CN104155312B - Grain intragranular portion pest detection method and apparatus based on near-infrared computer vision - Google Patents
Grain intragranular portion pest detection method and apparatus based on near-infrared computer vision Download PDFInfo
- Publication number
- CN104155312B CN104155312B CN201410404698.3A CN201410404698A CN104155312B CN 104155312 B CN104155312 B CN 104155312B CN 201410404698 A CN201410404698 A CN 201410404698A CN 104155312 B CN104155312 B CN 104155312B
- Authority
- CN
- China
- Prior art keywords
- grain
- image
- infrared
- computer vision
- collecting cassette
- 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.)
- Expired - Fee Related
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 41
- 241000607479 Yersinia pestis Species 0.000 title abstract description 16
- 235000013339 cereals Nutrition 0.000 claims abstract description 487
- 238000000034 method Methods 0.000 claims abstract description 50
- 241000238631 Hexapoda Species 0.000 claims abstract description 45
- 238000005070 sampling Methods 0.000 claims abstract description 20
- 230000008569 process Effects 0.000 claims abstract description 19
- 230000003287 optical effect Effects 0.000 claims abstract description 15
- 230000000266 injurious effect Effects 0.000 claims abstract 11
- 230000005540 biological transmission Effects 0.000 claims description 17
- 238000005457 optimization Methods 0.000 claims description 11
- 239000000284 extract Substances 0.000 claims description 9
- 235000013305 food Nutrition 0.000 claims description 8
- 230000008676 import Effects 0.000 claims description 6
- 230000005484 gravity Effects 0.000 claims description 5
- 230000011218 segmentation Effects 0.000 claims description 5
- 239000011800 void material Substances 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000005452 bending Methods 0.000 claims description 4
- 238000009825 accumulation Methods 0.000 claims description 3
- 238000013144 data compression Methods 0.000 claims description 3
- 238000003709 image segmentation Methods 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 3
- 238000003909 pattern recognition Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 230000003595 spectral effect Effects 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 2
- 238000001914 filtration Methods 0.000 claims description 2
- 230000009471 action Effects 0.000 claims 1
- 238000010586 diagram Methods 0.000 claims 1
- 238000009434 installation Methods 0.000 claims 1
- 230000006870 function Effects 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 6
- 241000209140 Triticum Species 0.000 description 4
- 235000021307 Triticum Nutrition 0.000 description 4
- 238000003333 near-infrared imaging Methods 0.000 description 4
- 238000004497 NIR spectroscopy Methods 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 238000004140 cleaning Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000004069 differentiation Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 238000003756 stirring Methods 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000001931 thermography Methods 0.000 description 2
- 230000004580 weight loss Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 101100117236 Drosophila melanogaster speck gene Proteins 0.000 description 1
- 241000233866 Fungi Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 206010036590 Premature baby Diseases 0.000 description 1
- 241000318997 Rhyzopertha dominica Species 0.000 description 1
- 241000254179 Sitophilus granarius Species 0.000 description 1
- 241000254152 Sitophilus oryzae Species 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 239000002551 biofuel Substances 0.000 description 1
- 235000008429 bread Nutrition 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 210000003608 fece Anatomy 0.000 description 1
- 239000012634 fragment Substances 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 208000015181 infectious disease Diseases 0.000 description 1
- 238000003331 infrared imaging Methods 0.000 description 1
- 230000013011 mating Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 239000000725 suspension Substances 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Landscapes
- Investigating Or Analysing Materials By Optical Means (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention discloses a kind of grain intragranular portion injurious insect detector based on near-infrared computer vision, the detection means is made up of grain grain sampling part, grain grain hop and computer vision process part；Described grain grain sampling part makes quantitative grain grain quickly and efficiently be separated from feeder using the driving row's grain wheel rotation of grain motor is arranged；Grain grain hop can convey collecting cassette to specified location, and the grain grain that can be cleared up automatically in collecting cassette；Computer vision process part is made up of near infrared camera, light source, optical filter runner, lighting box, computer and corresponding software, after carrying out multi-wavelength near-infrared image collection to the grain grain in collecting cassette, whether attitude and the grain grain that grain of putting out cereal is automatically determined by identification model are infected by insect.The invention also discloses the corresponding detection method of grain intragranular portion injurious insect detector based on near-infrared computer vision.The present invention realizes real-time, the accurate, automatic detection of grain intragranular portion's insect pest.
Description
Technical field
The invention discloses a kind of grain intragranular portion pest detection method and apparatus based on near-infrared computer vision, specifically
It is related to the grain intragranular portion pest detection method based on near-infrared hyper-spectral image technique
Background technology
" bread is the staff of life ", " food is with grain as first ", grain security is related to national economy, of prosperous economy and social development.
However, Disastrous climate takes place frequently in recent years, bio-fuel strives grain, and world food yield deposit declines, and grain security situation is increasingly
It is severe.In addition, with the improvement of people ' s living standards, requirement higher is proposed to grain quality.According to data, the whole world
Grain weight loss after results is about 10-15%, and the grain at least 5% during annual storage is ravaged by insect.China
《Grain and oil storage technology specification》Clear stipulaties, pest density statistics in worm grain classification standard is the quantity of worm of living, borer population living
It is the inside and outside borer population sum living of grain grain to measure.Therefore, the detection of grain worm includes the outside insect of grain grain and grain intragranular portion insect
Detection.
The boring pests such as rice weevil, lesser grain borer directly endanger complete grain grain, and its prematurity larva grows in grain intragranular portion,
Until adult just from grain intragranular out, after spawning it is a few week in be difficult visually to be found, it is difficult to from external visualization detect.Evil
Worm not only causes the weight loss of grain, and its excreta, corpse, cot etc. can also produce serious pollution to grain, easily increase
Plus the infection of fungi, cause the decline of grain quality.It can be seen that, the harm that grain intragranular portion insect is caused is sizable, and detection
It is relatively difficult.Therefore, how early detection, and accurately, the insect in automatic detection grain intragranular portion be very important.
Traditional grain intragranular portion pest detection method has artificial organoleptic examination method, insect fragment method of inspection, suspension method, conductance
Rate method etc..These methods have one or more shortcomings, it is such as too subjective, complicated it is cumbersome, inaccurate, time-consuming, with destructiveness.Closely
Nian Lai, occurs in that some new detection methods, such as Electronic Nose method, near infrared spectroscopy, computer vision method.Electronic Nose method is to inspection
The seal requirement for surveying sample container is higher, the overlong time that sample prepares and samples.Near infrared spectroscopy is for infecting level
Than relatively low situation, the quantization to the grain of grain containing worm is less reliable, it is impossible to which differentiate single grain intragranular portion insect infects situation.
Computer vision method detection grain intragranular portion insect, comprising thermal imaging method, grenz ray imaging method, near infrared imaging method
Deng.Thermal imaging method detection process is excessively cumbersome, and sample time is long.Grenz ray imaging method can determine that early stage grain worm is infected
Grain grain, due to safety problem and cost, have impact on its practical application.
Near infrared imaging method compared with other imaging systems, all has at the aspect such as hardware costs, simple, compact, security
There is certain advantage, be the grain intragranular portion very promising developing direction of pest detection, researcher is studied in this respect,
Make some progress.Ridgway etc. uses the difference image artificial visual of the 1202nm and 1300nm wavelength of selection by hand
Pestinfestation in detection wheat, but without research automatic discrimination algorithm.Ridgway etc. goes out with selected by near infrared spectroscopy
Wavelength, the near-infrared grain weevil that have developed 981nm infects detecting system, and the differentiation rate of the method is detected better than artificial visual, but grain
Influence of the extraction effect of intragranular portion speck to infecting detection is larger, and detection efficiency awaits further raising.Singh etc. is to wheat
Grain carries out high spectrum image collection, and the statistical picture that wheat is extracted under two sensitive wave lengths of 1101.69nm and 1305.05nm is special
Levy, whether distinguish in wheat has the accuracy of worm more than 85%, but the attitude of grain grain is placed for manual, and only have studied grain grain
Ventral groove upward with ventral groove two kinds of attitudes down.
Near infrared imaging method in computer vision method can realize the detection of grain intragranular portion insect, be grain intragranular portion insect
The direction of development is detected, but current near-infrared computer vision system China Oil and Food Import and Export Corporation grain is artificial sampling mode, cannot also be obtained automatically
Detection grain grain is taken, it is necessary to improve the automaticity of detection, whether grain intragranular portion has the discrimination precision of worm also to need further raising.
Therefore, it is necessary to study grain intragranular portion insect automatic testing method and its device, automatic, the accurate inspection of grain intragranular portion insect is realized
Survey.
The content of the invention
The technical problems to be solved by the invention are：For the defect of prior art, there is provided a kind of to be calculated based on near-infrared
The grain intragranular portion insect automatic testing method and its device of machine vision, can automatically take out part grain grain from grain sample, and certainly
Dynamic to be transferred to computer vision system and carry out IMAQ, whether attitude and the grain grain that grain of putting out cereal is automatically determined by identification model are received
To infecting for insect, real-time, the accurate, automatic detection of grain intragranular portion's insect pest is realized.
The present invention uses following technical scheme to solve above-mentioned technical problem：
Grain intragranular portion insect automatic detection device based on near-infrared computer vision of the present invention is sampled by grain grain
Partly, grain grain hop and computer vision process part composition.Grain grain sampling part and computer vision process part are all
Above grain grain hop.
The function that described grain grain sampling part is realized is quantitative grain grain is quickly and efficiently isolated from feeder
Come, mainly including feeder, row's grain wheel, row grain motor and clear grain brush.Existing grain intragranular portion insect pest computer vision system is all
Do not possess automatic ration sampled functions, grain grain sampling part of the invention is directed to grain intragranular portion's insect pest computer vision and examines automatically
Survey and design.Feeder is fixed on device frame, the surface in row's grain wheel.The circumferential surface for arranging grain wheel is uniform-distribution with
4 grooves of flat trapeze, by single belt and row's grain motor connection, height of the difference in height slightly larger than collecting cassette with conveyer belt 9
Degree.Clear grain brush is fixed together with feeder, and the Way out with grain grain is consistent, and between the outer surface holding necessarily of row's grain wheel
Gap.Therefore, after grain sample pours into feeder, quantitative is carried out by the above-mentioned motor-driven row's grain wheel of row's grain automatically
Sample.
The function that described grain grain hop is realized is accurately to receive grain grain, conveying the first collecting cassette and the second collecting cassette
Underface to computer vision process part is for IMAQ, and the grain grain in 2 collecting cassettes, main to include control mould
Block, transmission electrical machine, disposable box, conveyer belt, the first collecting cassette, the second collecting cassette, the first photoelectric sensor and the second photoelectric sensing
Device.Control module is by data wire and the first photoelectric sensor, the second photoelectric sensor, row's grain motor, computer and transmission
Motor is connected, and computer receives the trigger signal from the first photoelectric sensor and the second photoelectric sensor by it, and passes through
It sends control instruction to row's grain motor and transmission electrical machine.Conveyer belt uses chain Chain conveyer transmission mechanism, by transmitting electricity
Machine drives operating.First collecting cassette and the second collecting cassette are uniformly distributed on chain, and 2 collecting cassette front ends are hinged on chain
On, rear end is free end, can flexible rotating.2 photoelectric sensors are separately mounted under the outlet of row's grain wheel, near infrared camera
Side, the inboard of conveyer belt, and the slightly less than position of collecting cassette height.When they detect 2 collecting cassettes move to row grain take turns out
When mouth, the underface of near infrared camera, the trigger signal that the sampling of grain grain and multi-wavelength image shoot successively is provided to control module.
Disposable box is used for the grain grain in collecting cassette, and the underface on the right side of conveyer belt should be greater than collection apart from the difference in height of conveyer belt
Height when box stands upside down, collecting cassette is performed button and is stirred after work topples over grain grain can continue to run forward.
Described computer vision process part, major function is the multi-wavelength near-infrared image for gathering grain grain, by near red
Outer camera, optical filter runner, near-infrared light source, lighting box, computer and corresponding software composition.Lighting box is arranged on conveyer belt
There is the light source of the near infrared band of uniform illumination surface, the inside, is uniformly diffused for the grain grain in collecting cassette is provided.Closely
Infrared camera is arranged in lighting box, is vertically mounted on the surface of conveyer belt, with the collecting cassette direction of motion in the same plane,
Apart from the height of conveyer belt 9 the collecting cassette bottom surface effective range for collecting should be made as big as possible, and clear shooting, collecting can be distinguished
The multi-wavelength near-infrared image of box China Oil and Food Import and Export Corporation grain.The spectral region of near infrared camera is 900-1700nm, and its front end carries near-infrared
Optical filter runner.Near infrared camera is connected through data wire with computer, and the image transmitting that will be shot is to computer.Optical filter turns
Wheel the inside is provided with the near infrared filter of several different wave lengths, and the half-band width of near infrared filter is respectively less than 10nm.Optical filter
Runner is connected through data wire with computer, the control information from computer is received, to control during multi-wavelength image automatic data collection
The automatic switchover of several optical filters.
Examined automatically suitable for the grain intragranular portion insect based on near-infrared computer vision the invention also discloses a kind of
The detection method of device is surveyed, is comprised the following steps that：
(1) grain grain sampling part is automatic extracts quantitative grain grain from feed appliance, and the grain grain is automatically fallen into collecting cassette.
(2) grain grain hop is sent to collecting cassette the underface of near infrared camera, computer vision process part pair
Grain grain in collecting cassette carries out multi-wavelength near-infrared image collection, and transmits to computer.
(3) the first wave length image to grain grain carries out image procossing, according to target area in the picture after image segmentation
Information, accurately determines effective target (grain grain), records coordinate, the center of gravity of all grain grain targets in first wave length image etc.
Information, is partitioned into single complete grain grain, obtains the completely black grayscale sub-image for only containing single grain grain of background.
(4) for the grayscale sub-image of single grain grain under first wave length, by asking for grain intragranular contouring and Threshold segmentation
The difference of bianry image afterwards, according to target in difference image and the difference of the ultimate range on grain grain border, can automatic discrimination go out this
Whether target is border or inner void.If grain grain target has border hole or inner void, grain grain is directly judged to
Not Wei pestinfestation grain grain, if grain grain target without hole, into the poses discrimination of next step.
(5) for the grayscale sub-image of single grain grain under first wave length, invariant moment features (such as not bending moment 1, not bending moment are extracted
2nd ..., not bending moment 7 etc.), along the absolute value etc. of grain grain length axle two ends area invariant moment features difference；Placed an order for first wave length
Individual grain grain bianry image, extracts the spies such as the nearest equivalent distances on flexibility, circle, major axis and its minimum enclosed rectangle side long
Levy, form the original feature space of grain grain attitude.Using data compression technique, such as principal component analysis, Fisher discriminant analyses,
To grain grain original feature space Dimension Reduction Analysis, characteristic parameter information of the accumulation contribution rate higher than 90% is extracted, form grain grain attitude
Optimization feature space.
(6) for the optimization feature space of grain grain attitude, using nonlinear pattern recognition sorting technique, such as neutral net point
Class device, support vector machine classifier, Fuzzy Classifier, extension classification device etc., set up spy after the grain optimization of first wave length image China Oil and Food Import and Export Corporation
The relational model of parameter and grain grain attitude is levied, grain grain attitude common are ventral groove down, upward, laterally, and Model Identification precision
More than 95%, the attitude of each grain grain is automatically determined out by grain grain gesture recognition model.
(7) for the grain grain gray level image under multi-wavelength, extract respectively its statistical nature (such as maximum, minimum value, in
Value, standard deviation etc.), histogram transform characteristics, textural characteristics etc., the primitive character for being formed and being optimized under different grain grain attitudes is empty
Between.According to the grain grain of different attitudes, using intelligent mode sorting technique, grain grain image after optimizing in multi-wavelength image is set up respectively
Whether characteristic parameter is subject to the relational model of pestinfestation with grain grain, and ensures 3 disaggregated model precision of attitude 95%
More than, can determine that whether grain of putting out cereal is infected by insect with grain intragranular portion insect identification model.
(8) the grain grain number amount of pestinfestation is subject in the multi-wavelength image acquired in counting, if in the presence of grain grain is infected, remembering
Record the information such as the multi-wavelength image, the center of gravity for infecting grain grain.
(9) when computer is analyzed and processed to the multi-wavelength near-infrared image for obtaining, grain grain hop is continued to run with, when
Run to the grain grain in automatic cleaning removal collecting cassette during extreme position.
Further, described image procossing includes removal collecting cassette background, filtering enhancing image, segmentation figure picture, grain grain
It is accurately positioned.
The intensity of light source and position in lighting box can be adjusted, and choose light source and the locus of suitable spectrum, and make to take the photograph
As forming uniform illumination in the visual field of head, camera is set to obtain clearly screenings image.
The present invention uses above technical scheme compared with prior art, with following technique effect：
(1) present invention uses near-infrared computer vision system to detect grain grain, automatic extraction part grain grain, and according to
According to the different attitudes of grain grain automatic discrimination is carried out to whether grain grain is infected by insect；
(2) present invention utilizes near-infrared imaging technology, by the attitude of automatic discrimination grain grain, and according to the different appearances of grain grain
State uses different disaggregated models, makes whether grain grain is subject to the classification accuracy rate of pestinfestation to reach more than 95%, improves near
The discrimination of infrared imaging method automatic detection grain intragranular portion insect；
(3) present invention is by grain grain automatic sampling transmitting device and includes the Computer Vision Detection System of high speed processing software
Combine, realize automatic, the accurate differentiation that the portion's insect pest of grain intragranular is infected, improve the insect pest detection of grain intragranular portion oneself
Dynamicization degree, reduces labour intensity.
Brief description of the drawings
Fig. 1 is structural representation of the invention；
Wherein, 1. feeder, 2. row grain wheel, 3. row grain motor, 4. transmission electrical machine, 5. near infrared camera, 6. optical filter turn
Wheel, 7. near-infrared light source, 8. clear grain brush, 9. conveyer belt, 10. lighting box, 11. computers, 12. disposable boxes, 13. second collections
Box, 14. second photoelectric sensors, 15. first collecting cassettes, 16. first photoelectric sensors, 17. control modules.
Fig. 2 is method of the present invention process chart.
Specific embodiment
Embodiments of the present invention are described below in detail, the example of the implementation method is shown in the drawings, wherein ad initio
Same or similar element or element with same or like function are represented to same or similar label eventually.Below by ginseng
The implementation method for examining Description of Drawings is exemplary, is only used for explaining the present invention, and is not construed as limiting the claims.
Technical scheme is described in further detail below in conjunction with the accompanying drawings：
Structural representation of the invention is as shown in figure 1, described detection means is by grain grain sampling part, grain grain hop
Constituted with computer vision process part.Grain grain sampling part and computer vision process part are all arranged on grain grain hop
Above.It is characterized in that：
The function that described grain grain sampling part is realized is quantitative grain grain is quickly and efficiently isolated from feeder
Come, mainly include feeder 1, row's grain wheel 2, row's grain motor 3 and clear grain brush 8.Existing grain intragranular portion insect pest computer vision system
System does not possess automatic ration sampled functions, and grain grain sampling part of the invention is directed to grain intragranular portion's insect pest computer vision certainly
Move detection and design.Feeder 1 is fixed on device frame, the surface in row's grain wheel 2.The circumferential surface of row's grain wheel 2 is equal
It is even to be dispersed with 4 grooves of flat trapeze, it is connected with row's grain motor 3 by single belt, it is slightly larger than with the difference in height of conveyer belt 9 and adopts
Collect the height of box.Clear grain brush 8 is fixed together with feeder 1, and the Way out with grain grain is consistent, and the outer surface for arranging grain wheel 2
Keep certain interval.Therefore, after grain sample pours into feeder 1, the row's grain driven by above-mentioned row's grain motor 3 takes turns 2 certainly
It is dynamic to carry out quantitative sampling.
The function that described grain grain hop is realized is accurately to receive grain grain, the collection of the first collecting cassette of conveying 15 and second
Box 13 is to the underface of computer vision process part for IMAQ, and cleaning the first collecting cassette 15 and the second collecting cassette 13
In grain grain, mainly include that control module 17, transmission electrical machine 4, disposable box 12, conveyer belt 9, the first collecting cassette 15, second are gathered
Box 13, the first photoelectric sensor 16 and the second photoelectric sensor 14.Control module 17 passes through data wire and the first photoelectric sensor
16th, the second photoelectric sensor 14, row grain motor 3, computer 11 and transmission electrical machine 4 be connected, computer 11 by it receive come
From the first photoelectric sensor 16 and the trigger signal of the second photoelectric sensor 14, and by it to row's grain motor 3 and transmission electrical machine 4
Send control instruction.Conveyer belt 9 uses chain Chain conveyer transmission mechanism, is driven by transmission electrical machine 4 and operated.First collecting cassette
15 and second collecting cassette 13 be uniformly distributed on chain, 2 collecting cassette front ends are hinged on chain, and rear end is free end,
Can flexible rotating.2 photoelectric sensors be separately mounted to row's grain wheel 2 export, the lower section of near infrared camera 5, in conveyer belt 9
Side, and the slightly less than position of the height of collecting cassette 15.When they detect 2 collecting cassettes move to row's grain wheel 2 export, near-infrared phase
During the underface of machine 5, the trigger signal that the sampling of grain grain and multi-wavelength image shoot successively is provided to control module 17.Adopt when first
The collection collecting cassette 13 of box 15 or second is performed to buckle and stirs work when running to extreme position, topples over the grain grain in collecting cassette, and by installing
Disposable box 12 immediately below it is collected.Disposable box 12 is located at the underface on the right side of conveyer belt 9, apart from the height of conveyer belt 9
Degree difference should be greater than height when collecting cassette stands upside down, and collecting cassette is performed after button stirs work can continue to run forward.
Described computer vision process part, major function is the multi-wavelength near-infrared image for gathering grain grain, by near red
Outer camera 5, optical filter runner 6, near-infrared light source 7, lighting box 10, computer 11 and corresponding software composition.Lighting box 10 is installed
In the surface of conveyer belt 9, there is the annular light source of the near infrared band of uniform illumination the inside, for the grain grain in collecting cassette is provided
Even diffuses.Near infrared camera 5 is arranged in lighting box 10, is vertically mounted on the surface of conveyer belt 9, is transported with collecting cassette
Dynamic direction in the same plane, should make the collecting cassette bottom surface effective range for collecting as big as possible apart from the height of conveyer belt 9, and
The multi-wavelength near-infrared image of clear shooting, collecting box China Oil and Food Import and Export Corporation grain can be distinguished.The spectral region of near infrared camera 5 is 900-
1700nm, its front end carries near infrared filter runner.Near infrared camera 5 is connected through data wire with computer 11, and will shoot
Image transmitting to computer 11.The inside of optical filter runner 6 is provided with the near infrared filter of several different wave lengths, near-infrared filter
The half-band width of mating plate is respectively less than 10nm.Optical filter runner is connected through data wire with computer 11, receives the control from computer 11
Information processed, to control the automatic switchover of several optical filters during multi-wavelength image automatic data collection.
Method of the present invention process chart as shown in Fig. 2 work when, grain sample feeding feeder 1 after, computer 11
Control instruction is sent to control module 17 and start row's grain motor 3, the rotation of the drive row's grain wheel 2 of row's grain motor 3, the grain grain row of falling into
In the first collecting cassette 15 below the outlet of grain wheel 2.Transmission electrical machine 4 drives conveyer belt 9 to rotate, when the first photoelectric sensor 16 is detected
To the first collecting cassette 15 in place after, computer 11 control near infrared camera 5 and optical filter runner 6 carry out multi-wavelength IMAQ,
After collection is finished, conveyer belt 13 moves on.First wave length image to grain grain carries out image procossing, according to mesh after image segmentation
Mark area information in the picture, accurately determines effective grain grain target, records all grain grain targets in first wave length image
In the information such as coordinate, center of gravity, be partitioned into single complete grain grain, obtain background it is completely black only containing the gray scale subgraph of single grain grain
Picture.For the grayscale sub-image of single grain grain under first wave length, the two-value after asking for grain grain (inside) profile and Threshold segmentation
The difference of image, according to target in difference image and the difference of the ultimate range on grain grain border, can automatic discrimination go out the target and be
No is border (inside) hole.If grain grain target has border hole or inner void, grain grain is directly identified as insect
Grain grain is infected, if grain grain target is without hole, into the poses discrimination of next step.For the gray scale of single grain grain under first wave length
Subgraph, extracts invariant moment features, the absolute value etc. along grain grain length axle two ends area invariant moment features difference；For first wave length
Under single grain grain bianry image, extract the nearest equivalent distances on flexibility, circle, major axis and its minimum enclosed rectangle side long etc.
Feature, forms the original feature space of grain grain attitude.Using data compression technique, to grain grain original feature space Dimension Reduction Analysis,
Characteristic parameter information of the accumulation contribution rate higher than 90% is extracted, the optimization feature space of grain grain attitude is formed.For grain grain attitude
Optimization feature space, using nonlinear pattern recognition sorting technique set up first wave length image China Oil and Food Import and Export Corporation grain optimization after characteristic parameter
With the relational model of grain grain attitude (ventral groove is down, upward, laterally), each grain is automatically determined out by grain grain gesture recognition model
The attitude of grain.For the grain grain gray level image under multi-wavelength, its statistical nature, histogram transform characteristics, texture are extracted respectively special
Levy, form and optimize the original feature space under different grain grain attitudes.According to the grain grain of different attitudes, using intelligent mode point
Class technology, sets up in multi-wavelength image whether grain grain image features are subject to the relation of pestinfestation with grain grain after optimization respectively
Model, and ensure that 3 disaggregated model precision of attitude more than 95%, can determine that with grain intragranular portion insect identification model
Whether grain grain is infected by insect.The grain grain number amount of pestinfestation is subject in multi-wavelength image acquired in counting, if depositing
Grain grain is being infected, the information such as the multi-wavelength image, the center of gravity for infecting grain grain are being recorded.When the first collecting cassette 15 moves to rightmost
Spin upside down, grain grain is automatically fallen into disposable box 12.After the first photoelectric sensor 16 detects the second collecting cassette 13 in place, pass
Transmission of electricity machine 4 is stalled and detected next time with wait.
Embodiments of the present invention are explained in detail above in conjunction with accompanying drawing, but the present invention is not limited to above-mentioned implementation
Mode, in the ken that those of ordinary skill in the art possess, can also be on the premise of present inventive concept not be departed from
Make a variety of changes.The above, is only presently preferred embodiments of the present invention, and any formal limit is not made to the present invention
System, although the present invention is disclosed above with preferred embodiment, but is not limited to the present invention, any to be familiar with this professional skill
Art personnel, without departing from the scope of the present invention, when using the technology contents of the disclosure above make it is a little change or
The Equivalent embodiments of equivalent variations are modified to, as long as being without departing from technical solution of the present invention content, according to technology reality of the invention
Matter, within the spirit and principles in the present invention, any simple modification, equivalent and the improvement made to above example
Deng still falling within the protection domain of technical solution of the present invention.
Claims (9)
1. a kind of detection method of the grain intragranular portion injurious insect detector based on near-infrared computer vision, it is characterised in that institute
State detection means to be made up of grain grain sampling part, grain grain hop and computer vision process part, the grain grain sampling portion
Divide the surface that the grain grain hop is arranged at computer vision process part, wherein：
The grain grain sampling part includes feeder, row's grain wheel, row grain motor and clear grain brush, and described feeder and row's grain wheel leads to
Support fixed installation is crossed, row's grain wheel is installed on the underface of feeder outlets；The surface of row's grain wheel is evenly distributed with flat
Groove, row's grain wheel is connected by single belt with row's grain motor；The exit of described feeder is fixedly connected with clear grain brush, clear grain
Gap is left between the outer surface of brush and row's grain wheel；
The computer vision process part include lighting box and computer, wherein, at lighting box inside center from the bottom to top according to
Secondary to be provided with near-infrared light source, optical filter runner and near infrared camera, the optical filter runner and near infrared camera are via number
It is connected with computer according to line；
The grain grain hop includes control module, transmission electrical machine, disposable box, conveyer belt, the first collecting cassette, the second collection
Box, the first photoelectric sensor and the second photoelectric sensor, wherein, the control module passes through data wire and the first photoelectric transfer respectively
Sensor, the second photoelectric sensor, row's grain motor, computer are connected with transmission electrical machine, and computer is received by control module and come from
The trigger signal of the first photoelectric sensor and the second photoelectric sensor, then row's grain motor and transmission electrical machine are sent out via control module
Go out control signal；Transmission electrical machine is connected with one end of conveyer belt, drives conveyer belt operating, and it is another that the disposable box is arranged at conveyer belt
The lower section of one end；The first, second collecting cassette proportional spacing is arranged on the conveyer belt, first, second collecting cassette
Front end is hinged conveyer belt, and rear end is free end, and collecting cassette can be rotated by axle of hinged place；First, second photoelectric sensor
Conveyer belt at immediately below the row's of being respectively arranged at grain wheel exit and lighting box is inboard, when the photoelectric sensor detects collection
When box is moved to the underface of near infrared camera in the outlet of row's grain wheel and lighting box, photoelectric sensor sends grain grain to control module
The trigger signal that sampling and multi-wavelength image shoot successively；
With the height of first, second collecting cassette as reference, distance of the lower end that row's grain is taken turns away from conveyer belt is more than collection
The height of box；Distance of the lighting box away from conveyer belt is more than the height of collecting cassette and so that what is photographed near infrared camera adopts
Collection box area is maximum；Height when distance of the disposable box away from conveyer belt is erect more than collecting cassette, makes collecting cassette perform button and turns over
Action can continue to run forward after toppling over grain grain；
Specific steps include：
Step (1), grain grain sampling part extract quantitative grain grain from feed appliance, and grain grain is fallen into collecting cassette；
Step (2), grain grain hop are sent to collecting cassette the underface of near infrared camera, computer vision process part pair
Grain grain in collecting cassette carries out multi-wavelength near-infrared image collection, and collection result is transported into computer；
Step (3), the first wave length near-infrared collection image to grain grain carry out image procossing, according to grain grain after image segmentation in figure
Area information as in, determines effective grain grain target, records parameter letter of all grain grain targets in first wave length image
Breath, is partitioned into single complete grain grain, obtains the completely black grayscale sub-image for only containing single grain grain of background；
Step (4), the grayscale sub-image for single grain grain under first wave length, ask for two after grain intragranular contouring and Threshold segmentation
It is worth the difference of image, the difference of target and grain the grain object boundary ultimate range in difference image determines the difference diagram
Whether the target as in is border hole or inner void：
(401) when grain grain target has border hole or inner void, then judge that grain grain is pestinfestation grain grain；
(402) when grain grain target is without hole, into the poses discrimination of next step；
Step (5), the grayscale sub-image for single grain grain under first wave length, extract corresponding gray level image feature, for the
Single grain grain bianry image, extracts corresponding bianry image feature under one wavelength, and the primitive character for ultimately forming grain grain attitude is empty
Between, data compression technique is recycled, to grain grain original feature space Dimension Reduction Analysis, extract feature of the accumulation contribution rate higher than 90%
Parameter information, forms the optimization feature space of grain grain attitude；
Step (6), the optimization feature space for grain grain attitude, first wave length is set up using nonlinear pattern recognition sorting technique
Characteristic parameter and the relational model of grain grain attitude, each grain grain is determined by grain grain attitude relational model after the grain optimization of image China Oil and Food Import and Export Corporation
Attitude；
Step (7), under multi-wavelength grain grain grayscale sub-image, its statistical nature, histogram transform characteristics, line are extracted respectively
Reason feature, forms and optimizes the original feature space under different grain grain attitudes, according to the grain grain of different attitudes, using intelligent mode
Sorting technique, sets up in multi-wavelength image whether grain grain image features are closed with grain grain by pestinfestation after optimization respectively
Whether it is model, is infected by insect with the grain intragranular portion insect identification model grain that determines to put out cereal；
Step (8), to count and be subject to the grain grain number amount of pestinfestation in acquired multi-wavelength image, and record and infect grain grain
Multi-wavelength image and position of centre of gravity information；
Step (9), computer to obtain multi-wavelength near-infrared image analyze and process when, grain grain hop continue to run with,
When collecting cassette runs to conveyer belt free end extreme position, the grain grain in collecting cassette is poured onto disposable box.
2. the detection method of the grain intragranular portion injurious insect detector of near-infrared computer vision is based on as claimed in claim 1, its
It is characterised by：The conveyer belt is chain Chain conveyer transmission mechanism.
3. the detection method of the grain intragranular portion injurious insect detector of near-infrared computer vision is based on as claimed in claim 1, its
It is characterised by：The equally distributed flat groove in surface of row's grain wheel is trapezoidal, and quantity is four.
4. the detection method of the grain intragranular portion injurious insect detector of near-infrared computer vision is based on as claimed in claim 1, its
It is characterised by：The spectral region of the near infrared camera is 900-1700nm.
5. the detection method of the grain intragranular portion injurious insect detector of near-infrared computer vision is based on as claimed in claim 1, its
It is characterised by：The near infrared filter of a plurality of different wave lengths, and the near infrared filter are installed in the optical filter runner
The half-band width of piece is respectively less than 10nm.
6. the detection method of the grain intragranular portion injurious insect detector of near-infrared computer vision is based on as claimed in claim 1, its
It is characterised by：In step (3), described image procossing includes removal collecting cassette background, filtering enhancing image, segmentation figure picture, grain
Grain is accurately positioned.
7. the detection method of the grain intragranular portion injurious insect detector of near-infrared computer vision is based on as claimed in claim 1, its
It is characterised by：In step (3), parameter information of the grain grain target in first wave length image includes co-ordinate position information and weight
Heart positional information.
8. the detection method of the grain intragranular portion injurious insect detector of near-infrared computer vision is based on as claimed in claim 1, its
It is characterised by：In step (5), the characteristics of image of the grayscale sub-image of single grain grain includes under the first wave length：Bending moment is not special
The absolute value sought peace along grain grain length axle two ends area invariant moment features difference；Single grain grain bianry image under the first wave length
Characteristics of image includes：The nearest equivalent distances on flexibility, circle, major axis and its minimum enclosed rectangle side long.
9. the detection method of the grain intragranular portion injurious insect detector of near-infrared computer vision is based on as claimed in claim 1, its
It is characterised by：The grain attitude of grain described in step (6) includes：Ventral groove is down, upward and laterally.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410404698.3A CN104155312B (en) | 2014-08-11 | 2014-08-11 | Grain intragranular portion pest detection method and apparatus based on near-infrared computer vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410404698.3A CN104155312B (en) | 2014-08-11 | 2014-08-11 | Grain intragranular portion pest detection method and apparatus based on near-infrared computer vision |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104155312A CN104155312A (en) | 2014-11-19 |
| CN104155312B true CN104155312B (en) | 2017-06-16 |
Family
ID=51880871
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410404698.3A Expired - Fee Related CN104155312B (en) | 2014-08-11 | 2014-08-11 | Grain intragranular portion pest detection method and apparatus based on near-infrared computer vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104155312B (en) |
Families Citing this family (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104749189A (en) * | 2015-02-28 | 2015-07-01 | 华北水利水电大学 | Grain interior insect pest detection device based on multi-spectral imaging technology |
| CN105187727A (en) * | 2015-06-17 | 2015-12-23 | 广州市巽腾信息科技有限公司 | Image information acquisition device, image acquisition method and use of image information acquisition device |
| CN105187726B (en) * | 2015-06-17 | 2021-05-18 | 巽腾(广东)科技有限公司 | Multifunctional mobile image processing device and processing method |
| CN106127226B (en) * | 2016-06-14 | 2019-09-03 | 河南工业大学 | The flexible grain quality detection method of grain grain and grain grain test sample |
| WO2018038123A1 (en) * | 2016-08-22 | 2018-03-01 | キユーピー株式会社 | Food inspecting device, food inspecting method, and learning method for identification means of food inspecting device |
| CN106770002A (en) * | 2016-11-21 | 2017-05-31 | 无锡艾科瑞思产品设计与研究有限公司 | A kind of near infrared ray domestic food detection means |
| CN106769980A (en) * | 2016-11-22 | 2017-05-31 | 南京财经大学 | A kind of method for distinguishing cereal and insect infrared radiation absorption characteristic |
| CN107610122B (en) * | 2017-09-30 | 2020-07-28 | 华北水利水电大学 | Micro-CT-based single-grain cereal internal insect pest detection method |
| IL256031B (en) * | 2017-11-30 | 2021-02-28 | Yosef Akrav Pinhas | Automatic system for separating and identifying insects on leaves |
| CN107894419B (en) * | 2017-12-29 | 2023-10-27 | 南京艾龙信息科技有限公司 | Device and method for detecting raw grain pests |
| CN109034269A (en) * | 2018-08-22 | 2018-12-18 | 华北水利水电大学 | A kind of bollworm female male imago method of discrimination based on computer vision technique |
| CN111768404A (en) * | 2020-07-08 | 2020-10-13 | 北京滴普科技有限公司 | Mask appearance defect detection system, method and device and storage medium |
| CN112505049B (en) * | 2020-10-14 | 2021-08-03 | 上海互觉科技有限公司 | Mask inhibition-based method and system for detecting surface defects of precision components |
| CN112415009B (en) * | 2020-12-14 | 2024-06-11 | 河南牧业经济学院 | Wheat grain non-adhesion image acquisition method and system based on viscous lattice |
| CN113866176A (en) * | 2021-09-18 | 2021-12-31 | 北京中仪智控科技有限公司 | Full-automatic analyzer for imperfect grains |
| CN114037704B (en) * | 2022-01-10 | 2022-04-22 | 安徽高哲信息技术有限公司 | Feeding system, control method and control device thereof, and storage medium |
| CN117074353B (en) * | 2023-08-18 | 2024-05-03 | 广东省农业科学院设施农业研究所 | Nondestructive detection method and system for litchi fruit Di-moths |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201382896Y (en) * | 2009-03-05 | 2010-01-13 | 华中农业大学 | Sorting experimental bench for multisource information detection of chesnut quality |
| CN101701915A (en) * | 2009-11-13 | 2010-05-05 | 江苏大学 | Device and method for detecting stored-grain insects based on visible light-near infrared binocular machine vision |
| CN102072883A (en) * | 2010-07-07 | 2011-05-25 | 北京农业智能装备技术研究中心 | Device and method for detecting comprehensive quality of crop seeds |
| CN102095481A (en) * | 2010-12-02 | 2011-06-15 | 江苏大学 | Device and method for automatically discriminating cocoon sexuality |
| CN103706574A (en) * | 2013-12-25 | 2014-04-09 | 中国科学院半导体研究所 | Automatic solid grain sorting system |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH1183733A (en) * | 1997-09-03 | 1999-03-26 | Kubota Corp | Grade measuring instrument for grain |
| DE102009010438B3 (en) * | 2009-02-26 | 2010-08-12 | Kws Saat Ag | Method and device for determining ingredients in root crops |
| JP5877373B2 (en) * | 2012-05-16 | 2016-03-08 | パナソニックＩｐマネジメント株式会社 | Plastic mixture separation method and plastic mixture separation device |
-
2014
- 2014-08-11 CN CN201410404698.3A patent/CN104155312B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201382896Y (en) * | 2009-03-05 | 2010-01-13 | 华中农业大学 | Sorting experimental bench for multisource information detection of chesnut quality |
| CN101701915A (en) * | 2009-11-13 | 2010-05-05 | 江苏大学 | Device and method for detecting stored-grain insects based on visible light-near infrared binocular machine vision |
| CN102072883A (en) * | 2010-07-07 | 2011-05-25 | 北京农业智能装备技术研究中心 | Device and method for detecting comprehensive quality of crop seeds |
| CN102095481A (en) * | 2010-12-02 | 2011-06-15 | 江苏大学 | Device and method for automatically discriminating cocoon sexuality |
| CN103706574A (en) * | 2013-12-25 | 2014-04-09 | 中国科学院半导体研究所 | Automatic solid grain sorting system |
Non-Patent Citations (4)
| Title |
|---|
| Detection techniques for stored-product insects in grain;S.Neethirajan et al.;《Food Control》;20071231;第18卷(第2期);第157-162页 * |
| Identifying Stored-Grain Insects Using Near-Infrared Spectroscopy;F.E.DOWELL et al.;《JOURNAL OF ECONOMIC ENTOMOLOGY》;19990228;第92卷(第1期);第165-169页 * |
| 储粮害虫图像识别中的特征提取;张红涛 等;《农业工程学报》;20090228;第25卷(第2期);第126-130页 * |
| 基于机器视觉的大豆籽粒精选技术;王润涛 等;《农业工程学报》;20110831;第27卷(第8期);第355-359页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104155312A (en) | 2014-11-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104155312B (en) | Grain intragranular portion pest detection method and apparatus based on near-infrared computer vision | |
| Momin et al. | Machine vision based soybean quality evaluation | |
| Zhou et al. | Strawberry maturity classification from UAV and near-ground imaging using deep learning | |
| CN101701915B (en) | Device and method for detecting stored-grain insects based on visible light-near infrared binocular machine vision | |
| CN103063585B (en) | Melon and fruit degree of ripeness Rapid non-destructive testing device and detection system method for building up | |
| CN104198324B (en) | Computer vision-based method for measuring proportion of cut leaves in cut tobacco | |
| CN102072883B (en) | Device and method for detecting comprehensive quality of crop seeds | |
| JP5812321B2 (en) | Bug image processing apparatus, image processing method, image processing program, and computer-readable storage medium | |
| Stajnko et al. | Modelling apple fruit yield using image analysis for fruit colour, shape and texture | |
| CN107952696A (en) | A kind of detection grading plant and detection method suitable for fresh tobacco leaf | |
| CN202010662U (en) | Real-time inspection and grading system device for fruit appearance quality | |
| Lee et al. | Sensing systems for precision agriculture in Florida | |
| CN110705655A (en) | Tobacco leaf classification method based on coupling of spectrum and machine vision | |
| CN101701906A (en) | Method and device for detecting stored-grain insects based on near infrared super-spectral imaging technology | |
| CN104990892B (en) | The spectrum picture Undamaged determination method for establishing model and seeds idenmtification method of seed | |
| CN103954570A (en) | Food insect attack degree distinguishing method based on spectral imaging technology | |
| CN210465246U (en) | Spectrum acquisition device of wheat infection scab grade and disease grain rate hyperspectral nondestructive detection device | |
| CN105548124A (en) | Method and device for detecting citrus canker | |
| Roseleena et al. | Assessment of palm oil fresh fruit bunches using photogrammetric grading system. | |
| CN113820325B (en) | Corn grain direct-harvest impurity-containing rate and breakage rate online detection system and method | |
| Shin et al. | Postharvest citrus mass and size estimation using a logistic classification model and a watershed algorithm | |
| CN114419311B (en) | Multi-source information-based passion fruit maturity nondestructive testing method and device | |
| CN103528967A (en) | Hyperspectral image based overripe Lonicera edulis fruit identification method | |
| CN104749189A (en) | Grain interior insect pest detection device based on multi-spectral imaging technology | |
| CN106556602A (en) | A kind of detection method and its device of fish body freshness |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170616 |