CN111570330B - Machine vision selection device and method for clear eggs in whole-tray duck egg planting - Google Patents
Machine vision selection device and method for clear eggs in whole-tray duck egg planting Download PDFInfo
- Publication number
- CN111570330B CN111570330B CN202010547278.6A CN202010547278A CN111570330B CN 111570330 B CN111570330 B CN 111570330B CN 202010547278 A CN202010547278 A CN 202010547278A CN 111570330 B CN111570330 B CN 111570330B
- Authority
- CN
- China
- Prior art keywords
- whole
- eggs
- egg
- tray
- duck
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 235000013601 eggs Nutrition 0.000 title claims abstract description 345
- 241000272525 Anas platyrhynchos Species 0.000 title claims abstract description 118
- 238000000034 method Methods 0.000 title claims description 19
- 230000007246 mechanism Effects 0.000 claims abstract description 29
- 238000009395 breeding Methods 0.000 claims abstract description 27
- 230000001488 breeding effect Effects 0.000 claims abstract description 27
- 230000012447 hatching Effects 0.000 claims abstract description 11
- 238000010187 selection method Methods 0.000 claims abstract description 4
- 238000001514 detection method Methods 0.000 claims description 24
- 230000003287 optical effect Effects 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 13
- 230000004720 fertilization Effects 0.000 claims description 10
- 238000002372 labelling Methods 0.000 claims description 9
- 230000017448 oviposition Effects 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 8
- 238000005516 engineering process Methods 0.000 claims description 7
- 238000013135 deep learning Methods 0.000 claims description 5
- 238000013527 convolutional neural network Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 3
- 239000006096 absorbing agent Substances 0.000 abstract 2
- 230000006870 function Effects 0.000 description 7
- 244000144977 poultry Species 0.000 description 5
- 241000272522 Anas Species 0.000 description 4
- 230000008030 elimination Effects 0.000 description 3
- 238000003379 elimination reaction Methods 0.000 description 3
- 238000012946 outsourcing Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 235000013305 food Nutrition 0.000 description 2
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 1
- 241001465754 Metazoa Species 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000009432 framing Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 235000013372 meat Nutrition 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 239000003973 paint Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000003307 slaughter Methods 0.000 description 1
- 229910001220 stainless steel Inorganic materials 0.000 description 1
- 239000010935 stainless steel Substances 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/34—Sorting according to other particular properties
- B07C5/342—Sorting according to other particular properties according to optical properties, e.g. colour
- B07C5/3422—Sorting according to other particular properties according to optical properties, e.g. colour using video scanning devices, e.g. TV-cameras
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01K—ANIMAL HUSBANDRY; AVICULTURE; APICULTURE; PISCICULTURE; FISHING; REARING OR BREEDING ANIMALS, NOT OTHERWISE PROVIDED FOR; NEW BREEDS OF ANIMALS
- A01K43/00—Testing, sorting or cleaning eggs ; Conveying devices ; Pick-up devices
- A01K43/04—Grading eggs
- A01K43/10—Grading and stamping
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/02—Measures preceding sorting, e.g. arranging articles in a stream orientating
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/36—Sorting apparatus characterised by the means used for distribution
- B07C5/361—Processing or control devices therefor, e.g. escort memory
- B07C5/362—Separating or distributor mechanisms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C2501/00—Sorting according to a characteristic or feature of the articles or material to be sorted
- B07C2501/0081—Sorting of food items
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Environmental Sciences (AREA)
- Animal Husbandry (AREA)
- Biodiversity & Conservation Biology (AREA)
- Multimedia (AREA)
- Wrapping Of Specific Fragile Articles (AREA)
Abstract
The invention discloses a machine vision selection device and a machine vision selection method for eggs without eggs in the whole duck egg tray, wherein the machine vision selection device comprises a crawler-type conveyor belt, the crawler-type conveyor belt is connected with a first motor in a driving way, a light chamber is arranged at the central position of the crawler-type conveyor belt, an egg tray is arranged on the crawler-type conveyor belt, a plurality of duck eggs incubated for 5 days are vertically arranged in the egg tray, a whole egg tray absorber is arranged in the light chamber, the whole egg tray absorber is connected to the inner side wall of the light chamber in a sliding way, an image acquisition mechanism is arranged on the light chamber, and the image acquisition mechanism is connected with a control mechanism in a signal way; the image acquisition mechanism is used for acquiring images of the whole tray of the breeding duck eggs and identifying clear eggs and fertilized eggs; the control mechanism is used for controlling the whole tray egg sucker to move and grab the breeding duck eggs belonging to the clear eggs; the whole-tray egg sucker is used for grabbing duck eggs planted on an egg tray, and the problem of low automation and intelligence degree in the egg hatching industry can be solved practically.
Description
Technical Field
The invention relates to the technical field of online nondestructive automatic detection and selection of agricultural products, in particular to a machine vision selection device and method for clear eggs in a whole tray of duck eggs.
Background
According to data provided by the animal husbandry association in China and the Food and Agricultural Organization (FAO) in the world in 2018, the slaughtering amount of meat ducks in the world is about 47.4 hundred million, the stocking amount of laying ducks is about 1.87 hundred million, and the duck breeding in China accounts for 74.3% of the duck breeding amount in the world. This indicates that billions of duck seedlings need to be hatched every year in the duck egg hatching industry of China to meet the domestic breeding demand. However, the domestic breeding mode of breeding ducks mainly adopts free-range breeding, so that the natural fertilization rate of the breeding ducks is lower than 80% (Li Qingxu, 2020, agricultural machinery science and newspaper). At present, in the poultry egg hatching industry in China, the mode of removing the eggs without eggs is mainly manual egg lighting, and has the disadvantages of high labor intensity, time and labor waste and low efficiency (aspirations, 2015, agricultural engineering bulletins). The poultry egg hatching industry in China urgently needs an efficient and intelligent device for automatically removing the clear eggs, and at present, students at home and abroad make a lot of researches on the poultry egg hatching aspect. Zhang et al (Zhang Wei, 2012, the report of agricultural machinery) applied the machine vision technology to combine with knocking vibration to carry on the nondestructive test to the fertilization information of the single kind duck egg that hatches for 5 days, utilize the color information of the kind duck egg image and knocking vibration's signal characteristic, set up LVQ neural network and distinguished the model, distinguish the accuracy to be 98%, but this method stability and practicability are bad; the method has the advantages that Li Qingxu and the like (Li Qingxu, 2020, agricultural machinery science and newspaper) utilize deep learning to detect fertilization information of single-breed duck eggs incubated for 3 days, an 11-layer convolutional neural network is built, the detection precision of the method is 97.97%, the detection efficiency of the single-breed duck eggs needs to be improved, and the method has guiding significance for detecting the fertilization information of the poultry eggs by using the deep learning method; wulinfeng and the like (Wulinfeng, 2019, food and machinery) apply a machine vision combined weighing method to carry out nondestructive detection on fertilized eggs in group-bred duck eggs, RGB, HIS and gray level mean values of the duck egg images are extracted after segmentation and smooth de-noising of the group-bred duck egg images, and an SVM discrimination model is established by combining the change condition of the egg weight of the bred duck eggs in the hatching process, the discrimination accuracy of the fertilized eggs in the group-bred duck eggs is 96.7 percent, the method has low practicability, needs to weigh the eggs every day, cannot detect the bred duck eggs in an egg tray, but has a guiding function on the fertilization information detection of the group-bred eggs. Through retrieval, the nondestructive detection research on fertilization information of a single egg is more at home and abroad, and the nondestructive detection research on fertilized eggs and non-fertilized eggs in a whole set of duck eggs by only using a machine vision technology is not developed at present.
Disclosure of Invention
The invention aims to provide a machine vision selection device and a method for selecting clear eggs in a whole tray of duck eggs, so as to solve the problems in the prior art and practically solve the problems of low automation and intelligence degree in the poultry egg hatching industry.
In order to achieve the purpose, the invention provides the following scheme: the invention provides a machine vision selecting device for eggs without eggs in sperm in a whole tray of breeding duck eggs, which comprises a crawler-type conveyor belt, wherein the crawler-type conveyor belt is connected with a first motor in a driving manner, a light chamber is arranged at the central position of the crawler-type conveyor belt, an egg tray is placed on the crawler-type conveyor belt, a plurality of breeding duck eggs for hatching for 5 days are vertically placed in the egg tray, a whole tray of egg sucker is arranged in the light chamber, the whole tray of egg sucker is connected to the inner side wall of the light chamber in a sliding manner, an image collecting mechanism is arranged on the light chamber, and the image collecting mechanism is connected with a control mechanism in a signal manner;
the image acquisition mechanism is used for acquiring images of the whole tray of the duck eggs and identifying clear eggs and fertilized eggs;
the control mechanism is used for controlling the whole tray egg sucker to move and grab the breeding duck eggs belonging to the clear eggs;
the whole-tray egg sucker is used for grabbing duck eggs planted on the egg tray.
Preferably, the crawler-type conveyor belt comprises a first conveyor belt and a second conveyor belt, the first conveyor belt is perpendicular to the second conveyor belt, the first conveyor belt is arranged on the left side and the right side of the optical chamber, and the second conveyor belt is arranged on the front side of the optical chamber.
Preferably, the acquisition mechanism comprises a photoelectric sensor and an industrial camera, the photoelectric sensor is positioned at the right end of the optical chamber, and the industrial camera is positioned at the top end of the optical chamber.
Preferably, the control mechanism comprises a PLC controller, the PLC controller is in signal connection with a computer, the computer is in signal connection with the industrial camera, and the PLC controller is in signal connection with the photoelectric sensor.
Preferably, the PLC controller is in signal connection with a second motor, the second motor is in driving connection with a group of guide rails, the guide rails are fixed on the inner side wall of the light chamber, and the whole-tray egg sucker is in sliding connection with the guide rails.
Preferably, the bottom end of the whole-tray egg sucker is provided with a whole-tray egg candler, and the whole-tray egg candler is arranged on the bottom wall of the inner side of the light chamber.
A machine vision selection method for clear eggs in a whole tray of duck eggs comprises the following steps: the method comprises the following steps: collecting images of the whole duck eggs, namely placing the whole duck eggs on a crawler-type conveyor belt, operating a first motor to enable the crawler-type conveyor belt to move forward, triggering a photoelectric sensor when the whole duck eggs are conveyed into a light chamber, taking pictures by an industrial camera to collect color images of the whole duck eggs, and providing a light source by a whole egg candler and keeping the whole duck eggs in a normally open state;
step two: manually marking fertilized eggs and clear eggs in the whole tray of duck egg images, and adjusting the size of a color image to 1634 × 1234, 320 × 320; manually labeling fertilized eggs and non-fertilized eggs in the whole set of duck egg images by using a Labelimage labeling tool, wherein the labeling labels of the fertilized eggs are shoujin and the non-fertilized eggs are wujin;
step three: establishing a fertilization information identification model of the whole tray of duck eggs by using a deep learning algorithm;
step four: and identifying the clear eggs in the image of the whole plate of the duck eggs and removing the clear eggs.
Preferably, the third step includes a, for the marked image data, according to 7: 3, dividing the principle into a training set and a test set, and keeping partial images as a verification set without marking;
b. the front 17 layers of the MobileNet V3_ large network are used as a basic network for automatically extracting the characteristics of the image of the whole plate of the breeding duck eggs;
c. a 5-layer convolutional neural network is built as an auxiliary network and used for outputting the category and position information of the whole set of duck egg images, and standard convolution is replaced by inverse residual convolution introducing an attention mechanism;
d. b and c, constructing a complete ssdlite-MobileNet V3_ large target detection network, and inputting the image data after the training set and the test set are manually labeled into the ssdlite-MobileNet V3_ large network for training;
e. after 10000 times of training, saving parameters of the ssdlite-MobileNet V3_ large network and exporting a pb file, wherein the size of the file is about 9.10 MB;
preferably, said step four comprises
The method comprises the steps that i, a pb file of a ssdlite-MobileNet V3_ large network is used for distinguishing an image of a whole set of duck eggs, fertilized eggs and non-sperm eggs in the image can be identified, when the probability that one hatching egg is a non-sperm egg is larger than 0.5, the position of the non-sperm egg in the image is returned, and the fertilized eggs are similar to the non-sperm eggs;
ii, calculating the real position of the clear eggs in the egg tray by using an industrial camera calibration technology;
iii, enabling the whole-tray egg sucker with the control valve to fall over the duck egg through a guide rail by the PLC, and enabling the whole-tray egg sucker to contact the surface of the duck egg;
iv, transmitting the real position information of the clear eggs to a PLC (programmable logic controller), and controlling a valve of a whole-disc egg sucker right above the clear eggs to be opened by the PLC to finish sucking and lifting the clear eggs;
and v, after the non-egg-laying eggs are sucked and lifted, the PLC makes the whole tray of egg-laying devices with the control valves return to the initial position again through the guide rails, the control valves of the whole tray of egg-laying devices are all closed, and the non-egg-laying eggs fall into the second conveyor belt to complete nondestructive online detection and selection of the non-egg-laying eggs in the whole tray of duck eggs.
The invention discloses the following technical effects: 1. by adopting a machine vision technology, the automatic detection and elimination steps of the clear eggs in the whole duck egg tray are researched, so that the automatic detection and elimination of the clear eggs in the whole duck egg tray can be completed;
2. the method has the characteristics of high efficiency and high intelligence, applies the artificial intelligence technology to the nondestructive detection of fertilization information of the whole tray of duck eggs, can realize nondestructive online detection, and has wide application prospect.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a schematic structural view of a machine vision selection device for clear eggs in a whole tray of duck eggs according to the invention;
FIG. 2 is a schematic structural view of the whole-tray egg sucker;
FIG. 3 is an image of a whole tray of duck eggs according to the present invention;
FIG. 4 is a diagram showing the recognition effect of the whole plate of duck eggs;
wherein 1 is a computer; 2 is an industrial camera; 3 is a light chamber; 4 is a guide rail; 5, a whole-tray egg sucker; 6, a whole tray egg candler; 7 is a breeding duck egg; 8 is a photoelectric sensor; 9 is an egg tray; 10 is a first motor; 11 is a crawler belt; 11-1 is a first conveyor belt; 11-2 is a second conveyor belt; 12 is a PLC controller; and 13 is a second motor.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Referring to fig. 1-2, the invention provides a machine vision selecting device for selecting clear eggs in a whole tray of duck eggs, which comprises a crawler belt 11, wherein the crawler belt 11 is in driving connection with a first motor 10, a light chamber 3 is arranged at the central position of the crawler belt 11, an egg tray 9 is placed on the crawler belt 11, a plurality of duck eggs 7 incubated for 5 days are vertically placed in the egg tray 9, a whole tray egg sucker 5 is arranged in the light chamber 3, the whole tray egg sucker 5 is connected to the inner side wall of the light chamber 3 in a sliding manner, black background paper is pasted in the box wall at the left side of the light chamber 3, an image acquisition mechanism is arranged on the light chamber 3, and the image acquisition mechanism is in signal connection with a control mechanism;
the image acquisition mechanism is used for acquiring images of a whole tray of the duck eggs 7 and identifying clear eggs and fertilized eggs;
the control mechanism is used for controlling the whole tray egg sucker 5 to move and grab the breeding duck eggs 7 belonging to the clear eggs;
the whole-tray egg sucker 5 is used for grabbing the breeding duck eggs 7 on the egg tray 9.
In a further optimized scheme, the crawler-type conveyor belt 11 comprises a first conveyor belt 11-1 and a second conveyor belt 11-2, the first conveyor belt 11-1 is perpendicular to the second conveyor belt 11-2, the first conveyor belt 11-1 is arranged on the left side and the right side of the optical chamber 3, and the second conveyor belt 11-2 is arranged on the front side of the optical chamber 3.
In a further optimized scheme, the acquisition mechanism comprises a photoelectric sensor 8 and an industrial camera 2, the photoelectric sensor 8 is positioned at the right end of the optical chamber 3, and the industrial camera 2 is positioned at the top end of the optical chamber 3.
According to a further optimized scheme, the control mechanism comprises a PLC (programmable logic controller) 12, the PLC 12 is in signal connection with a computer 1, the computer 1 is in signal connection with the industrial camera 2, and the PLC 12 is in signal connection with the photoelectric sensor 8.
According to a further optimized scheme, the PLC 12 is in signal connection with a second motor 13, the second motor 13 is in driving connection with a set of guide rails 4, the guide rails 4 are fixed on the inner side wall of the light chamber 3, and the whole-tray egg sucker 5 is in sliding connection with the guide rails 4.
In a further optimized scheme, the bottom end of the whole-tray egg sucker 5 is provided with a whole-tray egg candler 6, and the whole-tray egg candler 6 is arranged on the bottom wall of the inner side of the light chamber 3.
Referring to fig. 3-4, a machine vision selection method for clear eggs in a whole tray of duck eggs comprises the following steps: collecting images of the whole duck egg 7, namely placing the whole duck egg 7 on a crawler-type conveyor belt 11, operating a first motor 10 to enable the crawler-type conveyor belt 11 to move forward, triggering a photoelectric sensor 8 when the whole duck egg 7 is conveyed into a light chamber 3, taking a picture by an industrial camera 2 to collect a color image of the whole duck egg 7, and providing a light source by a whole duck egg candler 6 and keeping the whole duck egg candler in a normally open state;
step two: manually marking fertilized eggs and clear eggs in the whole tray of the duck egg 7 images, and adjusting the size of the color image to 1634 × 1234, 320 × 320; manually labeling fertilized eggs and non-fertilized eggs in the images of the whole set of duck eggs 7 by using a Labelimage labeling tool, wherein the labeling labels of the fertilized eggs are shoujin and the non-fertilized eggs are wujin;
step three: establishing a fertilization information identification model of the whole tray of duck eggs by using a deep learning algorithm;
step four: and identifying the clear eggs in the image of the whole plate of the duck eggs and removing the clear eggs.
And further optimizing the scheme, wherein the third step comprises a, according to 7: 3, dividing the principle into a training set and a test set, and keeping partial images as a verification set without marking;
b. the front 17 layers of the MobileNet V3_ large network are used as a basic network for automatically extracting the characteristics of the image 7 of the whole plate of the breeding duck eggs;
c. a 5-layer convolutional neural network is built as an auxiliary network and used for outputting the category and position information of the whole set of duck egg 7 images, and standard convolution is replaced by inverse residual convolution introducing an attention mechanism;
d. b and c, constructing a complete ssdlite-MobileNet V3_ large target detection network, and inputting the image data after the training set and the test set are manually labeled into the ssdlite-MobileNet V3_ large network for training;
e. after 10000 times of training, saving parameters of the ssdlite-MobileNet V3_ large network and exporting a pb file, wherein the size of the file is about 9.10 MB;
further optimizing the scheme, the step four comprises
The method comprises the steps that i, a pb file of a ssdlite-MobileNet V3_ large network is used for distinguishing a whole set of duck egg 7 images, fertilized eggs and non-fertilized eggs in the images can be identified, when the probability that one hatching egg is a non-fertilized egg is larger than 0.5, the position of the non-fertilized egg in the images is returned, and the fertilized eggs are similar to the fertilized eggs;
ii, calculating the real position of the clear eggs in the egg tray 9 by using the calibration technology of the industrial camera 2;
iii, enabling the whole-tray egg sucker 5 with a control valve to fall over the breeding duck eggs 7 through the guide rail 4 by the PLC 12, and enabling the whole-tray egg sucker 5 to contact the surfaces of the breeding duck eggs 7;
iv, transmitting the real position information of the clear eggs to the PLC 12, and controlling the valve of the whole-tray egg sucker 5 right above the clear eggs to be opened by the PLC 12 to finish sucking and lifting the clear eggs;
and v, after the non-egg-laying eggs are sucked and lifted, the PLC 12 enables the whole-tray egg sucker 5 with the control valve to return to the initial position again through the guide rail 4, the control valve of the whole-tray egg sucker 5 is completely closed, and the non-egg-laying eggs fall into the second conveyor belt 11-2 to complete nondestructive online detection and selection of the non-egg-laying eggs in the whole-tray duck eggs 7.
Firstly, horizontally placing a whole tray of duck eggs 7 above a first conveyor belt 11-1, turning on a first motor 10 to enable the whole tray of duck eggs 7 to move along the first conveyor belt 11-1, triggering a switch when the whole tray of duck eggs 7 move to a photoelectric sensor 8 on one side of an optical chamber 3, receiving a trigger signal by a PLC (programmable logic controller) 12 and sending the trigger signal to a computer 1 (upper computer), thereby controlling an industrial camera 2 to acquire images of the whole tray of duck eggs 7; after the computer 1 adjusts the size of the image, calling a pb file of a ssdlite-mobilenetV 3-large network to identify clear eggs and fertilized eggs in the image; then the position information of the non-fine eggs is sent to a PLC (programmable logic controller) 12, the PLC 12 controls the whole-tray egg sucker 5 with a control valve to fall into a correct position by operating a second motor 13 and a guide rail 4, and the control valve of the whole-tray egg sucker 5 which is in contact with the non-fine eggs is opened, so that the non-fine eggs are sucked and lifted; and finally, the PLC 12 controls the whole-tray egg sucker 5 with the control valve to return to the original position through the guide rail 4, the control valve of the whole-tray egg sucker 5 is closed, the non-finished eggs fall into the second conveyor belt 11-2, the fertilized eggs advance along the first conveyor belt 11-1 in the egg tray 9, and finally, the on-line nondestructive detection and elimination of the non-finished eggs are realized.
The invention mainly accomplishes the following work:
1. collecting images of a whole plate of duck eggs 7;
2. adjusting the size of the collected whole tray of duck egg 7 images;
3. building a ssdlite-MobileNet V3_ large target detection network, and finishing network training and pb file export;
4. detecting fertilized eggs and non-fertilized eggs of the whole tray of the duck eggs 7 by calling the pb file, and framing the positions of the fertilized eggs and the non-fertilized eggs;
5. the position information of the non-egg is utilized to realize accurate suction and lifting of the non-egg, and the rejecting work of the non-egg is completed.
The industrial camera 2 and the PLC 12 are connected with the computer 1, the photoelectric sensor 8, the second motor 13, the first motor 10 and the whole-tray egg sucker 5 with a control valve are connected with the PLC 12, the first motor 10 is connected with the crawler belt 11, the second motor 13 is connected with the whole-tray egg sucker 5 with a control valve, the industrial camera 2 is right opposite to the egg tray 9, and the egg tray 9 is right opposite to the whole-tray egg sucker 6.
The working principle of the invention is as follows: make whole set of kind of duck egg 7 get into in the light room 3 and with whole set of alignment about the egg ware 6 through crawler-type conveyer belt 11 and photoelectric sensor 8, through PLC controller 12, computer 1 triggers industrial camera 2 and shoots whole set of kind of duck egg 7, obtain the image of whole set of kind of duck egg 7, supply detection and analysis, make whole set of egg ware 5 of inhaling that has control valve through PLC controller 12 and guide rail 4 according to the information after detecting and accomplish the inhaling of not having smart egg and carry to realizing the rejection of not having smart egg.
The invention adopts the following functional components:
1) computer 1
Is a universal outsourcing element, such as a CPU Intel i5-2450M CPU @2.50GHz processor, a 64-bit Windows7 system; the function of the detection method is to call a pb file of ssdlite-MobileNet V3_ large to detect and control the whole tray of duck egg images in real time.
2) Industrial camera 2
The industrial camera is a general purchased part, such as a double-channel industrial camera 2 selected from JAI company; the function of the device is to collect the color images of the whole tray of the duck eggs 7.
3) Optical cell 3
The light chamber 3 is a rectangular stainless steel box body, and black paint is sprayed in the box; the function of the device is to ensure that the collected image is not influenced by external illumination and is used for installing and fixing the industrial camera 2 and the whole egg candler 6.
4) Guide rail 4
The guide rail 4 is a universal outsourcing member, such as a synchronous belt sliding table (WBD-TL 4080); the function of the device is to convey the whole tray egg sucker 5 with a control valve.
5) Whole tray egg sucker 5 with control valve
The whole tray egg sucker 5 with the control valve is a self-made component and totally comprises 7 multiplied by 9 sucking and lifting units, the interval between the sucking and lifting units is the same as the interval between the duck eggs 7 planted on the egg tray 9, meanwhile, each sucking and lifting unit is provided with an electromagnetic control valve, and the control valve is opened to finish the sucking and lifting of the duck eggs 7.
6) Whole plate egg candler 6
The whole egg candler 6 is a general outsourcing member, such as 7 x 9 whole egg candler produced by Eisen corporation; the interior of the duck egg rack contains an LED light source which can irradiate the whole tray of the duck eggs 7 simultaneously.
7) Photoelectric sensor 8
The photoelectric sensor 8 is a general-purpose outsourced component, such as a diffuse reflection type photoelectric sensor (E3Z-D61); its function is to receive and send signals to the PLC controller 12.
8) PLC controller 12
The PLC controller 12 is a general-purpose component, such as S7-300; the function of the device is to receive a trigger signal and send the signal to the computer 1, and the signal sent by the computer 1 is received to control the second motor 13 and the control valve of the whole egg sucker 5.
9) Crawler belt conveyor 11, egg tray 9, first motor 10 and second motor 13
All are general purchased parts. The egg tray 9 is used for placing the duck egg 7, the first motor 10 and the second motor 13 provide power for the device, and the crawler-type conveyor belt 11 can convey the whole tray of the duck egg 7;
the image processing software has the working flow:
a. reading an image;
b. resizing the image to 320 × 320;
c. directly importing the adjusted image into a detection model;
d. detecting clear eggs and fertilized eggs in the image by using the detection model, wherein the label of the fertilized eggs is 1, and the label of the clear eggs is 0;
e. selecting the positions of the clear eggs and the fertilized eggs in the image;
the detection result is obtained by the method:
the test samples of this example were the eggs of the Jinyun sheldrake purchased from the Shendan breeding duck breeding base in Hubei province, wherein the ratio of fertilized eggs to clear eggs was 1:1, 2700 fertilized eggs were trained, 927 fertilized eggs were tested, 261 fertilized eggs were verified, 11 duck eggs were misjudged (wherein clear eggs were judged as 7 fertilized eggs, fertilized eggs were judged as 4 clear eggs, and 0 was missed), and the accuracy and recall rate were 98.09% and 97.32%, respectively.
In the description of the present invention, it is to be understood that the terms "longitudinal", "lateral", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", and the like, indicate orientations or positional relationships based on those shown in the drawings, are merely for convenience of description of the present invention, and do not indicate or imply that the referenced devices or elements must have a particular orientation, be constructed and operated in a particular orientation, and thus, are not to be construed as limiting the present invention.
The above-described embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solutions of the present invention can be made by those skilled in the art without departing from the spirit of the present invention, and the technical solutions of the present invention are within the scope of the present invention defined by the claims.
Claims (4)
1. The utility model provides a device is selected to machine vision of no smart egg in whole dish kind duck's egg which characterized in that: the device comprises a crawler belt (11), wherein the crawler belt (11) is connected with a first motor (10) in a driving manner, a light chamber (3) is arranged at the center position of the crawler belt (11), an egg tray (9) is placed on the crawler belt (11), a plurality of seed duck eggs (7) which are hatched for 5 days are vertically placed in the egg tray (9), a whole egg sucker (5) is arranged in the light chamber (3), the whole egg sucker (5) is connected to the inner side wall of the light chamber (3) in a sliding manner, an image acquisition mechanism is arranged on the light chamber (3), and the image acquisition mechanism is in signal connection with a control mechanism;
the image acquisition mechanism is used for acquiring images of the whole tray of the duck eggs (7) and identifying clear eggs and fertilized eggs;
the control mechanism is used for controlling the whole tray egg sucker (5) to move and grab the breeding duck eggs (7) belonging to the clear eggs;
the whole-tray egg sucker (5) is used for grabbing duck eggs (7) on an egg tray (9);
the crawler-type conveyor belt (11) comprises a first conveyor belt (11-1) and a second conveyor belt (11-2), the first conveyor belt (11-1) is perpendicular to the second conveyor belt (11-2), the first conveyor belt (11-1) is arranged on the left side and the right side of the optical chamber (3), and the second conveyor belt (11-2) is arranged on the front side of the optical chamber (3);
the acquisition mechanism comprises a photoelectric sensor (8) and an industrial camera (2), the photoelectric sensor (8) is positioned at the right end of the optical chamber (3), and the industrial camera (2) is positioned at the top end of the optical chamber (3);
the bottom end of the whole-tray egg sucker (5) is provided with a whole-tray egg candler (6), and the whole-tray egg candler (6) is arranged on the bottom wall of the inner side of the light chamber (3);
a selection method of the machine vision selection device for the clear eggs in the whole tray of the duck eggs comprises the following steps: the method comprises the following steps: collecting images of the whole-tray duck eggs (7), placing the whole-tray duck eggs (7) on a crawler-type conveyor belt (11), operating a first motor (10) to enable the crawler-type conveyor belt (11) to advance, triggering a photoelectric sensor (8) when the whole-tray duck eggs (7) are conveyed into a light chamber (3), taking pictures by an industrial camera (2) to collect color images of the whole-tray duck eggs (7), and providing a light source by a whole-tray egg candler (6) and keeping the whole-tray duck eggs in a normally open state;
step two: manually labeling fertilized eggs and non-fertilized eggs in the image of the whole tray of the planted duck eggs (7), and adjusting the size of a color image to 1634 multiplied by 1234, and to 320 multiplied by 320; utilizing a Labelimage labeling tool to manually label fertilized eggs and non-fertilized eggs in the images of the whole set of duck eggs (7), wherein the labeling labels of the fertilized eggs are shoujing and the non-fertilized eggs are wujing;
step three: establishing a fertilization information identification model of the whole tray of duck eggs by using a deep learning algorithm;
step four: identifying and removing clear eggs in the image of the whole plate of the breeding duck eggs;
step three comprises a, according to 7: 3, dividing the principle into a training set and a test set, and keeping partial images as a verification set without marking;
b. the front 17 layers of the MobileNet V3_ large network are used as a basic network for automatically extracting the characteristics of the images of the whole plate of the breeding duck eggs (7);
c. a 5-layer convolutional neural network is built as an auxiliary network and used for outputting the category and position information of the image of the whole tray of the duck eggs (7), and the standard convolution is replaced by the inverse residual convolution introducing an attention mechanism;
d. b and c, constructing a complete ssdlite-MobileNet V3_ large target detection network, and inputting the image data after the training set and the test set are manually labeled into the ssdlite-MobileNet V3_ large network for training;
e. after 10000 times of training, the parameters of the ssdlite-MobileNetV3_ large network are saved and a pb file is exported, the file size being about 9.10 MB.
2. The machine vision selection device of clear eggs among a whole tray of breed duck eggs of claim 1, characterized in that: the control mechanism comprises a PLC (programmable logic controller) (12), the PLC (12) is in signal connection with a computer (1), the computer (1) is in signal connection with the industrial camera (2), and the PLC (12) is in signal connection with the photoelectric sensor (8).
3. The machine vision selection device of clear eggs among whole-plate breed duck eggs of claim 2, characterized in that: PLC controller (12) signal connection has second motor (13), second motor (13) drive is connected with a set of guide rail (4), guide rail (4) are fixed on the inside wall of light room (3), whole dish egg sucker (5) sliding connection be in on guide rail (4).
4. The machine vision selection device of clear eggs among a whole tray of breed duck eggs of claim 1, characterized in that: the fourth step comprises
The method comprises the steps that i, a pb file of a ssdlite-MobileNet V3_ large network is used for distinguishing an image of a whole tray of breeding duck eggs (7), fertilized eggs and non-fertilized eggs in the image can be identified, when the probability that one hatching egg is a non-fertilized egg is larger than 0.5, the position of the non-fertilized egg in the image is returned, and the fertilized egg is similar to the non-fertilized egg;
ii, calculating the real position of the clear eggs in the egg tray (9) by using an industrial camera (2) calibration technology;
iii, enabling the whole-tray egg sucker (5) with a control valve to fall over the breeding duck eggs (7) through a guide rail (4) by the PLC (12), and enabling the whole-tray egg sucker (5) to contact the surfaces of the breeding duck eggs (7);
iv, transmitting the real position information of the clear eggs to a PLC (programmable logic controller) (12), and controlling a valve of a whole-tray egg sucker (5) right above the clear eggs to be opened by the PLC (12) to finish sucking and lifting the clear eggs;
and v, after the non-egg-laying sucking is finished, the PLC (12) enables the whole-tray egg sucking device (5) with the control valve to return to the initial position again through the guide rail (4), the control valve of the whole-tray egg sucking device (5) is completely closed, and the non-egg falls into the second conveyor belt (11-2) to finish the nondestructive online detection and selection of the non-egg in the whole-tray duck egg (7).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010547278.6A CN111570330B (en) | 2020-06-16 | 2020-06-16 | Machine vision selection device and method for clear eggs in whole-tray duck egg planting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010547278.6A CN111570330B (en) | 2020-06-16 | 2020-06-16 | Machine vision selection device and method for clear eggs in whole-tray duck egg planting |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111570330A CN111570330A (en) | 2020-08-25 |
| CN111570330B true CN111570330B (en) | 2021-10-01 |
Family
ID=72120209
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010547278.6A Active CN111570330B (en) | 2020-06-16 | 2020-06-16 | Machine vision selection device and method for clear eggs in whole-tray duck egg planting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111570330B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113396839B (en) * | 2021-05-24 | 2022-05-20 | 杭州华杰禽业有限公司 | Egg sorting equipment before vaccine sowing |
| CN114544630A (en) * | 2022-02-25 | 2022-05-27 | 河南科技大学 | Group egg image segmentation fertilization information detection device and method based on deep learning |
| CN114532253B (en) * | 2022-04-26 | 2022-07-22 | 华南农业大学 | Automatic intelligent detection device for hatching egg activity |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO1999027362A1 (en) * | 1997-11-20 | 1999-06-03 | Fps Food Processing Systems B.V. | Method and device for detecting dirt as present on articles, for example eggs |
| CN103026984A (en) * | 2012-12-12 | 2013-04-10 | 华中农业大学 | Visual inspection grading device and method for survival rates of eggs incubated in groups |
| WO2014041416A1 (en) * | 2012-09-11 | 2014-03-20 | Visio Nerf | Device for controlling travelling parts |
| CN106645605A (en) * | 2016-09-05 | 2017-05-10 | 中国农业大学 | Poultry fertilization egg detection method based on machine vision |
| CN106872465A (en) * | 2016-12-27 | 2017-06-20 | 中国农业大学 | Birds fertile egg detection screening plant based on machine vision |
| CN107064150A (en) * | 2017-05-27 | 2017-08-18 | 华中农业大学 | A kind of brown shell infertile egg and fertile egg identification device and discrimination method based on machine vision technique |
| CN108522358A (en) * | 2018-03-09 | 2018-09-14 | 仲恺农业工程学院 | Waterfowl high-quality variety breeding intelligent decision method and system based on big data |
-
2020
- 2020-06-16 CN CN202010547278.6A patent/CN111570330B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO1999027362A1 (en) * | 1997-11-20 | 1999-06-03 | Fps Food Processing Systems B.V. | Method and device for detecting dirt as present on articles, for example eggs |
| WO2014041416A1 (en) * | 2012-09-11 | 2014-03-20 | Visio Nerf | Device for controlling travelling parts |
| CN103026984A (en) * | 2012-12-12 | 2013-04-10 | 华中农业大学 | Visual inspection grading device and method for survival rates of eggs incubated in groups |
| CN106645605A (en) * | 2016-09-05 | 2017-05-10 | 中国农业大学 | Poultry fertilization egg detection method based on machine vision |
| CN106872465A (en) * | 2016-12-27 | 2017-06-20 | 中国农业大学 | Birds fertile egg detection screening plant based on machine vision |
| CN107064150A (en) * | 2017-05-27 | 2017-08-18 | 华中农业大学 | A kind of brown shell infertile egg and fertile egg identification device and discrimination method based on machine vision technique |
| CN108522358A (en) * | 2018-03-09 | 2018-09-14 | 仲恺农业工程学院 | Waterfowl high-quality variety breeding intelligent decision method and system based on big data |
Non-Patent Citations (1)
| Title |
|---|
| 基于深度学习的种鸭蛋孵化早期受精信息无损检测;李庆旭;《农业机械学报》;20191112;第188-193页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111570330A (en) | 2020-08-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111570330B (en) | Machine vision selection device and method for clear eggs in whole-tray duck egg planting | |
| CN102217559B (en) | On-line automatic detection and sorting equipment and method for incubated egg quality | |
| CN100562245C (en) | Seedling replanting system based on machine vision | |
| CN102835329B (en) | Automatic hatching egg identification and sorting equipment and method | |
| CN103026984B (en) | Visual inspection grading device and method for survival rates of eggs incubated in groups | |
| CN100366147C (en) | Methods and apparatus for selectively processing eggs having identified characteristics | |
| CN100352343C (en) | Methods and apparatus for identifying live eggs by detecting embryo heart rate and/or motion | |
| CN110443262B (en) | Computer vision-based rapid nondestructive testing method and device for freshness of fish body | |
| CN102179374B (en) | Automatic detecting and sorting device for poultry egg quality and method thereof | |
| CN103752531B (en) | A kind of Nicotiana tabacum L. sorter based on machine vision | |
| CN208283385U (en) | Underwater detecting robot for culture fishery | |
| CN107064150B (en) | Machine vision technology-based brown shell clear egg and fertilized egg identification device and identification method | |
| CN208976281U (en) | A kind of bar code recognition sorting system | |
| CN202037127U (en) | Automatic egg quality detection and separation equipment | |
| CN116636481A (en) | Intelligent determination method and equipment for egg laying performance and egg quality of poultry | |
| CN113569971B (en) | Image recognition-based catch target classification detection method and system | |
| CN109663747A (en) | A kind of Chinese chestnut small holes caused by worms intelligent detecting method | |
| CN110208472A (en) | A kind of automatic inspection egg machine | |
| CN110115234A (en) | A kind of the aquaculture management monitoring system and intelligent weighing method of intelligent weighing | |
| CN117461575A (en) | Intelligent healthy breeding data system for domestic pigs and platform thereof | |
| CN109507204A (en) | A kind of Fresh Grade Breast lignifying stage division and its device based on curvature detection | |
| CN203786131U (en) | Vaginitis detection work station | |
| CN109158329A (en) | A kind of equipment for the vanning of dress-goods automatic sorting | |
| CN202999054U (en) | Visual inspection grading device for survival rate of group hatched eggs | |
| CN111802281B (en) | Optical fiber spectrum grading detection device and method for fertilized eggs in pre-incubated breeding duck eggs |
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 |