CN110887707B - Real-time monitoring system and method for impurity-containing crushing state of grains based on U-Net network - Google Patents
Real-time monitoring system and method for impurity-containing crushing state of grains based on U-Net network Download PDFInfo
- Publication number
- CN110887707B CN110887707B CN201910804445.8A CN201910804445A CN110887707B CN 110887707 B CN110887707 B CN 110887707B CN 201910804445 A CN201910804445 A CN 201910804445A CN 110887707 B CN110887707 B CN 110887707B
- Authority
- CN
- China
- Prior art keywords
- grain
- grains
- net network
- pixel
- impurity
- 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
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N1/00—Sampling; Preparing specimens for investigation
- G01N1/02—Devices for withdrawing samples
- G01N1/10—Devices for withdrawing samples in the liquid or fluent state
- G01N1/20—Devices for withdrawing samples in the liquid or fluent state for flowing or falling materials
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N2021/8466—Investigation of vegetal material, e.g. leaves, plants, fruits
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
Abstract
The invention discloses a grain collecting device and a grain impurity-containing crushing state real-time monitoring system and method based on a U-Net network.A rotating plate of the grain collecting device is driven by a stepping motor to realize the opening and closing of a side panel of a sampling box; during grain collection, the ARM processor controls the camera to shoot images, an improved U-Net network model is called to conduct image segmentation, the pixel area of each label is obtained, the CAN bus of the ARM processor receives the water content from the upper computer, the grain impurity content rate and the breakage rate are comprehensively converted out, data are stored locally and displayed through the display screen, and the data are sent to the upper computer through the CAN peripheral equipment. The data stored locally is beneficial to further off-line research, the grain images can be rapidly processed by improving the U-Net network, and the robustness of the images shot under different illumination conditions is strong; the invention can also adjust the relevant operation parameters of the combine harvester such as the gap of the concave plate, the rotating speed of the separating roller, the rotating speed of the fan and the like in real time.
Description
Technical Field
The invention relates to the field of combine harvester working performance and image segmentation, in particular to a method and a system for monitoring the impurity-containing crushing state of grains in real time based on a U-Net network.
Background
When the combine harvester works, grains are crushed when abnormal working parameters such as overlarge gap between concave plates, excessively high rotating speed of a separating roller and the like occur; when the fan rotates too slowly and the angle of the air distributing plate is improper, a large amount of impurities such as branches and stalks can be mixed in the grains. The grain breakage rate and the impurity rate are two important parameter indexes for measuring the grain quality.
Most of devices and detection modules developed by foreign agricultural machinery are integrated into multi-module mass production, and the price is high, so that the popularization of the national precision agriculture is not facilitated. The study of domestic scholars on impurity rate and breakage rate of grains is mostly carried out in a laboratory, only a few scholars study devices and processing methods for real-time monitoring, but the sampling is not sufficient when grain samples are collected, grain releasing is not rapid enough, and the conditions of false identification and false segmentation of grains and impurities exist by applying a traditional image segmentation method. For example: the K-Means algorithm and the watershed algorithm are combined for image segmentation, and then a BP neural network model is used for identifying sundry and broken grains, but the watershed algorithm has an over-segmentation phenomenon, is complex, and has large calculation amount and long time consumption of the BP neural network.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a grain collection device, a grain impurity-containing crushing state real-time monitoring method and a grain impurity-containing crushing state real-time monitoring system based on a U-Net network, and the grain and impurity segmentation precision is improved.
The invention adopts the following technical scheme:
the utility model provides a cereal collection device, includes that the top is fixed the sampling box on a grain outlet support, the box body of sampling box is for leaking hopper-shaped, and the box body bottom designs into the slope form, is equipped with the rotor plate near the sampling box side of grain outlet department, can detain with the buckle on the sampling box side when the rotor plate rotates, and the rotor plate is controlled by step motor.
The grain impurity-containing crushing state real-time monitoring system based on the U-Net network comprises the grain collecting device and an ARM processor, wherein the ARM processor is in signal connection with the grain collecting device, an image collecting device and a display screen and is also communicated with an upper computer; and the ARM processor calls an improved U-Net network to combine with the water content to obtain the impurity content and the breakage rate of the grains.
In the technical scheme, the Dropout is added after the convolution of the U-Net network, and the number of the filters is modified.
In the technical scheme, the improved U-Net network segments the grain image acquired by the image acquisition device to obtain a pixel a of broken grains, a pixel b of intact grains and a pixel c of stem branches.
In the technical scheme, the impurity content of the grains
The grain breakage rate
Wherein v is a pixel corresponding to 1 g of grains, and u is a pixel corresponding to 1 g of branch, stalk and stalk impurities.
A grain impurity-containing crushing state real-time monitoring method based on a U-Net network is characterized in that a grain image acquired by an image acquisition device is sent to an ARM processor, pixels a of crushed grains, pixels b of intact grains and pixels c of stem branches obtained after the processing of the ARM processor are combined with the water content sent by an upper computer, and the impurity-containing rate and the crushing rate of grains are obtained.
The invention has the beneficial effects that:
1. the real-time monitoring system for the impurity-containing crushing state of the grain can be used for collecting and releasing the grain in real time by controlling the rotating speed direction of the stepping motor to drive the rotating plate of the sampling box when the combine harvester works, the grain at the grain outlet can be fully collected by the structural design and the installation mode of the sampling box, and the process of collecting the grain by closing the sampling box is more stable by the round hole of the rotating plate and the buckle design above the sampling box. Sampling box bottom slope, sample cereal can flow out more fast from the box, and step motor's voltage and electric current are little, approximate the current-voltage value of other modules, and the concentrated power supply is safer.
2. Compared with the traditional image segmentation, the improved U-Net network provided by the invention has a simple structure, maximally retains the information of impurities of complete grains, broken grains, branches and stalks, and has better robustness to pictures shot under different illumination conditions; the class label smooth regularization network model improves the adaptability of the model and can improve the classification precision.
3. The invention displays the impurity rate and the crushing rate on the display screen in real time, so that the working personnel can adjust the gap of the concave plate and other performance parameters such as the rotating speed of the feeding auger; in addition, the impurity rate and the breakage rate are saved, so that the further research and analysis after offline are facilitated, and powerful data support is provided for building a control mechanism motion model.
4. The whole system device is convenient to install and simple to operate, and can be controlled to work only by one ARM processor.
Drawings
FIG. 1 is a schematic structural diagram of a real-time monitoring system for impurity-containing crushing state of grains based on a U-Net network;
FIG. 2 is a schematic view showing the installation of the grain collecting device of the present invention, FIG. 2(a) is a schematic view showing the installation of the grain collecting device, and FIG. 2 (b) is a detailed view showing the installation of the stepping motor of the grain collecting device;
FIG. 3 is a network model training flow diagram of the present invention;
fig. 4 is a structure diagram of the improved U-Net network of the present invention.
The notation in the figure is: the method comprises the following steps of 1-a camera, 2-an image shooting chamber, 3-a light source, 4-a transparent acrylic plate, 5-a stepping motor, 6-an inclined thin plate, 7-a mounting bracket A, 8-a buckle, 9-a screw rod, 10-a rotating plate, 11-a grain inlet, 12-a grain outlet, 13-a mounting bracket B and 14-a coupler.
Detailed Description
The technical scheme of the invention is further explained by combining the attached drawings.
As shown in figure 1, the grain impurity-containing and breakage state real-time monitoring system based on the U-Net network comprises a grain collecting device, an image collecting device and a display device, wherein the image collecting device comprises a camera 1 and a light source 3, the display device selects a display screen, the display screen is installed in a cab and is connected with an ARM processor, and the current grain impurity-containing rate and breakage rate data are displayed in real time.
The ARM processor is positioned in a cab of the harvester, controls the forward rotation and the reverse rotation of the stepping motor 5 and the image shooting frequency of the camera 1, processes grain images acquired by the camera 1 in real time, calls a U-Net network to perform image segmentation, displays and saves the impurity rate and the breakage rate of grains on a display screen, and is communicated with an upper computer through a CAN bus; and the image acquired by the camera 1 and the processing result of the ARM processor are locally stored in real time.
The CAN bus is communicated with the upper computer, and the moisture content of the upper computer is received through the CAN bus peripheral of the ARM processor, and the impurity-containing rate and the breakage rate are calculated and then sent to the upper computer. At first, the CAN controller of the ARM processor enters a reset mode, a GPIO pin used for CAN communication is initialized, and baud rate, working mode and interruption are set, wherein:
wherein, TqFor each communication node on the bus, N occupies T for each data bitqThe number of (2); and after the initialization is finished, exiting the reset mode and setting the filter.
The CAN controller receives the water content data, namely the CAN receives the message of the upper computer, and the message receiving mode comprises an inquiry control receiving mode and an interrupt control mode; the CAN controller sends the data of the impurity rate and the breakage rate to an upper computer, and sets a message to be sent, including the contents of the ID, the extended ID, the IDE bit and the RTR bit of the message, the DLC section and the message data section. When a request for sending a message is received, periodically inquiring the state of a sending buffer, if the sending buffer is locked, waiting until the sending buffer is released by the CAN controller; if the sending buffer is released, the CAN controller writes the message into the sending buffer and sets a sending request TR mark in the command register, and finally starts to send the message.
FIG. 2 is a schematic view of the installation of the grain collecting device, which comprises a sampling box and a stepping motor 5, wherein the top end of the sampling box is fixed on a bracket of a grain outlet 12 through a mounting bracket A7, the top end of the sampling box is a grain inlet 11, and the grain throwing direction is opposite to the grain inlet 11 of the sampling box; the sampling box is characterized in that the box body of the sampling box is funnel-shaped, the bottom end of the box body is designed to be inclined, namely the sampling box is formed by closing an inclined thin plate 6 and the side surface of the box body, the side surface of the sampling box, which is close to a grain outlet 12, is provided with a rotating plate 10, the length of the upper end of the rotating plate 10 is greater than the vacancy of the side surface of the sampling box, so that a round hole in the rotating plate 10 can be buckled with a buckle 8 on the side surface of the sampling box, the rotating plate 10 is fixed on the sampling box through a lead screw 9, the end part, which is close to the sampling box, of the lead screw 9 is fixed with a coupler 14, the coupler 14 is also fixed with a motor shaft of a stepping motor 5, and the stepping motor 5 is fixed on the sampling box through a mounting bracket B13; the rotating plate 10 is controlled by the stepping motor 5, initializes GPIO and is set as a control pin of the stepping motor 5, determines the excitation mode of the stepping motor 5, drives each beat, controls the output time sequence of the pin to control the rotating direction, and sets the delay time of each beat to control the rotating speed; the stepping motor 5 rotates reversely to drive the rotating plate 10 to overturn to a vertical position, the side panel of the sampling box is opened, and grains at the bottom of the box flow out rapidly along the gradient of the inclined thin plate; when the grain in the sampling box is released, the ARM processor controls the stepping motor 5 to rotate forward for a certain angle, the rotating plate 10 rotates to the side face of the box body, the round hole buckles the side face box body buckle 8, the sampling box is closed in the process, and sampling is repeated.
Inside being close to grain inlet 11 department of sampling box is equipped with confined image room 2 of making a video recording, and transparent acrylic plate 4 is adopted in the bottom of image room 2 of making a video recording, and all the other sides are the same with sampling box body material, are equipped with light source 3 on the 2 sides of image room of making a video recording, and light source 3 crosses the wire and links to each other with external power source, and 2 tops of image room of making a video recording are fixed with camera 1, and camera 1's trigger signal is controlled by the ARM treater.
A real-time monitoring method for the impurity-containing crushing state of grains based on a U-Net network comprises the following steps:
s1: the grain collecting device is fixed with a grain outlet 12 bracket of the harvester through four mounting brackets A7 on the box body, the rotating plate 10 is controlled to be buckled with the buckle 8, and the light source 3 is turned on; when the grain outlet 12 of the harvester is thrown out by grains, the grain inlet 11 can collect the grains to the maximum extent, the grains flow into the sampling box along the side surface of the box body, the time of accumulating the grains to the transparent acrylic plate 4 can be calculated through empirical parameters, and the ARM processor triggers the camera 1 to shoot the image of the current grains.
S2: utilizing an improved U-Net network (in an ARM processor) to segment the grain image, and calculating pixels of segmented stems, stalks, broken grains and intact grains to obtain the impurity content and the breakage rate of the grains;
s2.1: improvement of original U-Net network
The original U-Net network consists of an input layer, a convolution layer, a pooling layer, an up-sampling layer and an output layer, an original structure is easy to generate an overfitting condition, robustness and generalization capability are not ideal, and Dropout is added after convolution to improve regularization of the network.
As shown in fig. 4, the improved U-Net network design structure is left-right symmetric, and has 11 convolutional layers, 2 maximum pooling layers, 2 upsampling layers, and 11 RELU layers, 2 cuts and copies, the left side is a contraction path, mainly having 4 convolutional layers and 2 pooling layers, and the right side is an expansion path, mainly having 5 convolutional layers and 2 upsampling layers.
The first module contains Conv _1, convolution kernel 3 x 3, filters 32, 1 RELU layer, 1 Dropout; conv _1, convolution kernel 3 × 3, filter 32, 1 RELU layer; 1 maximum pooling layer, 2 × 2 convolution kernels, 32 filters;
the second module contains Conv _3, convolution kernel 3 x 3, 64 filters, 1 RELU layer, 1 Dropout; conv _4, convolution kernel 3 × 3, 64 filters, 1 RELU layer; 1 max pooling layer, convolution kernel 2 x 2, 64 filters;
the third module comprises Conv _5, convolution kernel 3 x 3, 64 filters, 1 RELU layer, 1 Dropout; conv _6, convolution kernel 3 × 3, 64 filters, 1 RELU layer;
the fourth module comprises 1 upsampling layer, 2 × 2 convolution kernels and 128 filters; conv _7, convolution kernel 3 × 3, 192 filters, 1 RELU layer, 1 Dropout; conv _8, convolution kernel 3 × 3, 64 filters, 1 RELU layer;
the fifth module comprises 1 upsampling layer, convolution kernel 2 x 2 and 64 filters; conv _9, convolution kernel 3 × 3, filters 96, 1 RELU layer, 1 Dropout; conv _10, convolution kernel 3 × 3, filter 32, 1 RELU layer;
the sixth module contains 1 Conv _11, convolution kernel 1 x 1, filter 3, 1 RELU layer.
Wherein the Conv _4 feature information copy is superimposed on Conv _7 and the Conv _2 feature information copy is superimposed on Conv _ 9. The output layer adopts Softmax as an activation function, and specifically comprises the following steps:
wherein: z[L]Is the output layer vector, σ is the K-dimensional real vector.
S2.2: training improved U-Net network
As shown in fig. 3, S2.2.1: data preparation
The grain image is divided into a training set and a testing set, each existing grain image is cut into 800 × 800 sizes by taking different areas at will, the training set is subjected to image preprocessing, category labels are manually marked according to broken grains, intact grains and stem branches, and data are subjected to normalization processing. The image preprocessing mainly comprises operations of image enhancement, gray level transformation, adaptive histogram equalization and the like, and the image enhancement comprises image turning, distortion, translation, random noise addition and the like.
S2.2.2: training data
Sending the training set data in S2.2.1 into an improved U-Net network, improving the network learning efficiency by using an Adam optimization algorithm, optimizing the training result, and setting parameters: the method comprises the following steps of (1) continuously updating weight and parameters by using an initial learning rate, a training batch size, a learning rate reduction coefficient, a maximum iteration number and a Dropout value, and taking the cross entropy of the class label regularization LSR as a loss function, wherein the method specifically comprises the following steps:
wherein: y isi,kProbability of true value; epsilon is a smoothing parameter, and epsilon is 0.1 in the embodiment; u (k) is a prior distribution of k class labels, the present invention assumes a uniform distribution of class labels,
pi,ka probability of predicting a kth class label value for the ith sample;
and when the recorded training Loss value Loss does not decrease any more, stopping training, inputting a test set to verify the reliability of the network, and finally storing the trained U-Net network.
The step length updating function of the Adam optimization algorithm specifically comprises the following steps:
wherein: alpha is the initial learning rate;
is a gradient mean value;
is the gradient variance; beta is a constant, usually 10-8。
S2.2.3: testing
And sending the test set data into the trained U-Net network, verifying and improving the reasonability of the U-Net network, and storing.
And finally, sending the grain image acquired by the camera 1 into a trained improved U-Net network for image segmentation.
S2.3: s2.2 is carried out for a plurality of times, until the loss function no longer descends, the pixel a of the broken cereal that obtains, the pixel b of intact cereal, the pixel c of stem branch stalk combine present cereal moisture content d (the host computer sends through the CAN bus) to obtain cereal trash content rate and percentage of damage, show on the display screen, send simultaneously for the host computer and save locally, specifically do:
wherein: v is the pixel corresponding to 1 g of grains, and u is the pixel corresponding to 1 g of branch, stalk and stalk impurities.
S3: when the impurity rate and the breakage rate of the grains are displayed on the display screen, a driver adjusts the gap of the concave plate, the rotating speed of the threshing cylinder and the rotating speed of the fan of the combine harvester according to the impurity rate and the breakage rate.
The present invention is not limited to the above-described embodiments, and any obvious improvements, substitutions or modifications can be made by those skilled in the art without departing from the spirit of the present invention.
Claims (7)
1. Cereal contains miscellaneous broken state real-time monitoring system based on U-Net network, its characterized in that: the grain collection device comprises a grain collection device and an ARM processor, wherein the ARM processor is in signal connection with the grain collection device, an image acquisition device and a display screen and is also communicated with an upper computer; the ARM processor calls an improved U-Net network to combine with the water content to obtain the impurity content and the breakage rate of the grains;
the improved U-Net network is that Dropout is added after the convolution of the U-Net network, and the number of filters is modified;
the grain impurity content
Wherein v is a pixel corresponding to 1 g of grains, and u is a pixel corresponding to 1 g of branch, stem and stalk impurities;
the grain breakage rate
Wherein: a is a pixel of broken grains, b is a pixel of intact grains, c is a pixel of stems and branches, d is the water content of the current grains, v is a pixel corresponding to 1 g of grains, and u is a pixel corresponding to 1 g of stem and stem impurities.
2. The U-Net network-based cereal impurity-containing breakage state real-time monitoring system according to claim 1, characterized in that: after the improved U-Net network segments the grain image collected by the image collecting device, a pixel a of broken grains, a pixel b of intact grains and a pixel c of stem branches are obtained.
3. The U-Net network-based cereal impurity-containing breakage state real-time monitoring system according to claim 1, characterized in that: cereal collection device includes that the top is fixed the sampling box on a grain outlet (12) support, the box body of sampling box is for leaking hopper-shaped, and the box body bottom designs into the slope form, and the sampling box side that is close to a grain outlet (12) department is equipped with rotor plate (10), can detain with buckle (8) on the sampling box side when rotor plate (10) rotate, and rotor plate (10) are controlled by step motor (5).
4. A monitoring method of a U-Net network based cereal impurity-containing crushing state real-time monitoring system according to any one of claims 1-3, characterized in that: the grain image collected by the image collecting device is sent to the ARM processor, and the grain impurity content and the breakage rate are obtained by combining the water content sent by the upper computer with the pixel a of the broken grains, the pixel b of the intact grains and the pixel c of the stem branches obtained after the processing of the ARM processor.
5. The monitoring method according to claim 4, wherein: the pixels a of the broken grains, the pixels b of the intact grains and the pixels of the stems and the stalks are obtained by processing an improved U-Net network in an ARM processor.
6. The monitoring method according to claim 5, wherein: the grain impurity content
Wherein v is a pixel corresponding to 1 g of grains, u is a pixel corresponding to 1 g of branch, stalk and stalk impurities, and d is the water content of the current grains.
7. The monitoring method according to claim 6, wherein: the grain breakage rate
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910804445.8A CN110887707B (en) | 2019-08-28 | 2019-08-28 | Real-time monitoring system and method for impurity-containing crushing state of grains based on U-Net network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910804445.8A CN110887707B (en) | 2019-08-28 | 2019-08-28 | Real-time monitoring system and method for impurity-containing crushing state of grains based on U-Net network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110887707A CN110887707A (en) | 2020-03-17 |
| CN110887707B true CN110887707B (en) | 2022-03-22 |
Family
ID=69745897
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910804445.8A Active CN110887707B (en) | 2019-08-28 | 2019-08-28 | Real-time monitoring system and method for impurity-containing crushing state of grains based on U-Net network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110887707B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112014399B (en) * | 2020-07-22 | 2023-08-25 | 湖北工业大学 | Belt-type grain crushing rate and impurity content detection device and method in grain tank |
| CN113298085B (en) * | 2021-04-06 | 2024-03-22 | 江苏大学 | Corn impurity-crushing identification method and system based on Mask R-CNN |
| CN113644298B (en) * | 2021-07-09 | 2022-08-23 | 江苏大学 | Test device for simulating water drive in gas diffusion layer of fuel cell and control method |
| CN115399140B (en) * | 2022-10-28 | 2023-03-24 | 潍柴雷沃智慧农业科技股份有限公司 | Working parameter self-adaptive adjusting method and system of combine harvester |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103706574A (en) * | 2013-12-25 | 2014-04-09 | 中国科学院半导体研究所 | Automatic solid grain sorting system |
| CN104392430A (en) * | 2014-10-22 | 2015-03-04 | 华南农业大学 | Machine vision-based super hybrid rice bunch seeding quantity detection method and device |
| CN105806751A (en) * | 2016-03-24 | 2016-07-27 | 江苏大学 | On-line monitoring system and method for crushing of cereals in grain tank of combine harvester |
| CN106404679A (en) * | 2016-09-26 | 2017-02-15 | 江苏大学 | Device and method for monitoring impurity rate and breakage rate of grains in grain tank |
| CN107909581A (en) * | 2017-11-03 | 2018-04-13 | 杭州依图医疗技术有限公司 | Lobe of the lung section dividing method, device, system, storage medium and the equipment of CT images |
-
2019
- 2019-08-28 CN CN201910804445.8A patent/CN110887707B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103706574A (en) * | 2013-12-25 | 2014-04-09 | 中国科学院半导体研究所 | Automatic solid grain sorting system |
| CN104392430A (en) * | 2014-10-22 | 2015-03-04 | 华南农业大学 | Machine vision-based super hybrid rice bunch seeding quantity detection method and device |
| CN105806751A (en) * | 2016-03-24 | 2016-07-27 | 江苏大学 | On-line monitoring system and method for crushing of cereals in grain tank of combine harvester |
| CN106404679A (en) * | 2016-09-26 | 2017-02-15 | 江苏大学 | Device and method for monitoring impurity rate and breakage rate of grains in grain tank |
| CN107909581A (en) * | 2017-11-03 | 2018-04-13 | 杭州依图医疗技术有限公司 | Lobe of the lung section dividing method, device, system, storage medium and the equipment of CT images |
Non-Patent Citations (5)
| Title |
|---|
| Design of sampling device for rice grain impurity sensor in grain-bin of combine harvester;Chen Jin等;《农业工程学报》;20190331;第18-25页 * |
| IMAGE PROCESSING BASED ANN WITH BAYESIAN REGULARIZATION LEARNING ALGORITHM FOR CLASSIFICATION OF WHEAT GRAINS;Ahmet Kayabasi;《IEEE 10th International Conference on Electrical and Electronics Engineering》;20171231;第1166-1170页 * |
| 基于机器视觉的水稻杂质及破碎籽粒在线识别方法;陈进等;《农业工程学报》;20180731;第187-194页 * |
| 基于深度卷积神经网络的小麦赤霉病高光谱病症点分类方法;金秀等;《浙江农业学报》;20190225;第315-325页 * |
| 采用Ｕ－Ｎｅｔ卷积网络的桥梁裂缝检测方法;朱苏雅等;《西安电子科技大学学报》;20190820;第35-42页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110887707A (en) | 2020-03-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110887707B (en) | Real-time monitoring system and method for impurity-containing crushing state of grains based on U-Net network | |
| US11275011B2 (en) | Automated airborne particulate matter collection, imaging, identification, and analysis | |
| Fu et al. | Image recognition method of multi-cluster kiwifruit in field based on convolutional neural networks | |
| US10716255B2 (en) | Adaptive control system for threshing separation load of tangential flow and longitudinal axial flow device | |
| CN104737707A (en) | Combine harvester cleaning impurity rate self-adaptive control device and self-adaptive control cleaning method | |
| WO2016138675A1 (en) | Combine harvester self-adaptive cleaning control apparatus and self-adaptive cleaning method thereof | |
| CN103418554B (en) | Based on control system and the control method of the small-sized Intelligent agricultural product separator of DSP | |
| CN204741825U (en) | Combine harvester cleans dirt percentage adaptive control device | |
| CN204733622U (en) | Controlling means is cleaned to combine harvester self -adaptation | |
| CN108107049A (en) | Combined harvester tanker seed percentage of impurity and percentage of damage real-time monitoring device and method | |
| CN105806751A (en) | On-line monitoring system and method for crushing of cereals in grain tank of combine harvester | |
| CN113298085B (en) | Corn impurity-crushing identification method and system based on Mask R-CNN | |
| CN111359907B (en) | Real-time online detection system and method for wheat scab infection particle rate based on embedded deep learning | |
| Mahirah et al. | Monitoring harvested paddy during combine harvesting using a machine vision-Double lighting system | |
| CN109566111A (en) | Threshing and cleaning testing stand | |
| CN114882468A (en) | Self-adaptive adjustment method for angle of sweeping brush of sweeper | |
| CN113330915A (en) | Self-adaptive cotton harvesting method based on binocular vision recognition and intelligent mechanical harvesting device | |
| CN108090910B (en) | Night outdoor tomato plant image segmentation algorithm based on information entropy gradient simplified PCNN model | |
| CN207336378U (en) | A kind of device of automatic detection rice phenotypic parameter | |
| CN207147533U (en) | A kind of Miniature digital rice species test machine | |
| CN112017208A (en) | Multicellular ball identification and classification method based on deep learning | |
| US20220375228A1 (en) | Residue Spread Monitoring | |
| GB2606741A (en) | Residue spread monitoring | |
| Pezzementi et al. | Going against the grain: Real-time classification of grain quality | |
| CN220819050U (en) | Intelligent pest and disease damage identification and green fertilization detection system based on solar scanner |
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 |