extracted in step 4 to both sides；
Step 6: pipe tobacco after extracting center framework in step 4 is segmented；
Step 7: to pipe tobacco pixel wide value and the pixel wide value of whole pipe tobacco between each section after being segmented in step 6；
Step 8: linear fit equation is calculated by pixel wide value and developed width value；
Step 9: obtaining the pixel wide value of a collection of pipe tobacco image and go out the width of pipe tobacco with the linear fit prediction equation of step 8
Angle value, and detect actual width value with projector and be compared, calculate mean absolute error。
2. true to the pipe tobacco got in step 2 the method according to claim 1, wherein in the step 3
Coloured picture picture carries out the pretreatment of image, and step includes:
1) median filter process is carried out using nonlinear smoothing filter to the rgb image of pipe tobacco, obtains the ash of original pipe tobacco image
Degree figure, median filtering are a kind of typical nonlinear filtering techniques, and basic thought is a kind of can have based on sequencing statistical theory
Effect inhibits the nonlinear signal processing technology of noise, the basic principle is that digital picture or the gray value of Serial No. pixel
It is replaced with the intermediate value of each point gray value in a neighborhood of the point, the true value for making the pixel value of surrounding close, to eliminate orphan
Vertical noise spot can retain image edge detailss again simultaneously, because it is independent of those are very big with representative value difference in neighborhood
Value.Median filter is similar with the working method of linear filter when handling consecutive image window function, but filtering is not
It is ranking operation again.This design philosophy is exactly to check the sampling in input signal and judge whether it represents signal, is used
The observation window of odd-numbered samples composition realizes this functionality, is ranked up to the gray values in watch window, is located at observation window
Intermediate gray scale intermediate value is as output.Then, earliest value is abandoned, new sampling is obtained, repeats calculating process above.It is counted
Calculation method are as follows:
If k1, k2, k3... knFor the sequence gray value in watch window, rear gray value is sorted in ascending order:
km1≤km2≤km3......≤kmn(n is gray value number in window, and m is the serial number of gray value sequence)
Or gray value after sorting in descending order:
km1≥km2≥km3......≥kmn
Then gray scale intermediate value are as follows:
In practical applications, n generally takes odd number, so
kIntermediate value=km(n+1)/2
2) method binary conversion treatment is iterated to the gray level image after obtained median filtering, gets the two of original pipe tobacco image
Value image, its calculation formula is:
K in formulamaxAnd kminThe maximum gradation value and minimum gradation value of image after median filtering are respectively indicated, gray threshold is then remembered
For T；
The gray value of image is divided into two groups of R1 and R2 according to gray threshold T；
μ in formula1And μ2It is the average gray value of R1 and two groups of R2, T1 is new gray threshold；
3) pixel number is carried out to the pipe tobacco image extracted and calculates i.e. region area calculating, gray scale is less than for region area
The image of threshold value T1 range is rejected, and remains larger than the image of gray threshold T1 range as pretreated pipe tobacco image.
3. the method according to claim 1, wherein in the step 4, to pipe tobacco pretreated in step 3
Image carries out center framework to extract being mainly to judge that the local feature acquisition current location of pipe tobacco is close by the method based on tracking
As center line trend, and combine pipe tobacco image in single pipe tobacco local feature information, it is transversal to obtain current position object
The central point in face, and all central points in regional scope are concatenated into a center line, the center framework of as single pipe tobacco,
It is implemented as follows:
(1) it marks deleted boundary point；
(2) point being marked is deleted；
(3) continue to mark deleted remaining boundary point；
(4) labeled point is deleted.
(5) until the point not being deleted, algorithm is terminated at this time, generates the skeleton in the region, i.e. pipe tobacco center framework.
4. the method according to claim 1, wherein in the step 5, in the pipe tobacco extracted in step 4
The distance on heart skeleton to both sides is used as single region pipe tobacco width, mainly by selecting number point to the pipe tobacco center framework extracted
The pipe tobacco edge of image carries out vertical range calculating after toward pretreatment, by same point to the sum of the pixel distance at pipe tobacco both sides edge
Pipe tobacco pixel wide as single region.
5. the method according to claim 1, wherein in the step 6, after extracting center framework in step 4
Pipe tobacco is segmented, and is mainly set by the pipe tobacco pixel wide threshold value that the center framework extracted to pipe tobacco carries out single region,
And coefficient of deviation P is carried out to the width of each point in the pipe tobacco pixel wide for calculating obtained center framework list region in step 5
The setting of worst error threshold value, while to the width established standards of all the points in the pipe tobacco pixel wide in pipe tobacco center framework list region
Deviation S threshold range more than pipe tobacco pixel wide meets simultaneously in the case where 3 points, carries out segmentation calculating to pipe tobacco, in section
Mean value be this section of pipe tobacco section between pixel wide value；Its calculation formula is:
K is the pipe tobacco pixel wide in pipe tobacco center framework list region, and n is in the pipe tobacco pixel wide in pipe tobacco center framework list region
The number of all the points；
Standard deviation calculation:
S represents the standard deviation value of the pipe tobacco pixel wide in single region, x in formulaiThe pipe tobacco pixel wide in single region is represented,
All the points mean value in the pipe tobacco pixel wide in single region is represented, n represents the quantity of the pipe tobacco pixel wide in single region；
Irrelevance calculates:
P is the pipe tobacco pixel wide coefficient of deviation in single region, k in formulaiBe pipe tobacco center framework list region pipe tobacco pixel it is wide
Degree,Mean value in section.
6. the method according to claim 1, wherein in the step 7, to each section after being segmented in step 6
Between pipe tobacco pixel wide value and the pixel wide value of whole pipe tobacco, by between after segmentation pipe tobacco carry out section pixel wide it is inclined
It is set from threshold range, while the setting of standard deviation threshold range is carried out to pixel wide between each section, to meeting the above two o'clock
Pixel wide carries out mean value computation between the pipe tobacco section of the threshold range of setting, pixel wide mean value between the section finally obtained, as
The pixel wide of single pipe tobacco, its calculation formula is:
V is pixel wide between each section of pipe tobacco center framework, and m is all the points in pixel wide between each section of pipe tobacco center framework
Number
Standard deviation calculation:
S in formulaSectionThe standard deviation value of pixel wide, v between representing each sectioniPixel wide between representing each section,It represents each
All the points mean value in pixel wide, the quantity of pixel wide between m represents each section between section.
7. the method according to claim 1, wherein passing through pixel wide value and developed width in the step 8
Value calculates linear fit equation, obtains in the main pipe tobacco width actual value and step 7 calculated by optical projector in step 1
The pipe tobacco pixel wide got carries out linear relationship fitting, and the pixel wide of pipe tobacco is converted to the developed width of pipe tobacco；Its
Calculation formula are as follows:
Linear relationship calculates:
Wherein y is the practical pipe tobacco width value after calculating,Be obtain section between pixel wide mean value be single pipe tobacco pixel
Width, k are y pairsRegression coefficient, that is, straight slope, b is that constant is intercept with y-axis.
8. the method according to claim 1, wherein the pixel for obtaining a collection of pipe tobacco image is wide in the step 9
Angle value and the width value for going out pipe tobacco with the linear fit prediction equation of step 8, and detect actual width value with projector and carry out
It compares, calculates mean absolute errorIts calculation formula:
tError=| yVery-yIn advance|
Wherein yVeryIt is the practical pipe tobacco width value of projector measurement, yIn advanceIt is the pipe tobacco width value with linear fit equation calculation
Mean absolute errorIts calculation formula:
Wherein t is the practical pipe tobacco width value of projector measurement and pipe tobacco width value error, a out of linear fit equation calculation are
Sample number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910761617.8A CN110345874A (en) | 2019-08-16 | 2019-08-16 | A kind of new method based on mechanical vision inspection technology measurement pipe tobacco width |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910761617.8A CN110345874A (en) | 2019-08-16 | 2019-08-16 | A kind of new method based on mechanical vision inspection technology measurement pipe tobacco width |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110345874A true CN110345874A (en) | 2019-10-18 |
Family
ID=68180657
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910761617.8A Pending CN110345874A (en) | 2019-08-16 | 2019-08-16 | A kind of new method based on mechanical vision inspection technology measurement pipe tobacco width |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110345874A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111412834A (en) * | 2020-04-08 | 2020-07-14 | 昆明超泰经贸有限公司 | Tobacco bale paper indentation data detection system and detection method thereof |
| CN112330663A (en) * | 2020-11-25 | 2021-02-05 | 中国烟草总公司郑州烟草研究院 | Computer vision tobacco shred width detection method based on variable diameter circle |
| CN113570584A (en) * | 2021-07-30 | 2021-10-29 | 河南中烟工业有限责任公司 | Cut tobacco structure detection method based on image recognition |
| CN113838123A (en) * | 2021-08-16 | 2021-12-24 | 湖南磐钴传动科技有限公司 | Method for measuring tobacco shred morphology characteristics based on image processing |
| CN115496753A (en) * | 2022-11-16 | 2022-12-20 | 山东中烟工业有限责任公司青州卷烟厂 | Tobacco shred structure detection method based on machine vision |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103439365A (en) * | 2013-08-22 | 2013-12-11 | 浙江工商大学 | Device and method for detecting water content of fresh tobacco leaves |
| CN104016102A (en) * | 2014-03-27 | 2014-09-03 | 上海创和亿电子科技发展有限公司 | Error-free whole-box tobacco nicotine chemical value transmission method |
| CN106250896A (en) * | 2016-03-03 | 2016-12-21 | 上海创和亿电子科技发展有限公司 | The recognition methods of the positive and negative of online Nicotiana tabacum L. based on image collecting device |
| CN106289070A (en) * | 2016-08-03 | 2017-01-04 | 上海创和亿电子科技发展有限公司 | The method measuring irregularly shaped object length and width |
| US20170238598A1 (en) * | 2014-09-12 | 2017-08-24 | International Tobacco Machinery Poland Sp. Z O.O. | Measuring apparatus and measuring method for multi-segment rod-like articles of tobacco industry |
| CN107798672A (en) * | 2016-08-31 | 2018-03-13 | 上海创和亿电子科技发展有限公司 | The width measurements method of pipe tobacco or stem, system and the server with the system |
| CN109829943A (en) * | 2018-11-13 | 2019-05-31 | 上海烟草集团有限责任公司 | Blade construction detection method, system, medium and equipment based on machine vision |
-
2019
- 2019-08-16 CN CN201910761617.8A patent/CN110345874A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103439365A (en) * | 2013-08-22 | 2013-12-11 | 浙江工商大学 | Device and method for detecting water content of fresh tobacco leaves |
| CN104016102A (en) * | 2014-03-27 | 2014-09-03 | 上海创和亿电子科技发展有限公司 | Error-free whole-box tobacco nicotine chemical value transmission method |
| US20170238598A1 (en) * | 2014-09-12 | 2017-08-24 | International Tobacco Machinery Poland Sp. Z O.O. | Measuring apparatus and measuring method for multi-segment rod-like articles of tobacco industry |
| CN106250896A (en) * | 2016-03-03 | 2016-12-21 | 上海创和亿电子科技发展有限公司 | The recognition methods of the positive and negative of online Nicotiana tabacum L. based on image collecting device |
| CN106289070A (en) * | 2016-08-03 | 2017-01-04 | 上海创和亿电子科技发展有限公司 | The method measuring irregularly shaped object length and width |
| CN107798672A (en) * | 2016-08-31 | 2018-03-13 | 上海创和亿电子科技发展有限公司 | The width measurements method of pipe tobacco or stem, system and the server with the system |
| CN109829943A (en) * | 2018-11-13 | 2019-05-31 | 上海烟草集团有限责任公司 | Blade construction detection method, system, medium and equipment based on machine vision |
Non-Patent Citations (1)
| Title |
|---|
| 赖长龙等: "基于旋转矢量不变性的烟叶宽度提取算法", 中国农机化学报, vol. 37, no. 8, 15 August 2016 (2016-08-15), pages 167 - 170 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111412834A (en) * | 2020-04-08 | 2020-07-14 | 昆明超泰经贸有限公司 | Tobacco bale paper indentation data detection system and detection method thereof |
| CN111412834B (en) * | 2020-04-08 | 2022-02-08 | 昆明超泰经贸有限公司 | Method for detecting cigarette packet paper indentation data |
| CN112330663A (en) * | 2020-11-25 | 2021-02-05 | 中国烟草总公司郑州烟草研究院 | Computer vision tobacco shred width detection method based on variable diameter circle |
| CN113570584A (en) * | 2021-07-30 | 2021-10-29 | 河南中烟工业有限责任公司 | Cut tobacco structure detection method based on image recognition |
| CN113838123A (en) * | 2021-08-16 | 2021-12-24 | 湖南磐钴传动科技有限公司 | Method for measuring tobacco shred morphology characteristics based on image processing |
| CN113838123B (en) * | 2021-08-16 | 2024-03-19 | 湖南磐钴传动科技有限公司 | Method for measuring appearance characteristics of cut tobacco based on image processing |
| CN115496753A (en) * | 2022-11-16 | 2022-12-20 | 山东中烟工业有限责任公司青州卷烟厂 | Tobacco shred structure detection method based on machine vision |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110345874A (en) | A kind of new method based on mechanical vision inspection technology measurement pipe tobacco width | |
| CN107255641B (en) | A method of Machine Vision Detection is carried out for self-focusing lens surface defect | |
| CN114445387A (en) | Fiberboard quality classification method based on machine vision | |
| CN106855520B (en) | Workpiece defect detection method based on machine vision | |
| CN109934811A (en) | A kind of optical element surface defect inspection method based on deep learning | |
| CN108073918B (en) | Method for extracting blood vessel arteriovenous cross compression characteristics of fundus retina | |
| CN106483129B (en) | A method of the leukorrhea trichomonad based on motion estimate detects automatically | |
| CN101201329A (en) | Method for extracting and recognizing nondestructive detecting defect based on radial damping energy field | |
| CN108288275A (en) | A kind of mechanically harvested seed cotton method for detecting impurities based on mobile intelligent terminal | |
| CN102637258A (en) | Method for creating online surface quality detection system defect library | |
| CN112330663A (en) | Computer vision tobacco shred width detection method based on variable diameter circle | |
| CN112233059A (en) | Light guide plate defect detection method based on segmentation and decision-making two-stage residual error attention network | |
| Shire et al. | Plain ceramic tiles surface defect detection using image processing | |
| CN111709919B (en) | Glass defect visual detection algorithm | |
| CN108776978A (en) | A kind of threshed redried strips piece shape characterizing method | |
| CN106706644A (en) | Method for detecting lens defects | |
| CN115866502A (en) | Microphone part surface defect online detection process | |
| CN115797361A (en) | Aluminum template surface defect detection method | |
| CN115266728A (en) | Method for identifying inclusion defects on surface of cold-rolled strip steel | |
| CN103177426B (en) | Morphology-based strong-interference laser edge image restoration method | |
| CN114985283A (en) | Tea leaf identification method based on multi-feature image identification technology | |
| CN117788464A (en) | Industrial gear oil impurity visual detection method | |
| CN111738991A (en) | Method for creating digital ray detection model of weld defects | |
| CN115908301A (en) | Defect detection method and device based on enhanced input and storage medium | |
| WO2023150064A1 (en) | Measure image quality of blood cell images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| DD01 | Delivery of document by public notice |
Addressee: Zheng Yuao Document name: Notice before the expiration of the time limit for the actual review request |
|
| DD01 | Delivery of document by public notice | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| DD01 | Delivery of document by public notice |
Addressee: Zheng Yuao Document name: Notification of Eligibility for Procedures Addressee: Nian Jiawu Document name: Notification of Eligibility for Procedures |
|
| DD01 | Delivery of document by public notice |