Farmland residual film pollution monitoring system and application method thereof

Abstract

The application belongs to the technical field of farmland residual film pollution supervision, and is a farmland residual film pollution supervision system, which mainly comprises a post-harvest residual film recovery machine, a pre-spring sowing residual film recovery machine, a server, a mobile user terminal and a UAV; the application method is to use the post-harvest residual film recovery machine, the installed post-harvest residual film recovery supervision system, the UAV, the UAV residual film pollution verification system, the pre-spring sowing residual film recovery machine and the installed pre-spring sowing residual film recovery supervision system, obtain images after residual film recovery through a motion camera and an on-board camera, and obtain the residual film coverage rate of the farmland after recovery on a farmland residual film pollution supervision information platform; the application combines deep learning, image processing and Internet of Things technology, uses fast data acquisition, uploading and processing, realizes efficient and accurate monitoring and verification of farmland residual film pollution at different stages, and is an efficient, convenient, economical and practical farmland residual film pollution supervision system and application method.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of farmland residual film pollution supervision, and particularly relates to a farmland residual film pollution supervision system and an application method. BACKGROUND

[0002] The mulching cultivation technology has brought huge benefits to the cotton production in Xinjiang, and has greatly solved the problem of weak agricultural production capacity in arid and cool regions. However, the mulching film is difficult to recycle, and the residual mulching film causes serious pollution to farmland and leads to soil fertility decline.

[0003] Mechanized recycling is an important way to control residual film pollution. Due to the use of mulching film that does not meet the requirements, it is very easy to break during the mechanized recycling process, resulting in a large amount of residual film that is not easy to recycle. The execution of existing laws, regulations and standards is not enough, and the supervision and restraint of users of mulching film that does not meet the standards need to be further strengthened.

[0004] The mechanized recycling of residual film in farmland is divided into two stages, the first stage is the residual film recycling process after crop harvest, and the second stage is the residual film recycling again before crop sowing in spring. There is no mature farmland residual film pollution supervision system available at present. Compared with other detection methods currently available, this method has the characteristics of low labor intensity, short evaluation period, simple operation, wide supervision range, high data accuracy, and data can be checked at any time. It can comprehensively monitor and verify the residual film recycling process in different stages, provide certain theoretical and technical support for the informationization supervision of residual film recycling, protect the soil environment of farmland, and promote the sustainable development of agricultural industry. SUMMARY

[0005] One of the purposes of the application is to provide a farmland residual film pollution supervision system that is efficient, convenient, economical and practical, can realize the rapid evaluation of the quality of residual film recycling after harvest, the monitoring and evaluation of residual film pollution before spring sowing, and the use of unmanned aerial vehicle residual film pollution evaluation and verification, and build a unified information platform for data collection in the process of residual film recycling. Another purpose of the application is to provide an application method of the farmland residual film pollution supervision system that is efficient, convenient, economical and practical, combines deep learning, image processing and Internet of Things technology, uses fast data collection, uploading and processing, realizes efficient and accurate monitoring and verification of farmland residual film pollution in different stages, and provides reference data for the decision of whether the film mulching crop field is fallow during spring sowing.

[0006] In order to achieve the above purpose, the application provides the following technical solutions:

[0007] A farmland residual film pollution supervision system mainly comprises a residual film recycling machine after harvest, a residual film recycling machine before spring sowing, an unmanned aerial vehicle, a server and a mobile user terminal.

[0008] The post-harvest residual film recovery machine is provided with a motion camera, a Beidou positioning device, an industrial computer and a 4G DTU module; the motion camera is provided with two, which are installed on the left and right sides of the cab of the machine, and is used for collecting images of farmland before and after the recovery of residual films; the Beidou positioning device is installed above the cab of the machine, and is used for collecting position information of the land; the industrial computer is installed in the cab of the machine, and is used for installing a post-harvest residual film recovery supervision system; and the 4G DTU module is installed on the industrial computer and transmits data in a wireless transmission mode.

[0009] The pre-spring planting residual film recovery machine is provided with a motion camera, a Beidou positioning device, an industrial computer and a 4G DTU module; the motion camera is provided with two, which are installed on the left and right sides of the cab of the machine, and is used for collecting images of farmland before and after the recovery of residual films; the Beidou positioning device is installed above the cab of the machine, and is used for collecting position information of the land; the industrial computer is installed in the cab of the machine, and is used for installing a pre-spring planting residual film recovery supervision system; and the 4G DTU module is installed on the industrial computer and transmits data in a wireless transmission mode.

[0010] The unmanned aerial vehicle is provided with a flight controller, an on-board camera, a GPS sensor and a data transmission link; the flight controller is used for stabilizing the flight state and executing flight instructions; the on-board camera is installed on a gimbal and is used for shooting images of farmland after the recovery of residual films by the residual film recovery machine; the GPS sensor is built into the unmanned aerial vehicle and is used for collecting position information of the land; and the data transmission link is used for transmitting photos shot by the on-board camera and position information of the land.

[0011] The server is installed with a farmland residual film pollution supervision information platform, which is used for identifying images of farmland before and after the recovery of residual films collected by the post-harvest residual film recovery machine, images of farmland before and after the recovery of residual films collected by the pre-spring planting residual film recovery machine, and images of farmland after the recovery of residual films shot by the unmanned aerial vehicle, and calculating the proportion of residual films in the images, recording position information of the recovery land and farmer information, and being available for checking at any time.

[0012] The mobile user terminal is installed with an unmanned aerial vehicle residual film pollution checking system, which is used for checking residual film residues after the recovery of the post-harvest residual film recovery machine and the pre-spring planting residual film recovery machine, and recording position information of the recovery land, farmer information, farmland area, basic information of the use of mulching film and the recovery mode of the unmanned aerial vehicle; the mobile user terminal is also used for deep learning training to obtain a model for identifying residual film images, logging in the cloud to check information and data at any time.

[0013] The post-harvest residual film recovery supervision system is used for recording images of farmland before and after the recovery of residual films, position information of the farmland, owner information of the post-harvest residual film recovery machine and farmer information of the farmland.

[0014] The spring pre-sowing residual film recovery supervision system is used for recording the images of the farmland before and after the recovery of the residual film, and the position information of the farmland, the information of the owner of the spring pre-sowing residual film recovery machine and the information of the owner of the farmland.

[0015] The mobile user terminal carries out deep learning training to obtain a model for identifying the residual film image, that is, the mobile user terminal divides the data into a training set, a test set and a verification set by using the residual film image information data on the farmland residual film pollution supervision information platform, carries out residual film labeling, carries out model training by using a deep learning network, and obtains the best identification model, wherein the model is a vehicle-mounted residual film identification model and a UAV residual film identification model.

[0016] An application method of the farmland residual film pollution supervision system, which utilizes the post-harvest residual film recovery machine and the installed post-harvest residual film recovery supervision system, the UAV, the UAV residual film pollution verification system, the spring pre-sowing residual film recovery machine and the installed spring pre-sowing residual film recovery supervision system, obtains the image after the recovery of the residual film by using a motion camera and an on-board camera, and obtains the residual film coverage rate of the farmland after the recovery on the farmland residual film pollution supervision information platform; mainly includes the following steps:

[0017] (1) when the residual film recovery is carried out after the harvest, the post-harvest residual film recovery machine is started, the field residual film image after the recovery is shot by using a motion camera, the residual film recovery machine is required to maintain a stable speed during the driving, and the image after the recovery of the residual film is collected in sequence according to the planned route;

[0018] (2) the image after the recovery of the residual film stored in the post-harvest residual film recovery supervision system is uploaded to the farmland residual film pollution supervision information platform, the image after the recovery of the residual film is identified by using a deep learning technology, the pixel point proportion of the residual film is calculated, and the average coverage rate of the residual film after the recovery of the post-harvest residual film recovery machine is obtained;

[0019] (3) after the post-harvest residual film recovery machine stops the operation after the recovery, the UAV is started, the sampling task is sent to the flight controller by using the UAV residual film pollution verification system, and the field residual film verification image after the recovery is shot by using an on-board camera;

[0020] (4) the field residual film verification image after the recovery of the residual film stored in the UAV residual film pollution verification system is uploaded to the farmland residual film pollution supervision information platform, the field residual film verification image after the recovery of the residual film is identified by using a deep learning technology, the pixel point proportion of the residual film is calculated, and the average coverage rate of the residual film after the verification of the recovery of the post-harvest residual film recovery machine is obtained;

[0021] (5) When the pre-spring ploughing residual film recycling machine is started, the residual film image after recycling is photographed by a motion camera, and the residual film recycling machine is required to maintain a stable speed during driving and collect the image after recycling residual film according to the planned route;

[0022] (6) The residual film image after recycling stored in the pre-spring ploughing residual film recycling supervision system is uploaded to the farmland residual film pollution supervision information platform, the deep learning technology is used to identify the residual film image after recycling, and the pixel point proportion of the residual film is calculated to obtain the average coverage rate of the residual film after recycling by the pre-spring ploughing residual film recycling machine;

[0023] (7) After the pre-spring ploughing residual film recycling machine stops working, the unmanned aerial vehicle is started, the sampling task is sent to the flight controller through the unmanned aerial vehicle residual film pollution verification system, and the residual film verification image after recycling is photographed by the on-board camera;

[0024] (8) The residual film verification image after recycling stored in the unmanned aerial vehicle residual film pollution verification system is uploaded to the farmland residual film pollution supervision information platform, the deep learning technology is used to identify the residual film verification image after recycling, and the pixel point proportion of the residual film is calculated to obtain the average coverage rate of the residual film verified after recycling by the pre-spring ploughing residual film recycling machine.

[0025] The application method of the farmland residual film pollution supervision system, wherein the average coverage rate of the residual film is calculated by the following formula: Wherein T represents the average coverage rate of the residual film; q i represents the number of residual film pixel points in the ith picture, Q i represents the total number of pixel points in the ith picture; and n represents the total number of pictures obtained.

[0026] The application method of the farmland residual film pollution supervision system, wherein the average coverage rate calculation method mainly includes the following steps: (1) for the residual film image after recycling and the residual film verification image after recycling, a corresponding model trained in the farmland residual film pollution supervision information platform is selected for residual film identification and segmentation; (2) the segmented image is calculated by using the pixel point counting code written by python to calculate the segmented residual film pixel point number and the pixel point number of the whole picture; (3) the total proportion of the residual film pixel point number of each picture relative to the pixel point number of the whole picture is calculated, and the average proportion, i.e. the average coverage rate of the residual film, is obtained.

[0027] The application method of the farmland residual film pollution monitoring system, the residual film in the picture is recognized by using a deep learning technology, mainly includes the following steps: (1) using the post-harvest residual film recovery monitoring system, the unmanned aerial vehicle residual film pollution verification system and the pre-spring planting residual film recovery monitoring system, obtaining the data required for deep learning, the residual film image data set collected by the vehicle is transmitted through the data connection established by the 4GDTU module installed on the industrial computer through wireless transmission, and the residual film image data set collected by the unmanned aerial vehicle is downloaded through the unmanned aerial vehicle residual film pollution verification system; (2) the mobile user end uses the deep learning network to train the above data to obtain the best identification model, and the model is a vehicle-mounted residual film identification model and an unmanned aerial vehicle residual film identification model; (3) upload the trained model to the server for identification of the residual film recovery image and the verification image of the post-harvest residual film pollution monitoring information platform.

[0028] The farmland residual film pollution monitoring system in the application mainly includes a post-harvest residual film recovery monitoring system, a pre-spring planting residual film recovery monitoring system, an unmanned aerial vehicle residual film pollution verification system, a farmland residual film pollution monitoring information platform, a motion camera, a Beidou positioning device, an industrial computer, a post-harvest residual film recovery machine, a pre-spring planting residual film recovery machine, a 4GDTU module, a flight controller, an on-board camera, a GPS sensor, a data link, a mobile user end and a server.

[0029] The post-harvest residual film recovery monitoring system is used for monitoring the post-harvest residual film pollution and recording the images, position information, vehicle owner information of the post-harvest residual film recovery machine and farmland owner information before and after the recovery of residual film; the motion camera on the post-harvest residual film recovery machine is installed on the suspension in the cab, the height and angle of the camera are adjusted by adjusting the position change of the suspension, so that the camera maintains an appropriate shooting posture during image acquisition; the Beidou positioning device is installed above the cab through a strong magnet; the industrial computer is installed on the ground of the cab; the 4GDTU module is installed on the industrial computer to establish data connection with the camera through wireless transmission, and the image data is stored in the industrial computer and can be uploaded to the farmland residual film pollution monitoring information platform.

[0030] The spring pre-sowing residual film recovery supervision system is used for monitoring the spring pre-sowing residual film pollution and recording the images, position information, vehicle owner information of the spring pre-sowing residual film recovery machine and farmland owner information of the farmland before and after the recovery of the residual film; the motion camera on the spring pre-sowing residual film recovery machine is installed on a suspension in the cab, the height and angle of the camera are adjusted by adjusting the position change of the suspension, so that the camera maintains a suitable shooting posture during image acquisition; the Beidou positioning device is installed above the cab by a strong magnet; the industrial computer is installed on the ground of the cab; the 4G DTU module is installed on the industrial computer to establish a data connection with the camera by wireless transmission and store the image data in the industrial computer, which can be uploaded to the farmland residual film pollution supervision information platform.

[0031] The unmanned aerial vehicle residual film pollution verification system is used for verifying the residual film residue after the recovery of the autumn post-harvest residual film recovery machine and the residual film residue after the recovery of the spring pre-sowing residual film recovery machine, and recording the position information, farmer information, farmland area, mulching film use and recovery method of the recovery plot of the unmanned aerial vehicle flight; the flight controller connects each part of the unmanned aerial vehicle remote control system, power system and gimbal system, and is used for receiving and executing the flight task sent by the mobile user terminal and controlling the on-board camera to take pictures; the on-board camera is connected to the gimbal of the unmanned aerial vehicle, the gimbal can adjust the pitch angle, and the on-board camera can shoot the image with the lens facing the opposite side; the data link is used to establish a data connection with the mobile user terminal to perform wireless transmission of the image.

[0032] The mobile user terminal is used for installing the unmanned aerial vehicle residual film pollution verification system, and can also be used for deep learning training to obtain a model for identifying residual film images, and can log in to the farmland residual film pollution supervision information platform to check the information and data uploaded at any time.

[0033] The server is used for installing the farmland residual film pollution supervision information platform.

[0034] The farmland residual film pollution supervision information platform is used for identifying the images of the farmland before and after the recovery of the residual film by the autumn post-harvest residual film recovery machine, the images of the farmland before and after the recovery of the residual film by the spring pre-sowing residual film recovery machine, and the images of the farmland after the recovery of the residual film by the unmanned aerial vehicle, calculating the residual film proportion in the images, recording the position information and farmer information of the recovery plot, and checking at any time.

[0035] Compared with the prior art, the farmland residual film pollution supervision system provided by the application is efficient, convenient, economical and practical, can realize rapid evaluation of the quality of residual film recovery operation after autumn harvest, rapid evaluation of residual film pollution before spring sowing, and data collection for residual film recovery process by using unmanned aerial vehicle residual film pollution evaluation and verification, and builds a unified information platform; the application method of the farmland residual film pollution supervision system provided by the application is economical and practical, can combine deep learning, image processing and Internet of Things technology, use fast data collection, uploading and processing, and realize efficient and accurate monitoring and verification of farmland residual film pollution at different stages.

[0036] The farmland residual film pollution supervision system and the application method provided by the application realize rapid evaluation of the quality of residual film recovery operation after autumn harvest, unmanned aerial vehicle residual film pollution evaluation and verification, and rapid evaluation of residual film pollution before spring sowing by using vehicle-mounted near-ground imaging and unmanned aerial low-altitude imaging methods, combining deep learning, image processing, Internet of Things and other technologies, and achieving residual film recovery and residual film residue monitoring at each link. The application method can be used not only in various residual film recovery scenes after autumn harvest and before spring sowing, but also can efficiently and accurately identify and evaluate, ensuring the supervision of the residual film recovery process at each stage and providing reference data for whether the film-covered crop field is fallow. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 The structure schematic diagram of the farmland residual film pollution supervision system disclosed by the embodiments of the application is shown in the figure.

[0038] Figure 2 The residual film pollution monitoring process schematic diagram of the cotton field at the stage after autumn harvest disclosed by the embodiments of the application is shown in the figure.

[0039] Figure 3 The residual film pollution monitoring process schematic diagram of the cotton field at the stage before spring sowing disclosed by the embodiments of the application is shown in the figure.

[0040] Figure 4 The residual film identification model acquisition process schematic diagram disclosed by the embodiments of the application is shown in the figure.

[0041] Figure 5 The average coverage rate calculation process schematic diagram disclosed by the embodiments of the application is shown in the figure. DETAILED DESCRIPTION

[0042] A farmland residual film pollution supervision system mainly comprises a residual film recovery machine after autumn harvest, a residual film recovery machine before spring sowing, an unmanned aerial vehicle, a server and a mobile user terminal.

[0043] The post-harvest residual film recycling machine is provided with a motion camera, a Beidou positioning device, an industrial computer and a 4G DTU module; the motion camera is provided with two, which are installed on the left and right sides of the cab of the machine, and is used for collecting images of farmland before and after recycling residual films; the Beidou positioning device is installed above the cab of the machine, and is used for collecting position information of the land; the industrial computer is installed in the cab of the machine, and is used for installing a post-harvest residual film recycling supervision system; the 4G DTU module is installed on the industrial computer and transmits data in a wireless transmission mode.

[0044] The pre-spring planting residual film recycling machine is provided with a motion camera, a Beidou positioning device, an industrial computer and a 4G DTU module; the motion camera is provided with two, which are installed on the left and right sides of the cab of the machine, and is used for collecting images of farmland before and after recycling residual films; the Beidou positioning device is installed above the cab of the machine, and is used for collecting position information of the land; the industrial computer is installed in the cab of the machine, and is used for installing a pre-spring planting residual film recycling supervision system; the 4G DTU module is installed on the industrial computer and transmits data in a wireless transmission mode.

[0045] The unmanned aerial vehicle is provided with a flight controller, an on-board camera, a GPS sensor and a data transmission link; the flight controller is used for stabilizing the flight state and executing flight instructions; the on-board camera is installed on a gimbal and is used for shooting images of farmland after the residual film recycling machine recycles; the GPS sensor is built into the unmanned aerial vehicle and is used for collecting position information of the land; the data transmission link is used for transmitting photos shot by the on-board camera and position information of the land.

[0046] The server is installed with a farmland residual film pollution supervision information platform, which is used for identifying images of farmland before and after the post-harvest residual film recycling machine recycles residual films, images of farmland before and after the pre-spring planting residual film recycling machine recycles residual films, and images of farmland after the unmanned aerial vehicle shoots, and calculating residual film proportion in the images, recording position information of the recycling land and farmer information, and being available for checking at any time.

[0047] The mobile user terminal is installed with an unmanned aerial vehicle residual film pollution checking system, which is used for checking residual film residues after the post-harvest residual film recycling machine recycles and after the pre-spring planting residual film recycling machine recycles, and recording position information of the recycling land, farmer information, farmland area, basic information of land film use and recycling mode of the unmanned aerial vehicle; the mobile user terminal is also used for deep learning training to obtain a model for identifying residual film images, logging in the cloud to check information and data at any time.

[0048] The post-harvest residual film recycling supervision system is used for recording images of farmland before and after recycling residual films, and position information of the farmland, owner information of the post-harvest residual film recycling machine and farmer information of the farmland.

[0049] The spring pre-sowing residual film recovery supervision system is used for recording the images of the farmland before and after the recovery of the residual film, and the position information of the farmland, the information of the owner of the spring pre-sowing residual film recovery machine and the information of the owner of the farmland.

[0050] The mobile user terminal carries out deep learning training to obtain a model for identifying the residual film image, that is, the mobile user terminal divides the data into a training set, a test set and a verification set by using the residual film image information data on the farmland residual film pollution supervision information platform, carries out residual film labeling, carries out model training by using a deep learning network, and obtains the best identification model, wherein the model is a vehicle-mounted residual film identification model and a UAV residual film identification model.

[0051] An application method of the farmland residual film pollution supervision system is provided, which utilizes the post-harvest residual film recovery machine and the installed post-harvest residual film recovery supervision system, the UAV, the UAV residual film pollution verification system, the spring pre-sowing residual film recovery machine and the installed spring pre-sowing residual film recovery supervision system, obtains the image after the recovery of the residual film by using the motion camera and the on-board camera, and obtains the residual film coverage rate of the farmland after the recovery on the farmland residual film pollution supervision information platform; the method mainly includes the following steps:

[0052] (1) when the residual film recovery is carried out after the harvest, the post-harvest residual film recovery machine is started, the image of the residual film in the field after the recovery is shot by using the motion camera, the residual film recovery machine is required to maintain a stable speed during the driving, and the image after the recovery of the residual film is collected in sequence according to the planned route;

[0053] (2) the image of the residual film after the recovery stored in the post-harvest residual film recovery supervision system is uploaded to the farmland residual film pollution supervision information platform, the deep learning technology is used to identify the image of the residual film after the recovery, the pixel point proportion of the residual film is calculated, and the average coverage rate of the residual film after the recovery of the post-harvest residual film recovery machine is obtained;

[0054] (3) after the post-harvest residual film recovery machine stops the operation after the recovery, the UAV is started, the sampling task is sent to the flight controller by using the UAV residual film pollution verification system, and the verification image of the residual film in the field after the recovery is shot by using the on-board camera;

[0055] (4) the verification image of the residual film in the field after the recovery stored in the UAV residual film pollution verification system is uploaded to the farmland residual film pollution supervision information platform, the deep learning technology is used to identify the verification image of the residual film in the field after the recovery, the pixel point proportion of the residual film is calculated, and the average coverage rate of the residual film after the verification of the post-harvest residual film recovery machine is obtained;

[0056] (5) When the pre-spring ploughing residual film recycling machine is started, the residual film image after recycling is photographed by a motion camera, and the residual film recycling machine is required to maintain a stable speed during driving and collect the image after recycling residual film according to the planned route;

[0057] (6) The residual film image after recycling stored in the pre-spring ploughing residual film recycling supervision system is uploaded to the farmland residual film pollution supervision information platform, the deep learning technology is used to identify the residual film image after recycling, and the pixel point proportion of residual film is calculated to obtain the average coverage rate of residual film after recycling by the pre-spring ploughing residual film recycling machine;

[0058] (7) After the pre-spring ploughing residual film recycling machine stops working, the unmanned aerial vehicle is started, the sampling task is sent to the flight controller through the unmanned aerial vehicle residual film pollution verification system, and the residual film verification image after recycling is photographed by the on-board camera;

[0059] (8) The residual film verification image after recycling stored in the unmanned aerial vehicle residual film pollution verification system is uploaded to the farmland residual film pollution supervision information platform, the deep learning technology is used to identify the residual film verification image after recycling, and the pixel point proportion of residual film is calculated to obtain the average coverage rate of residual film after verification by the pre-spring ploughing residual film recycling machine.

[0060] The application method of the farmland residual film pollution supervision system, wherein the average coverage rate of residual film is calculated by the following formula: Wherein T represents the average coverage rate of residual film; q i represents the number of residual film pixel points in the i-th picture, Q i represents the total number of pixel points in the i-th picture; and n represents the total number of pictures obtained.

[0061] The application method of the farmland residual film pollution supervision system, wherein the average coverage rate calculation method mainly includes the following steps: (1) for the residual film image after recycling and the residual film verification image after recycling, a corresponding model trained in the farmland residual film pollution supervision information platform is selected for residual film identification and segmentation; (2) the segmented image is calculated by using the pixel point counting code written by python to calculate the segmented residual film pixel point number and the pixel point number of the whole picture; (3) the total proportion of the residual film pixel point number of each picture relative to the pixel point number of the whole picture is calculated, and the average proportion, i.e. the average coverage rate of residual film, is obtained.

[0062] The application method of the farmland residual film pollution supervision system, wherein the residual film in the picture is recognized by using a deep learning technology, mainly includes the following steps: (1) using the post-harvest residual film recovery supervision system, the unmanned aerial vehicle residual film pollution verification system and the pre-spring planting residual film recovery supervision system to obtain the data required for deep learning, transmitting the vehicle-mounted residual film image dataset to the mobile user terminal through the data connection established by the 4G DTU module installed on the industrial computer through wireless transmission, and downloading the residual film image dataset collected by the unmanned aerial vehicle through the unmanned aerial vehicle residual film pollution verification system; (2) the mobile user terminal trains the above data by using a deep learning network to obtain the best recognition model, and the model is a vehicle-mounted residual film recognition model and an unmanned aerial vehicle residual film recognition model; and (3) uploading the trained model to a server for identification of residual film recovery images and verification images by the farmland residual film pollution supervision information platform in the later period.

[0063] The application provides a farmland residual film pollution supervision system and an application method thereof, and the residual film recovery condition is effectively monitored from the following parts.

[0064] The method for recognizing the residual film in the picture by using the deep learning technology includes the following steps: (1) using the post-harvest residual film recovery supervision system, the unmanned aerial vehicle residual film pollution verification system and the pre-spring planting residual film recovery supervision system to obtain the data required for deep learning, transmitting the vehicle-mounted residual film image dataset to the mobile user terminal through the data connection established by the 4G DTU module installed on the industrial computer through wireless transmission, and downloading the residual film image dataset collected by the unmanned aerial vehicle through the unmanned aerial vehicle residual film pollution verification system; (2) the mobile user terminal divides the data into a training set, a test set and a verification set, and performs residual film labeling, trains the model by using a deep learning network to obtain the best recognition model, and the model is a vehicle-mounted residual film recognition model and an unmanned aerial vehicle residual film recognition model; and (3) uploading the trained model to a server for identification of residual film recovery images and verification images by the farmland residual film pollution supervision information platform in the later period.

[0065] The method for calculating the average coverage rate includes the following steps: (1) uploading the post-recovery residual film image and the post-recovery field residual film verification image to the farmland residual film pollution supervision information platform; (2) selecting the picture from the farmland residual film pollution supervision information platform to recognize and segment the residual film by using the corresponding trained model; (3) calculating the residual film pixel point number after segmentation and the pixel point number of the whole picture by using the pixel point counting code written by python; and (4) calculating the residual film pixel point number of each picture accounts for the pixel point number of the whole picture and summing up to calculate the average residual film pixel point ratio of each picture, that is, the average coverage rate of the residual film.

[0066] The application method of the farmland residual membrane pollution supervision system is as follows: (1) after the autumn harvest, the user records the information of the owner of the post-autumn residual membrane recycling machine, the information of the farmland owner and the longitude and latitude obtained by the Beidou positioning device as the position information in the post-autumn residual membrane recycling supervision system, and after the post-autumn residual membrane recycling machine starts, the residual membrane recycling machine keeps a stable speed during driving and collects the images of the residual membrane after recycling in turn according to the planned route; (2) after the residual membrane recycling machine stops working, the images of the residual membrane after recycling stored in the post-autumn residual membrane recycling supervision system are uploaded to the farmland residual membrane pollution supervision information platform, the residual membrane recognition model trained by the deep learning technology is used to identify the images of the residual membrane after recycling, the pixel point proportion of the residual membrane is calculated, and the average coverage rate of the residual membrane after recycling by the post-autumn residual membrane recycling machine is obtained; (3) after the post-autumn residual membrane recycling machine leaves the farmland, the user records the position information of the recycling plot, the information of the farmer, the farmland area, the mulching film use and the recycling method and other basic information in the unmanned aerial vehicle residual membrane pollution verification system, the unmanned aerial vehicle starts, sends a sampling task to the flight controller through the unmanned aerial vehicle residual membrane pollution verification system, and the on-board camera shoots the residual membrane verification images after recycling; (4) after the unmanned aerial vehicle returns, the residual membrane verification images after recycling stored in the unmanned aerial vehicle residual membrane pollution verification system are uploaded to the farmland residual membrane pollution supervision information platform, the residual membrane verification images after recycling are identified by the deep learning technology, the pixel point proportion of the residual membrane is calculated, and the average coverage rate of the residual membrane verified after recycling by the post-autumn residual membrane recycling machine is obtained; (5) when the residual membrane is recycled again before spring sowing, the user records the information of the owner of the pre-spring residual membrane recycling machine, the information of the farmland owner and the longitude and latitude obtained by the Beidou positioning device as the position information in the pre-spring residual membrane recycling supervision system, and after the pre-spring residual membrane recycling machine starts, the residual membrane recycling machine keeps a stable speed during driving and collects the images of the residual membrane after recycling in turn according to the planned route; (6) after the residual membrane recycling machine stops working, the images of the residual membrane after recycling stored in the pre-spring residual membrane recycling supervision system are uploaded to the farmland residual membrane pollution supervision information platform, the residual membrane recognition model trained by the deep learning technology is used to identify the images of the residual membrane after recycling, the pixel point proportion of the residual membrane is calculated, and the average coverage rate of the residual membrane after recycling by the post-autumn residual membrane recycling machine is obtained; (7) after the pre-spring residual membrane recycling machine leaves the farmland, the user records the position information of the recycling plot, the information of the farmer, the farmland area, the mulching film use and the recycling method and other basic information in the unmanned aerial vehicle residual membrane pollution verification system, the unmanned aerial vehicle starts, sends a sampling task to the flight controller through the unmanned aerial vehicle residual membrane pollution verification system, and the on-board camera shoots the residual membrane verification images after recycling.(8) After the UAV returns, the field residue film verification image of the residue film after recovery stored in the UAV residue film pollution verification system is uploaded to the farmland residue film pollution supervision information platform, the deep learning technology is used to identify the field residue film verification image after recovery, the pixel point proportion of the residue film is calculated, and the average coverage rate of the residue film verified after recovery by the spring pre-sowing residue film recovery machine is obtained;(9) After the end, the monitoring data and corresponding farmland related information of the land can be inquired at any time by logging in the farmland residue film pollution supervision information platform.

[0067] The system adopts a 4GDTU module as a data transmission module. The 4GDTU module is a wireless transmission module based on the Internet of Things, provides the functions of 4G wireless network and TCP / IP data communication, has the functions of serial data bidirectional conversion, supports automatic heartbeat and parameter setting, and has more advantages than traditional wired transmission and GPRS transmission.

[0068] The present application uses deep learning and image processing methods to realize the monitoring of the farmland residue film pollution in different stages. Compared with the prior art, the method has the characteristics of high image spatial resolution, low cost, wide monitoring range and high data accuracy, and is very suitable for rapid and non-contact detection on site. Since the image acquisition is completed in an instant, the deep learning method is a more effective measurement method, which can complete the accurate evaluation of the surface residue film of the farmland.

[0069] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. It should be pointed out that any modification, equivalent replacement and improvement made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A farmland residual film pollution monitoring system, characterized in that Mainly including post-harvest residual film recycling machine, pre-spring planting residual film recycling machine, unmanned aerial vehicle, server, mobile user terminal; The post-harvest residual film recycling machine is provided with a motion camera, a Beidou positioning device, an industrial computer and a 4G DTU module; the motion camera is provided with two, which are installed on the left and right sides of the cab of the machine, and is used for collecting images of farmland before and after recycling residual film; the Beidou positioning device is installed above the cab of the machine, and is used for collecting position information of the land; the industrial computer is installed in the cab of the machine, and is used for installing a post-harvest residual film recycling supervision system; the 4G DTU module is installed on the industrial computer and transmits data in a wireless transmission mode; The pre-spring planting residual film recycling machine is provided with a motion camera, a Beidou positioning device, an industrial computer and a 4G DTU module; the motion camera is provided with two, which are installed on the left and right sides of the cab of the machine, and is used for collecting images of farmland before and after recycling residual film; the Beidou positioning device is installed above the cab of the machine, and is used for collecting position information of the land; the industrial computer is installed in the cab of the machine, and is used for installing a pre-spring planting residual film recycling supervision system; the 4G DTU module is installed on the industrial computer and transmits data in a wireless transmission mode; The unmanned aerial vehicle is provided with a flight controller, an on-board camera, a GPS sensor and a data transmission link; the flight controller is used for stabilizing the flight state and executing flight instructions; the on-board camera is installed on a gimbal, and is used for shooting images of farmland after recycling by the residual film recycling machine; the GPS sensor is built into the unmanned aerial vehicle, and is used for collecting position information of the land; the data transmission link is used for transmitting photos shot by the on-board camera and position information of the land; The server is installed with a farmland residual film pollution supervision information platform, which is used for identifying images of farmland before and after recycling residual film collected by the post-harvest residual film recycling machine, images of farmland before and after recycling residual film collected by the pre-spring planting residual film recycling machine, and images of farmland after recycling by the residual film recycling machine shot by the unmanned aerial vehicle, and calculating residual film proportion in the images, recording position information of the recycling land and farmer information, and being available for checking at any time; The mobile user terminal is installed with an unmanned aerial vehicle residual film pollution checking system, which is used for checking residual film remaining conditions after recycling by the post-harvest residual film recycling machine and the pre-spring planting residual film recycling machine, and recording position information of the recycling land, farmer information, farmland area, mulching film use and basic information of recycling mode of the unmanned aerial vehicle; the mobile user terminal is also used for deep learning training to obtain a model for identifying residual film images, logging in the cloud to check information and data at any time, and uploading data. 2.The agricultural field residue film pollution monitoring system of claim 1, wherein The post-harvest residual film recycling supervision system is used for recording images of farmland before and after recycling residual film, and position information of the farmland, owner information of the post-harvest residual film recycling machine and farmer information of the farmland. 3.The agricultural field residue film pollution monitoring system of claim 1 or 2, wherein The pre-spring planting residual film recycling supervision system is used for recording images of farmland before and after recycling residual film, and position information of the farmland, owner information of the pre-spring planting residual film recycling machine and farmer information of the farmland. 4.The farmland residual film pollution monitoring system of claim 3, wherein, The mobile user terminal carries out deep learning training to obtain a model for identifying residual film images, that is, the mobile user terminal uses residual film image information data on a farmland residual film pollution supervision information platform, divides the data into a training set, a test set and a verification set, carries out residual film labeling, uses a deep learning network to carry out model training, and obtains an optimal identification model, which is a vehicle-mounted residual film identification model and an unmanned aerial vehicle-mounted residual film identification model.

5. The application method of the agricultural field residue film pollution monitoring system according to claim 4, characterized in that, The post-harvest residual film recovery machine, the installed post-harvest residual film recovery supervision system, the unmanned aerial vehicle, the unmanned aerial vehicle residual film pollution verification system, the pre-spring planting residual film recovery machine and the installed pre-spring planting residual film recovery supervision system are used to obtain images after residual film recovery through a motion camera and an on-board camera, and obtain the residual film coverage rate of farmland after recovery on a farmland residual film pollution supervision information platform; the method mainly comprises the following steps: (1) when residual film recovery is carried out after harvest, the post-harvest residual film recovery machine is started, a motion camera is used to shoot images of residual film in the field after recovery, and the residual film recovery machine is required to maintain a stable speed during driving and collect images after residual film recovery in sequence according to a planned route; (2) the images of residual film after recovery stored in the post-harvest residual film recovery supervision system are uploaded to the farmland residual film pollution supervision information platform, deep learning technology is used to identify the images of residual film after recovery, the pixel point proportion of residual film is calculated, and the average coverage rate of residual film after recovery by the post-harvest residual film recovery machine is obtained; (3) after the post-harvest residual film recovery machine stops working after recovery, the unmanned aerial vehicle is started, a sampling task is sent to a flight controller through the unmanned aerial vehicle residual film pollution verification system, and an on-board camera is used to shoot residual film verification images in the field after recovery; (4) the residual film verification images in the field after recovery stored in the unmanned aerial vehicle residual film pollution verification system are uploaded to the farmland residual film pollution supervision information platform, deep learning technology is used to identify the residual film verification images in the field after recovery, the pixel point proportion of residual film is calculated, and the average coverage rate of residual film after verification by the post-harvest residual film recovery machine is obtained; (5) when residual film is recovered again before spring planting, the pre-spring planting residual film recovery machine is started, a motion camera is used to shoot images of residual film in the field after recovery, and the residual film recovery machine is required to maintain a stable speed during driving and collect images after residual film recovery in sequence according to a planned route; (6) the images of residual film after recovery stored in the pre-spring planting residual film recovery supervision system are uploaded to the farmland residual film pollution supervision information platform, deep learning technology is used to identify the images of residual film after recovery, the pixel point proportion of residual film is calculated, and the average coverage rate of residual film after recovery by the post-harvest residual film recovery machine is obtained; (7) after the pre-spring planting residual film recovery machine stops working, the unmanned aerial vehicle is started, a sampling task is sent to a flight controller through the unmanned aerial vehicle residual film pollution verification system, and an on-board camera is used to shoot residual film verification images in the field after recovery; (8) The field residue film verification image after residue film recovery stored in the unmanned aerial vehicle residue film pollution verification system is uploaded to the farmland residue film pollution supervision information platform, the deep learning technology is used for identifying the field residue film verification image after recovery, the pixel point proportion of the residue film is calculated, and the average coverage rate of the residue film verified by the residue film recovery machine before spring planting is obtained.

6. The application method of the agricultural field residue film pollution monitoring system according to claim 5, wherein, The calculation formula of the average coverage of the residual film is: Wherein T represents the average coverage of the residual film; q i represents the number of residual film pixel points in the i th picture, Q i represents the total number of pixel points in the i th picture; n represents the total number of pictures acquired.

7. The application method of the agricultural field residue film pollution monitoring system according to claim 6, characterized in that, The method for calculating the average coverage rate mainly includes the following steps: (1) For the obtained residue film recovery image and the field residue film verification picture after recovery, a corresponding trained model in the farmland residue film pollution supervision information platform is selected for residue film identification and segmentation; (2) the segmented image is calculated by using the pixel point counting code written by python, and the pixel point number of the segmented residue film and the pixel point number of the whole picture are calculated; (3) the total proportion of the residue film pixel point number of each picture relative to the pixel point number of the whole picture is calculated, and the average proportion, i.e. the average coverage rate of the residue film, is obtained. 8.The method of claim 7, wherein the system is applied to a field in which a residue of a film is present. The deep learning technology is used for identifying the residue film in the picture, mainly including the following steps: (1) the data required for deep learning is obtained by using the residue film recovery supervision system after autumn harvest, the unmanned aerial vehicle residue film pollution verification system and the residue film recovery supervision system before spring planting, the residue film image data set collected by the vehicle is transmitted through the data connection established by the 4GDTU module installed on the industrial computer and the mobile user terminal by wireless transmission, and the residue film image data set collected by the unmanned aerial vehicle is downloaded through the unmanned aerial vehicle residue film pollution verification system; (2) the mobile user terminal trains the model by using the deep learning network, obtains the best identification model, and the model is the vehicle-mounted residue film identification model and the unmanned aerial vehicle residue film identification model; (3) the trained model is uploaded to the server, which is used for identifying the residue film recovery image and the verification image by the farmland residue film pollution supervision information platform.

Images