High-standard farmland management system and insect pest situation monitoring method

Abstract

The invention belongs to the technical field of intelligent control, and particularly relates to a high-standard farmland management system and an insect situation monitoring method. Dividing the farmland to be detected into a plurality of detection areas; acquiring a farmland image acquired by the unmanned aerial vehicle by adopting a first acquisition frequency, marking a blade contour in the farmland image, identifying a non-blade area in a blade contour area, and calculating a ratio of the total non-blade area to the total blade contour area in each detection area; and comparing the ratio in each detection area with the first threshold value and the second threshold value, and performing different control modes. The insect pest situation is measured by calculating the area ratio of the total non-leaf area to the total leaf contour area in each detection area through image recognition, so that a more accurate judgment standard is provided, and the insect pest situation can be monitored more strictly. By adjusting the acquisition frequency, differential acquisition is implemented, so that the cost can be reduced, and the computing resource consumption can be reduced.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent control technology, and more specifically, relates to a high-standard farmland management system and an insect monitoring method. Background Art

[0002] Pest management in farmland is a key technology that determines yield and income. Currently, it is mostly based on manual management and lacks flexible, effective and low-cost automated control management. Summary of the invention

[0003] The invention provides a high-standard farmland management system and an insect monitoring method.

[0004] A method for monitoring insect pests, comprising:

[0005] S1: Divide the farmland to be tested into several testing areas;

[0006] S2: acquiring a farmland image acquired by the UAV using a first acquisition frequency, marking the leaf contours in the farmland image, identifying the non-leaf area in the leaf contour area, and calculating the ratio of the total non-leaf area to the total leaf contour area in each detection area;

[0007] S3: Compare the ratio in each detection area with the first threshold and the second threshold,

[0008] If the ratio of a certain detection area is greater than the first threshold, the pesticide spraying device is controlled to spray pesticides on this detection area, and the image acquisition frequency of this detection area is adjusted to the second acquisition frequency; if the ratio of a certain detection area is not greater than the first threshold and not less than the second threshold, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency; if the ratio of a certain detection area is less than the second threshold, the image acquisition frequency of this detection area remains at the first acquisition frequency.

[0009] Preferably, the operation of marking the leaf contours in the farmland image and identifying the non-leaf areas in the leaf contour area is implemented by a neural network.

[0010] Preferably, in S3, if the ratio of a certain detection area is not greater than the first threshold and not less than the second threshold, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency, and then it also includes: if the growth rate of the ratio of a certain detection area exceeds the first growth rate threshold, the pesticide spraying device is controlled to spray pesticides on this detection area, and the image acquisition frequency of this detection area is adjusted to the second acquisition frequency.

[0011] Preferably, the second acquisition frequency>the third acquisition frequency>the first acquisition frequency.

[0012] Preferably, in S3, if the ratio of a certain detection area is greater than the first threshold, the drone is controlled to spray pesticides on the detection area, and the image acquisition frequency of this detection area is adjusted to the second acquisition frequency. It also includes that if after multiple rounds of detection, the growth rate of the ratio of the detection area is less than the second growth rate threshold, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency.

[0013] It is preferably suitable for the green growth stage, green rice stage or leaf expansion stage of crop growth.

[0014] Preferably, the selection of pesticides in S3 is based on the types of crops in the farmland and their growth stages.

[0015] Farmland management system, including:

[0016] Drone: Equipped with a camera device, it is used to receive instructions from the processor and collect images of the farmland to be inspected;

[0017] Pesticide spraying device: used to receive instructions from the processor and spray pesticides on the detection area to be sprayed with pesticides;

[0018] Processor: used to obtain the farmland image collected by the UAV using the first collection frequency, mark the leaf contours in the farmland image, identify the non-leaf area in the leaf contour area, calculate the ratio of the total non-leaf area to the total leaf contour area in each detection area; compare the ratio in each detection area with the first threshold and the second threshold,

[0019] If the ratio of a certain detection area is greater than the first threshold, the pesticide spraying device is controlled to spray pesticides on this detection area, and the image acquisition frequency of this detection area is adjusted to the second acquisition frequency; if the ratio of a certain detection area is not greater than the first threshold and not less than the second threshold, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency; if the ratio of a certain detection area is less than the second threshold, the image acquisition frequency of this detection area remains at the first acquisition frequency.

[0020] Beneficial effects of the present invention:

[0021] Through image recognition, the ratio of the total non-leaf area to the total leaf contour area in each detection area is calculated, and the insect situation is measured by checking whether the leaves in the detection area are intact, covered with pests and excrement, etc. This has a more accurate judgment standard, can monitor the insect situation more strictly, and spray the corresponding pesticides in time.

[0022] By adjusting the collection frequency, we can conduct intensive monitoring on areas with serious insect infestation and conduct loose management on areas with less serious infestation, which can reduce costs and reduce the consumption of computing resources. DETAILED DESCRIPTION

[0023] The technical solution of the present application is described in detail below in conjunction with specific embodiments.

[0024] A method for monitoring insect pests is more suitable for the green growth period, green rice period or leaf expansion period in the growth cycle of crops, that is, the period of rapid leaf growth, especially before heading, and includes the following steps:

[0025] S1: Divide the farmland to be tested into several testing areas.

[0026] By dividing the farmland to be inspected into different zones, more accurate monitoring can be achieved. This zone can be an actual zone in the farmland or a virtual zone based on the collection route and speed of the drone.

[0027] S2: Acquire the farmland image collected by the UAV using the first collection frequency, mark the leaf contours in the farmland image, identify the non-leaf area in the leaf contour area, and calculate the ratio of the total non-leaf area to the total leaf contour area in each detection area.

[0028] The leaf contours in the farmland image can be marked by a neural network and the non-leaf area in the leaf contour area can be identified. Manual identification is also possible.

[0029] If a neural network is used, a dual segmentation model of DeepLabV3+ and U-Net can be specifically adopted to distinguish the leaf contours and non-leaf areas in the farmland image to obtain the leaf contour area.

[0030] If an insect infestation occurs, the leaves will either be eaten up and have holes, or the leaves will be covered with insects, eggs, and excrement, and the color will be different from the leaves.

[0031] The next step is to extract pixel features from the leaf contour area, extract the green channel value, and mark the pixels in the leaf contour whose green channel value is less than the green threshold as non-leaf pixels. All non-leaf pixels constitute the non-leaf area. The non-leaf areas in the area to be detected are summed to obtain the total non-leaf area, and the leaf contour areas are summed to obtain the total leaf contour area. Then, the ratio of the total non-leaf area to the total leaf contour area in each detection area is calculated. If the ratio is large, it means that the insect situation is serious, and the insect situation of the leaves in the area to be detected can be better obtained.

[0032] S3: Compare the ratio in each detection area with the first threshold and the second threshold,

[0033] If the ratio of a certain detection area is greater than the first threshold, the pesticide spraying device is controlled to spray pesticides on this detection area, and the image acquisition frequency of this detection area is adjusted to the second acquisition frequency; if the ratio of a certain detection area is not greater than the first threshold and not less than the second threshold, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency; if the ratio of a certain detection area is less than the second threshold, the image acquisition frequency of this detection area remains at the first acquisition frequency.

[0034] The second acquisition frequency>the third acquisition frequency>the first acquisition frequency. The first threshold value>the second threshold value.

[0035] A first threshold value and a second threshold value are set, and different detection control methods are adopted by comparing the relationship between the ratio and the first threshold value and the second threshold value.

[0036] For example, if the ratio of a certain detection area is greater than the first threshold, it means that the insect situation is serious and needs to be controlled by spraying pesticides. Then the pesticide spraying device is controlled to spray pesticides on this detection area. The selection of pesticide types is based on the types of crops in the farmland and the growth stage they are in. For example, which type of insect situation may be more likely to occur in corn during this growth period, so the corresponding pesticide is sprayed. Since there are relatively few types of insect situations in farmland, this method does not involve the identification of insect species. After spraying pesticides, it is necessary to monitor the killing effect, so at this time, the image acquisition frequency of this detection area should also be adjusted at the same time, and a second acquisition frequency that is higher than the first acquisition frequency should be used for more intensive monitoring. After each image is collected, the above ratio still needs to be calculated to control the trend of insect situation changes so that the monitoring strategy can be adjusted at any time.

[0037] If the ratio of a certain detection area is less than the second threshold, it means that there is no insect infestation, and the image acquisition frequency of this detection area is maintained at the first acquisition frequency.

[0038] For example, if the ratio of a certain detection area is not greater than the first threshold and not less than the second threshold, it means that there is insect infestation in this detection area, but it is not serious. It is only necessary to adjust the image acquisition frequency of this detection area to a third acquisition frequency slightly higher than the first acquisition frequency for monitoring.

[0039] In addition to monitoring the ratio itself, it is also possible to monitor the rate at which the ratio increases or decreases.

[0040] For example, if the ratio of a certain detection area is not greater than the first threshold and not less than the second threshold, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency. If the ratio growth rate exceeds the first growth rate threshold during monitoring, it means that the insect situation is developing obviously and rapidly. At this time, it is necessary to control the pesticide spraying device to spray pesticides in this detection area, and adjust the image acquisition frequency of this detection area to the second acquisition frequency.

[0041] For another example, if the ratio of a certain detection area is greater than the first threshold, the drone is controlled to spray pesticides on the detection area, and the image acquisition frequency of this detection area is adjusted to the second acquisition frequency. After multiple rounds of detection, the growth rate of the ratio of the detection area is less than the second growth rate threshold, indicating that the insect situation is under control. At this time, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency.

[0042] If the ratio of a certain detection area is less than the second threshold, it means that there is no insect infestation, and the image acquisition frequency of this detection area is maintained at the first acquisition frequency. If during the monitoring process, the ratio growth rate exceeds the first growth rate threshold, it means that the insect infestation is developing obviously and rapidly. At this time, it is necessary to control the pesticide spraying device to spray pesticides in this detection area, and adjust the image acquisition frequency of this detection area to the second acquisition frequency.

[0043] The above monitoring strategy adjustments are just a list and not exhaustive.

[0044] Farmland management system, including:

[0045] Drone: Equipped with a camera device, it is used to receive instructions from the processor and collect images of the farmland to be inspected;

[0046] Pesticide spraying device: used to receive instructions from the processor and spray pesticides on the detection area to be sprayed with pesticides;

[0047] Processor: used to obtain the farmland image collected by the UAV using the first collection frequency, mark the leaf contours in the farmland image, identify the non-leaf area in the leaf contour area, calculate the ratio of the total non-leaf area to the total leaf contour area in each detection area; compare the ratio in each detection area with the first threshold and the second threshold,

[0048] If the ratio of a certain detection area is greater than the first threshold, the pesticide spraying device is controlled to spray pesticides on this detection area, and the image acquisition frequency of this detection area is adjusted to the second acquisition frequency; if the ratio of a certain detection area is not greater than the first threshold and not less than the second threshold, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency; if the ratio of a certain detection area is less than the second threshold, the image acquisition frequency of this detection area remains at the first acquisition frequency.

[0049] The processor is also used to adjust the image acquisition frequency of a certain detection area to a third acquisition frequency if the ratio of this detection area is not greater than the first threshold and not less than the second threshold. It also includes: if the growth rate of the ratio of a certain detection area exceeds the first growth rate threshold, control the drone to spray pesticides on this detection area, and adjust the image acquisition frequency of this detection area to the second acquisition frequency.

Claims

1. A method for monitoring insect pests, characterized in that: include: S1: Divide the farmland to be tested into several testing areas; S2: acquiring a farmland image acquired by the UAV using a first acquisition frequency, marking the leaf contours in the farmland image, identifying the non-leaf area in the leaf contour area, and calculating the ratio of the total non-leaf area to the total leaf contour area in each detection area; S3: Compare the ratio in each detection area with the first threshold and the second threshold, If the ratio of a certain detection area is greater than the first threshold, the pesticide spraying device is controlled to spray pesticides on this detection area, and the image acquisition frequency of this detection area is adjusted to the second acquisition frequency; if the ratio of a certain detection area is not greater than the first threshold and not less than the second threshold, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency; if the ratio of a certain detection area is less than the second threshold, the image acquisition frequency of this detection area remains at the first acquisition frequency.

2. The insect monitoring method according to claim 1, characterized in that: Also includes: The neural network is used to mark the leaf contours in the farmland image and identify the non-leaf areas in the leaf contour area.

3. The insect monitoring method according to claim 1, characterized in that: In S3, if the ratio of a certain detection area is not greater than the first threshold and not less than the second threshold, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency. It also includes: if the growth rate of the ratio of a certain detection area exceeds the first growth rate threshold, the drone is controlled to spray pesticides on this detection area, and the image acquisition frequency of this detection area is adjusted to the second acquisition frequency.

4. The insect monitoring method according to claim 1, characterized in that: The second acquisition frequency>the third acquisition frequency>the first acquisition frequency.

5. The insect monitoring method according to claim 1, characterized in that: In S3, if the ratio of a certain detection area is greater than the first threshold, the pesticide spraying device is controlled to spray pesticides on the detection area, and the image acquisition frequency of this detection area is adjusted to the second acquisition frequency. It also includes that if after multiple rounds of detection, the growth rate of the ratio of the detection area is less than the second growth rate threshold, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency.

6. The insect monitoring method according to claim 1, characterized in that: It is suitable for the green growth stage, green rice stage or leaf expansion stage of crops.

7. The insect monitoring method according to claim 1, characterized in that: The choice of pesticides in S3 is based on the type of crops in the field and their growth stage.

8. A farmland management system, characterized in that: include: Drone: Equipped with a camera device, it is used to receive instructions from the processor and collect images of the farmland to be inspected; Pesticide spraying device: used to receive instructions from the processor and spray pesticides on the detection area to be sprayed with pesticides; Processor: used to obtain the farmland image collected by the UAV using the first collection frequency, mark the leaf contours in the farmland image, identify the non-leaf area in the leaf contour area, calculate the ratio of the total non-leaf area to the total leaf contour area in each detection area; compare the ratio in each detection area with the first threshold and the second threshold, If the ratio of a certain detection area is greater than the first threshold, the pesticide spraying device is controlled to spray pesticides on this detection area, and the image acquisition frequency of this detection area is adjusted to the second acquisition frequency; if the ratio of a certain detection area is not greater than the first threshold and not less than the second threshold, the image acquisition frequency of this detection area is adjusted to the third acquisition frequency; if the ratio of a certain detection area is less than the second threshold, the image acquisition frequency of this detection area remains at the first acquisition frequency.

9. The farmland management system according to claim 8, characterized in that: The processor is also used to adjust the image acquisition frequency of a certain detection area to a third acquisition frequency if the ratio of this detection area is not greater than the first threshold and not less than the second threshold. It also includes: if the growth rate of the ratio of a certain detection area exceeds the first growth rate threshold, control the drone to spray pesticides on this detection area, and adjust the image acquisition frequency of this detection area to the second acquisition frequency.