CN110517226A - The offal method for extracting region of multiple features texture image fusion based on bilateral filtering - Google Patents
The offal method for extracting region of multiple features texture image fusion based on bilateral filtering Download PDFInfo
- Publication number
- CN110517226A CN110517226A CN201910669112.9A CN201910669112A CN110517226A CN 110517226 A CN110517226 A CN 110517226A CN 201910669112 A CN201910669112 A CN 201910669112A CN 110517226 A CN110517226 A CN 110517226A
- Authority
- CN
- China
- Prior art keywords
- image
- offal
- region
- probability
- bilateral filtering
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000001914 filtration Methods 0.000 title claims abstract description 34
- 230000002146 bilateral effect Effects 0.000 title claims abstract description 22
- 230000004927 fusion Effects 0.000 title claims abstract description 15
- 241000208125 Nicotiana Species 0.000 claims description 35
- 235000002637 Nicotiana tabacum Nutrition 0.000 claims description 35
- 230000011218 segmentation Effects 0.000 claims description 20
- 238000009826 distribution Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 6
- 235000019504 cigarettes Nutrition 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000005530 etching Methods 0.000 claims description 3
- 238000011017 operating method Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 230000000877 morphologic effect Effects 0.000 abstract description 2
- 239000000284 extract Substances 0.000 abstract 2
- 238000005286 illumination Methods 0.000 abstract 1
- 238000001514 detection method Methods 0.000 description 10
- 210000000349 chromosome Anatomy 0.000 description 6
- 238000010009 beating Methods 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 5
- 230000002068 genetic effect Effects 0.000 description 3
- 230000035772 mutation Effects 0.000 description 3
- 239000002699 waste material Substances 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000009472 formulation Methods 0.000 description 2
- 238000000691 measurement method Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- UPLPHRJJTCUQAY-WIRWPRASSA-N 2,3-thioepoxy madol Chemical compound C([C@@H]1CC2)[C@@H]3S[C@@H]3C[C@]1(C)[C@@H]1[C@@H]2[C@@H]2CC[C@](C)(O)[C@@]2(C)CC1 UPLPHRJJTCUQAY-WIRWPRASSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000155 melt Substances 0.000 description 1
- 230000037435 normal mutation Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N5/00—Analysing materials by weighing, e.g. weighing small particles separated from a gas or liquid
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N23/00—Investigating or analysing materials by the use of wave or particle radiation, e.g. X-rays or neutrons, not covered by groups G01N3/00 – G01N17/00, G01N21/00 or G01N22/00
- G01N23/02—Investigating or analysing materials by the use of wave or particle radiation, e.g. X-rays or neutrons, not covered by groups G01N3/00 – G01N17/00, G01N21/00 or G01N22/00 by transmitting the radiation through the material
- G01N23/04—Investigating or analysing materials by the use of wave or particle radiation, e.g. X-rays or neutrons, not covered by groups G01N3/00 – G01N17/00, G01N21/00 or G01N22/00 by transmitting the radiation through the material and forming images of the material
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- 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
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2223/00—Investigating materials by wave or particle radiation
- G01N2223/10—Different kinds of radiation or particles
- G01N2223/101—Different kinds of radiation or particles electromagnetic radiation
- G01N2223/1016—X-ray
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2223/00—Investigating materials by wave or particle radiation
- G01N2223/40—Imaging
- G01N2223/401—Imaging image processing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2223/00—Investigating materials by wave or particle radiation
- G01N2223/60—Specific applications or type of materials
- G01N2223/621—Specific applications or type of materials tobacco
-
- 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
- G06T2207/10052—Images from lightfield camera
-
- 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
- G06T2207/20028—Bilateral filtering
-
- 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
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses the offal method for extracting region of the multiple features texture image fusion based on bilateral filtering, this method it is even from uneven illumination and have very noisy offal x-ray image in extract offal target aiming at the problem that, it proposes based on the probability density image for constructing offal region using three the pixel probability of occurrence of original image, gray level entropy, gray scale textural characteristics weightings, probability density image is filtered using bilateral filtering function again, finally extracts accurate offal region from offal probability density image using mixed Gauss model and morphologic filtering method.
Description
Technical field
The present invention relates to the offal method for extracting region of the multiple features texture image fusion based on bilateral filtering, belong to optics
Detection technique field.
Background technique
During tobacco threshing and redrying, smoked sheet and offal after the leaf that needs to be fought each other using wind point mode are separated, with
The exact formulation of later period tobacco leaf is carried out, cigarette quality is improved.Due to beating in the mixing tobacco leaf after leaf, the smoked sheet containing stalk with dissociate it is pure
The floating velocity difference of net blade is little, the smoked sheet containing stalk can be made to be taken as pure blade during wind point and be difficult to isolate
Come, to make to increase the smoked sheet containing stalk in tobacco formulation, directly affects cigarette quality.Tobacco leaf rate containing stalk after beating leaf is to influence
One of most important factor of chopping quality, and the quality of pipe tobacco directly influences the quality (compactedness, again of finished product cigarette
The important indicators such as amount, density, loose-ends ratio, wire falling rate).Existing beating and double roasting national standard (YC/T 146-2001 and YC/T
It 147--2001) limits respectively in the tobacco leaf before baking and after roasting, rate containing stalk should be less than or be equal to 2.5%, and new standard is modified again
To should be less than or be equal to 2.0%.
According to the above national standard, each beating and double roasting enterprise will detect offal content, most redrying enterprises
Every half an hour it is necessary to detect a hypo-tobacco leaf rate containing stalk, there are mainly two types of the measurement methods of use: first is that, it will be to threshing machine
It surveys tobacco leaf all to smash, then measures rate containing stalk therein again, this measurement method will cause the destruction of tobacco leaf, after measurement
Tobacco leaf will cannot be used continuously, play leaf line at the end of the day and will all waste many tobacco leaves for every.Second is that current most of tobacco factories
It is to go after stalk to weigh to offal and blade respectively by sampling to realize the detection of tobacco leaf rate containing stalk, the detection side with Redrying Factory
Method generates compared with havoc tobacco leaf, and time-consuming, cannot achieve on-line real-time measuremen for detection.Both detection methods exist
Obvious drawback.
Current detection technique also rests in detection rate containing stalk, and there are no any open source literatures and patent can provide inspection
The method for surveying offal region, detection offal region have very important directive significance for actual production.
Summary of the invention
To solve the above problems, the invention discloses a kind of, and the multiple features texture image based on bilateral filtering melts
The offal method for extracting region of conjunction, specific technical solution are as follows:
The offal method for extracting region of multiple features texture image fusion based on bilateral filtering specifically includes following operation step
It is rapid:
Step 1: obtaining tobacco image: X-ray camera is installed above the belt of the outlet end of threshing machine, X-ray camera is taken pictures
The tobacco image transmitted on belt is obtained,
Step 2: obtain textural characteristics: X-ray camera sends image to leaf vision on-line checking software containing stalk, leaf view containing stalk
Feel that on-line checking software carries out gray proces to tobacco image and obtains grayscale image, Threshold segmentation obtains binary map, passes through calculating
Sobel operator calculates the approximate gradient of image grayscale function, and gray level entropy refers to the degree of irregularity or complexity of gray scale in image
Degree, it is assumed that gray value is g in imagejNumber of pixel be kj,
Then gray value is g in imagejPixel occur probability P (gj) are as follows:
Entire image gray level entropy E (I) are as follows:
Obtain gray value gj, gray value gjPixel probability of occurrence P (gj) and three texture spies of entire image gray level entropy E (I)
Sign；
Step 3: obtaining the probability density image in offal region: carrying out linear image fusion using weighting function and obtain finally
Offal region probability density image；
Step 4: filtering: there are the probability density image in the offal region that step 3 obtains many very noisies needs to filter out, and adopt
It is filtered with probability density image of the bilateral filtering to offal region；
Step 5: segmentation offal region and background: can be realized by establishing gauss hybrid models by offal region and background
Accurately segmentation, the offal area image after being divided；
Step 6: the small-sized cavity of the offal area image after removal segmentation: after closing operation of mathematical morphology removal segmentation
The small-sized cavity of offal area image, smooth objects' contour and fills up profile thread breakage, finally obtains offal region.
Further, the bilateral filtering specific steps of the step 4 are as follows:
Effect while bilateral filtering is really two gaussian filterings, the weight of a calculating spatial neighbor degree, another
The weight for being responsible for calculating pixel value similarity passes through the space proximity of each point to central point in the principle of gaussian filtering
The each weight calculated optimizes, and is optimized for the weight of space proximity calculating and the power of pixel value similarity calculation
The product of value, the weight after optimization make convolution algorithm with image again, thus achieve the effect that protect side denoising,
Its specific formula are as follows:
Wherein, g (i, j) represents output point, and S (i, j) refers to the model of the size of (2N+1) (2N+1) centered on (i, j)
It enclosing, f (k, l) represents multiple input points, and w (i, j, k, l), which is represented, passes through two calculated values of Gaussian function,
Assuming that w (i, j, k, l)=m in formula, then have:
If+the mn=M m1+m2+m3 ... then has:
It can be seen that, this is obviously the convolution algorithm of image array and core at this time, first point that wherein m1/M is represented
Weight, and image array and core make weighted sum by convolution operator, finally obtain output valve,
And w (i, j, k, l)=ws*wr, wherein
σ is respectively the standard deviation of current pixel information and current pixel position.
Further, gauss hybrid models are established in the step 5 by the detailed process in offal region and background segment are as follows:
The probability value that each pixel belongs to some in K Gauss model is calculated using gauss hybrid models, utilizes this probability value
Original pixel value is replaced to achieve the purpose that segmented image, gauss hybrid models refer to the probability distribution mould with following form
Type:
Wherein, αkIt is coefficient,φ(y|θk) it is Gaussian distribution density,
Referred to as k-th of sub-model,
Then, histogram-fitting is carried out to the offal area image after segmentation, i.e., constantly adjusts the parameter of Gauss model.
Further, histogram-fitting is carried out to the offal area image after segmentation in the step 5, i.e., constantly adjustment is high
The parameter of this model, detailed process are as follows:
(4) initial value of parameter is taken to start iteration
(5) E is walked: according to "current" model parameter, calculating sub-model k to observation data yjResponsiveness
(6) M is walked: calculating the model parameter of new round iteration
(4) (2) step and (3) step are repeated, until convergence.
Further, the operation of closing operation of mathematical morphology is first to expand post-etching in the step 6, i.e. dst=close
(src, element)=erode (dilate (src, element)).
The beneficial effects of the present invention are:
The present invention can provide the specific region of offal, can be straight targeted specifically when the rate containing stalk that detects is exceeded
It sees to know offal position, directly finds offal, there is very important operation instruction meaning.
In actual production, it is used simultaneously in conjunction with the existing online detection instrument of rate containing stalk, when the rate containing stalk that detects is exceeded
When, production scene directly obtains the position of offal, and the tobacco in offal region is directly grabbed by manipulator, plays and targetedly grasps
Make, operated on against lesion, avoid the operation that a traditional basket is all done over again, improve working efficiency, avoids repeating to beat leaf,
It repeats that tiny tobacco leaf is to beat excessively fine crushing the drawbacks of beating leaf, can also reduce the quality of cigarette, or in screening process
In, tiny tobacco leaf is blown away by wind, increases the waste of tobacco leaf, increases manufacturing procedure, process time, equipment use.And pass through this
The technology of patent can directly be directed at offal region online, grab offal, reduce many processes, in actual production, greatly
Processing cost and tobacco waste are reduced greatly, are avoided tobacco leaf from being repeated and are beaten leaf, increase underproof excessively fine tobacco content.
Specific embodiment
With reference to embodiment, the present invention is furture elucidated.It should be understood that following specific embodiments are only used for
It is bright the present invention rather than limit the scope of the invention.
The image method tobacco based on multi-feature fusion detection method of rate containing stalk of the invention, including following operating procedure:
The present invention is based on the offal method for extracting region of the multiple features texture image of bilateral filtering fusion, specifically include following
Operating procedure: the strong noise x-ray image offal extracted region of the multiple features texture image fusion based on bilateral filtering utilizes original
The probability density image in three textural characteristics weighting building offal regions of highlighted dot density, gray level entropy, gray scale of image, then adopt
Probability density image is filtered with bilateral filtering function, finally using mixed Gauss model and morphologic filtering method from
Accurate offal region, detailed process are extracted in offal probability density image are as follows:
Step 1: obtaining tobacco image: X-ray camera is installed above the belt of the outlet end of threshing machine, X-ray camera is taken pictures
Obtain the tobacco image transmitted on belt；
Step 2: obtain textural characteristics: X-ray camera sends image to leaf vision on-line checking software containing stalk, and leaf is containing stalk
Vision on-line checking software carries out gray proces to tobacco image and obtains grayscale image, and Threshold segmentation obtains binary map, passes through calculating
Sobel operator calculates the approximate gradient of image grayscale function, and gray level entropy refers to the degree of irregularity or complexity of gray scale in image
Degree, it is assumed that gray value is g in imagejNumber of pixel be kj, then gray value is g in imagejPixel occur probability P (gj)
Are as follows:Entire image gray level entropy E (I) are as follows:Obtain gray value
gj, gray value gjPixel probability of occurrence P (gj) and three textural characteristics of entire image gray level entropy E (I)；
Step 3: obtaining the probability density image in offal region: carrying out linear image fusion using weighting function and obtain most
The probability density image in whole offal region；
Step 4: filtering: there are the probability density image in the offal region that step 3 obtains many very noisies needs to filter out, and adopt
It is filtered with probability density image of the bilateral filtering to offal region, detailed process are as follows:
Effect while bilateral filtering is really two gaussian filterings, the weight of a calculating spatial neighbor degree, another
The weight for being responsible for calculating pixel value similarity passes through the space proximity of each point to central point in the principle of gaussian filtering
The each weight calculated optimizes, and is optimized for the weight of space proximity calculating and the power of pixel value similarity calculation
The product of value, the weight after optimization make convolution algorithm with image again, thus achieve the effect that protect side denoising,
Its specific formula are as follows:
Wherein, g (i, j) represents output point, and S (i, j) refers to the model of the size of (2N+1) (2N+1) centered on (i, j)
It enclosing, f (k, l) represents multiple input points, and w (i, j, k, l), which is represented, passes through two calculated values of Gaussian function,
Assuming that w (i, j, k, l)=m in formula, then have:
If+the mn=M m1+m2+m3 ... then has:
It can be seen that, this is obviously the convolution algorithm of image array and core at this time, first point that wherein m1/M is represented
Weight, and image array and core make weighted sum by convolution operator, finally obtain output valve,
And w (i, j, k, l)=ws*wr, wherein
σ is respectively the standard deviation of current pixel information and current pixel position.
Step 5: segmentation offal region and background: can be realized by establishing gauss hybrid models by offal region and back
Scape is accurately divided, the offal area image after being divided, wherein establishing gauss hybrid models for offal region and background point
The detailed process cut are as follows: calculate the probability that each pixel belongs to some in K Gauss model using gauss hybrid models
Value replaces original pixel value to achieve the purpose that segmented image using this probability value, and gauss hybrid models refer to as follows
The probability Distribution Model of form:
Wherein, αkIt is coefficient,φ(y|θk) it is Gaussian distribution density, Then referred to as k-th of sub-model carries out histogram to the offal area image after segmentation
Fitting constantly adjusts the parameter of Gauss model.
Histogram-fitting is carried out to the offal area image after segmentation, i.e., constantly adjusts the parameter of Gauss model, it is specific to flow
Journey are as follows:
(1) initial value of parameter is taken to start iteration
(2) E is walked: according to "current" model parameter, calculating sub-model k to observation data yjResponsiveness
(3) M is walked: calculating the model parameter of new round iteration
(4) (2) step and (3) step are repeated, until convergence.
Step 6: the small-sized cavity of the offal area image after removal segmentation: after closing operation of mathematical morphology removal segmentation
The small-sized cavity of offal area image, smooth objects' contour and fills up profile thread breakage, finally obtains offal region,
Wherein the operation of closing operation of mathematical morphology is first to expand post-etching, i.e. dst=close (src, element)=erode (dilate
(src,element))。
After obtaining offal region by the above method, rate containing stalk is calculated by following methods: by obtaining after image procossing
Tobacco leaf picture using fitting algorithm go calculate rate containing stalk, detailed process are as follows:
Step (1): it obtains target data set: calculating a series of containing for tobacco finished product samples with traditional weight method
Stalk rate, using these values as target data set；
Step (2): it obtains input set: step 1 is calculated with a collection of tobacco finished product sample with the image method of software
Rate containing stalk out, using these values as the input set for needing fitting function,
Step (3): the optimal parameter of nonlinear fitting function is found out with Revised genetic algorithum, that is, meets genetic algorithm
Coefficient vector A=[a of end loop condition1, a2..., an] T, wherein genetic algorithm is applied to Cubic Spline Functions Fitting
Basic ideas are exactly based on operation GA and find coefficient vector A=[a1, a2..., an] optimum value of first element a1 in T, then
Successively find out remaining coefficient a2…an, so the parameter to be encoded in GA only has a1, it is located at entire curve fit interval [a, b]
On the corresponding evaluated error of the i-th generation j-th strip chromosome be eij, it can thus define the fitness function of the chromosome:
The specific steps of GA are as follows:
(1) parameter, including population size N, chromosome length L, crossover probability P are setc, normal variant probability Pmnor, maximum
Mutation probability Pmmax, run algebra Ge, run algebra Gt, estimated accuracy W enables i=0, k=0, m=0；
(2) N number of chromosome is initialized, i=i+1, k=0, m=0 are enabled；
(3) it decodes chromosome and evaluated error e is calculated to j-th strip chromosome in the i-th generationij, enable
(4) N number of male parent is selected using roulette wheel method, retains 1 optimal male parent and is not involved in cross and variation；
(5) pairing intersects, using two-point crossover method, with probability P in crossover processmMutation operation is called, if Pm=
Pmmax, restore normal mutation probability P after mutation operation completionm=Pmmor, the optimal male parent that step (4) are retained is randomly
Replace any of N number of daughter；
(6) ifOtherwise k=0, m=0；
(7) if N number of male parent is identical or k=Ge, make mutation probability Pm=Pmmax, enable k=0；
If m=Gt(or), then terminating algorithm, otherwise go to step (3).
For the practical application value for verifying this patent, chooses 5 groups of tobacco samples and tested, biography is respectively adopted in every group of sample
The weight method of system and the image method of this patent calculate rate containing stalk, and calculate the relative error of the two, test institute in below table
Record and the data calculated.
From the data in table above it may be concluded that the rate containing stalk that the image method of this patent calculates is obtained with weight method
Value relative error within 10%, be more accurate.
Taking the above-mentioned ideal embodiment according to the present invention as inspiration, through the above description, relevant staff is complete
Various changes and amendments can be carried out without departing from the scope of the technological thought of the present invention' entirely.The technology of this invention
Property range is not limited to the contents of the specification, it is necessary to which the technical scope thereof is determined according to the scope of the claim.
Claims (5)
1. the offal method for extracting region that the multiple features texture image based on bilateral filtering merges, which is characterized in that specifically include
Following operating procedure:
Step 1: obtaining tobacco image: X-ray camera is installed above the belt of the outlet end of threshing machine, X-ray camera is taken pictures acquisition
The tobacco image transmitted on belt,
Step 2: obtain textural characteristics: X-ray camera sends image to leaf vision on-line checking software containing stalk, and leaf vision containing stalk exists
Line inspection software carries out gray proces to tobacco image and obtains grayscale image, and Threshold segmentation obtains binary map, is calculated by calculating sobel
Son calculates the approximate gradient of image grayscale function, and gray level entropy refers to the degree of irregularity or complexity of gray scale in image, it is assumed that
Gray value is g in imagejNumber of pixel be kj,
Then gray value is g in imagejPixel occur probability P (gj) are as follows:
Entire image gray level entropy E (I) are as follows:
Obtain gray value gj, gray value gjPixel probability of occurrence P (gj) and three textural characteristics of entire image gray level entropy E (I)；
Step 3: obtaining the probability density image in offal region: carrying out linear image fusion using weighting function and obtain final cigarette
Obstruct the probability density image in region；
Step 4: filtering: there are the probability density image in the offal region that step 3 obtains many very noisies needs to filter out, using double
Side filtering is filtered the probability density image in offal region；
Step 5: segmentation offal region and background: by establish gauss hybrid models can be realized offal region and background is accurate
Segmentation, the offal area image after being divided；
Step 6: the small-sized cavity of the offal area image after removal segmentation: passing through the offal after closing operation of mathematical morphology removal segmentation
The small-sized cavity of area image, smooth objects' contour and fills up profile thread breakage, finally obtains offal region.
2. the offal method for extracting region of the multiple features texture image fusion according to claim 1 based on bilateral filtering,
It is characterized in that the bilateral filtering specific steps of the step 4 are as follows:
Effect while bilateral filtering is really two gaussian filterings, the weight of a calculating spatial neighbor degree, another is responsible for
The weight for calculating pixel value similarity is calculated in the principle of gaussian filtering by the space proximity of each point to central point
Each weight optimize, the weight of the weight and pixel value similarity calculation that are optimized for the calculating of space proximity multiplies
Product, the weight after optimization make convolution algorithm with image again, thus achieve the effect that protect side denoising,
Its specific formula are as follows:
Wherein, g (i, j) represents output point, and S (i, j) refers to the range of the size of (2N+1) (2N+1) centered on (i, j), f
(k, l) represents multiple input points, and w (i, j, k, l), which is represented, passes through two calculated values of Gaussian function,
Assuming that w (i, j, k, l)=m in formula, then have:
If+the mn=M m1+m2+m3 ... then has:
It can be seen that, this is obviously the convolution algorithm of image array and core at this time, the power for first point that wherein m1/M is represented
Value, and image array and core make weighted sum by convolution operator, finally obtain output valve,
And w (i, j, k, l)=ws*wr, wherein
σ is respectively the standard deviation of current pixel information and current pixel position.
3. the offal method for extracting region of the multiple features texture image fusion according to claim 1 based on bilateral filtering,
It is characterized in that establishing gauss hybrid models in the step 5 for the detailed process in offal region and background segment are as follows: utilize height
This mixed model calculates the probability value that each pixel belongs to some in K Gauss model, is replaced using this probability value former
The pixel value come achievees the purpose that segmented image, and gauss hybrid models refer to the probability Distribution Model with following form:
Wherein, αkIt is coefficient,φ(y|θk) it is Gaussian distribution density,
Referred to as k-th of sub-model,
Then, histogram-fitting is carried out to the offal area image after segmentation, i.e., constantly adjusts the parameter of Gauss model.
4. the offal method for extracting region of the multiple features texture image fusion according to claim 1 based on bilateral filtering,
It is characterized in that carrying out histogram-fitting to the offal area image after segmentation in the step 5, i.e., constantly adjust Gauss model
Parameter, detailed process are as follows:
(1) initial value of parameter is taken to start iteration
(2) E is walked: according to "current" model parameter, calculating sub-model k to observation data yjResponsiveness
(3) M is walked: calculating the model parameter of new round iteration
(4) (2) step and (3) step are repeated, until convergence.
5. the offal method for extracting region of the multiple features texture image fusion according to claim 1 based on bilateral filtering,
It is characterized in that the operation of closing operation of mathematical morphology is first to expand post-etching in the step 6, i.e.,
Dst=close (src, element)=erode (dilate (src, element)).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910669112.9A CN110517226B (en) | 2019-07-24 | 2019-07-24 | Tobacco stem region extraction method based on bilateral filtering and multi-feature texture image fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910669112.9A CN110517226B (en) | 2019-07-24 | 2019-07-24 | Tobacco stem region extraction method based on bilateral filtering and multi-feature texture image fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110517226A true CN110517226A (en) | 2019-11-29 |
| CN110517226B CN110517226B (en) | 2021-04-09 |
Family
ID=68623388
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910669112.9A Active CN110517226B (en) | 2019-07-24 | 2019-07-24 | Tobacco stem region extraction method based on bilateral filtering and multi-feature texture image fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110517226B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113805015A (en) * | 2021-08-06 | 2021-12-17 | 云南电网有限责任公司德宏供电局 | Arc image form detection method for multi-cavity arc extinguishing device |
| CN114067314A (en) * | 2022-01-17 | 2022-02-18 | 泗水县锦川花生食品有限公司 | Neural network-based peanut mildew identification method and system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104951795A (en) * | 2015-05-26 | 2015-09-30 | 重庆金山科技（集团）有限公司 | Image classification identifying and judging method |
| CN105654088A (en) * | 2016-03-07 | 2016-06-08 | 西安交通大学 | Tobacco stem extracting method based on local symmetric relation |
| US20160314567A1 (en) * | 2015-03-26 | 2016-10-27 | The Board Of Regents Of The University Of Texas System | Systems and methods for image/video recoloring, color standardization, and multimedia analytics |
| 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 |
| US20170278246A1 (en) * | 2016-03-22 | 2017-09-28 | Electronics And Telecommunications Research Institute | Apparatus and method for extracting object |
| CN109035252A (en) * | 2018-06-29 | 2018-12-18 | 山东财经大学 | A kind of super-pixel method towards medical image segmentation |
| CN109934802A (en) * | 2019-02-02 | 2019-06-25 | 浙江工业大学 | A kind of Fabric Defects Inspection detection method based on Fourier transformation and morphological image |
-
2019
- 2019-07-24 CN CN201910669112.9A patent/CN110517226B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160314567A1 (en) * | 2015-03-26 | 2016-10-27 | The Board Of Regents Of The University Of Texas System | Systems and methods for image/video recoloring, color standardization, and multimedia analytics |
| CN104951795A (en) * | 2015-05-26 | 2015-09-30 | 重庆金山科技（集团）有限公司 | Image classification identifying and judging method |
| CN105654088A (en) * | 2016-03-07 | 2016-06-08 | 西安交通大学 | Tobacco stem extracting method based on local symmetric relation |
| US20170278246A1 (en) * | 2016-03-22 | 2017-09-28 | Electronics And Telecommunications Research Institute | Apparatus and method for extracting object |
| 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 |
| CN109035252A (en) * | 2018-06-29 | 2018-12-18 | 山东财经大学 | A kind of super-pixel method towards medical image segmentation |
| CN109934802A (en) * | 2019-02-02 | 2019-06-25 | 浙江工业大学 | A kind of Fabric Defects Inspection detection method based on Fourier transformation and morphological image |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113805015A (en) * | 2021-08-06 | 2021-12-17 | 云南电网有限责任公司德宏供电局 | Arc image form detection method for multi-cavity arc extinguishing device |
| CN114067314A (en) * | 2022-01-17 | 2022-02-18 | 泗水县锦川花生食品有限公司 | Neural network-based peanut mildew identification method and system |
| CN114067314B (en) * | 2022-01-17 | 2022-04-26 | 泗水县锦川花生食品有限公司 | Neural network-based peanut mildew identification method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110517226B (en) | 2021-04-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110476839B (en) | Optimal regulation and control method and system based on fish growth | |
| CN110349153B (en) | Image method tobacco peduncle content detection method based on multi-feature fusion | |
| CN109308697B (en) | Leaf disease identification method based on machine learning algorithm | |
| CN110009013A (en) | Encoder training and characterization information extracting method and device | |
| CN102495005B (en) | Method for diagnosing crop water deficit through hyperspectral image technology | |
| CN103971342B (en) | A kind of image noise detection method based on convolutional neural networks | |
| CN109239082A (en) | Tobacco structure quality online test method and system based on machine vision technique | |
| CN110705655A (en) | Tobacco leaf classification method based on coupling of spectrum and machine vision | |
| CN110849828A (en) | Saffron crocus classification method based on hyperspectral image technology | |
| CN110517226A (en) | The offal method for extracting region of multiple features texture image fusion based on bilateral filtering | |
| CN109214298A (en) | A kind of Asia women face value Rating Model method based on depth convolutional network | |
| CN108550163A (en) | Moving target detecting method in a kind of complex background scene | |
| CN107590512A (en) | The adaptive approach and system of parameter in a kind of template matches | |
| CN116452594A (en) | Visualized monitoring and early warning method and system for power transmission line state | |
| CN111695730A (en) | ARIMA and RNN-based vertical mill vibration prediction method and device | |
| CN110427019B (en) | Industrial process fault classification method and control device based on multivariate discriminant analysis | |
| CN104933410B (en) | A kind of high spectrum image spectral domain and spatial domain joint classification method | |
| CN115471724A (en) | Fine-grained fish epidemic disease identification fusion algorithm based on self-adaptive normalization | |
| CN113077002B (en) | Machine olfaction visual sensing data analysis method based on space heterodyne Raman spectrum | |
| CN114511850A (en) | Method for identifying image of fruit size and granule of sunshine rose grape | |
| JP2019191854A (en) | Image recognition device, artificial pollination system, and program | |
| CN110610287B (en) | Genetic algorithm-based image-method stem content nonlinear fitting model calculation method | |
| CN115330868A (en) | Grape picking method based on deep learning and depth information fusion | |
| CN112668754A (en) | Power equipment defect diagnosis method based on multi-source characteristic information fusion | |
| CN112907503A (en) | Penaeus vannamei Boone quality detection method based on adaptive convolutional neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |