A leek maggot density estimation method and estimation system thereof

Abstract

The present application relates to a kind of leek maggot density estimation method and estimation system thereof.The leek maggot density estimation method is trained by constructing leek maggot identification model based on neural network and sample set composed of multiple appearance images in leek field, to realize the training of leek maggot identification model.The trained leek maggot identification model is used for the region identification of the image to be estimated, and the leek maggot region in the image to be estimated is identified, and then the target pixel total number corresponding to the leek maggot region in the whole image to be estimated is obtained.The number of leek maggots in the image to be estimated is calculated according to the average pixel number of a single leek maggot, and the number of leek maggots per unit area is calculated according to the actual area of the image to be estimated, that is, the leek maggot density of the image to be estimated.The method can effectively estimate the leek maggot density of each vegetable plot in leek field, and provide sufficient theoretical basis for later pest control.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of digital agriculture planting, in particular to a leek maggot density estimation method and system. BACKGROUND

[0002] Leek maggot (leek root maggot) is the natural enemy of leek, and leek root maggot can occur in leek greenhouse and open field. Leek maggot larvae feed on the underground part of leek, and the leek that is eaten will show symptoms such as weak, yellow, wilted and broken leaves. The larvae of leek maggot often gather in the root tuber or drill into the false stem to cause rot, and in severe cases, it can cause the whole row to be destroyed and the seeds to be lost, resulting in serious yield reduction and causing great losses to the growers.

[0003] Currently, agricultural experts have researched "high-temperature covering method", which uses high temperature to eliminate leek maggot. This method has good effect. Since leek has strong regenerative ability, it will grow again after harvesting, so the planting cycle of leek can be longer than that of general crops. Leek is not resistant to high temperature, and when the temperature exceeds 24℃, the growth of the plant will slow down, and when the temperature is higher than 30℃, the leek leaves will turn yellow. High temperature can eliminate leek maggot, but it also causes obvious damage to leek. When leek is attacked by pests during planting, this method has great limitations. Therefore, this method can only be used to eliminate pests before and after the planting cycle of leek, and the high temperature at that time cannot completely eliminate leek maggot, and the eggs in the deep soil will survive and reproduce when the soil temperature returns to normal.

[0004] In addition, it is difficult to estimate the number of leek maggot in the soil. Some people have proposed using attractants in farmland to attract leek maggot to the surface of the soil, and estimating the distribution of leek maggot in the field by manual visual estimation. However, after spraying the attractant, the leek maggot will quickly drill back into the soil, and this method has limited efficiency and cannot estimate the large area of leek field at the same time. In addition, the estimation by naked eye has deviation and is not convenient for statistics, which limits the prevention and control of pests and diseases in the later period. SUMMARY

[0005] Therefore, it is necessary to estimate the density of leek maggot in the farmland, which limits the prevention and control of pests and diseases in the later period. The present application provides a leek maggot density estimation method and system.

[0006] The present application discloses a leek maggot density estimation method, which comprises the following steps:

[0007] S1: constructing a leek maggot recognition model based on neural network.

[0008] S2: collecting a single frame image of leek field, and then performing segmentation processing on the single frame image to obtain a target image corresponding to a target vegetable plot area in the single frame image.

[0009] Wherein, the leek field is supplemented with light by a light supplementing module under shading condition, and then a single frame image is collected. The light supplementing frequency of the light supplementing module is consistent with the collection frequency of the single frame image.

[0010] S3: pre-processing the target image, and then obtaining an enhanced target image.

[0011] S4: respectively intercepting images of multiple appearances in the enhanced target image, obtaining leek maggot training samples and other appearance training samples, and then constituting a sample set for training the leek maggot recognition model.

[0012] S5: completing initialization of the leek maggot recognition model, setting a loss function and a training function, iteratively training the leek maggot recognition model by using the sample set, and retaining verified neural network model parameters, and then completing training of the leek maggot recognition model.

[0013] S6: collecting leek field images to be estimated in density according to steps S2 and S3, and sequentially pre-processing and enhancing the leek field images after segmentation.

[0014] S7: using the trained leek maggot recognition model to recognize the region of the leek field image to be estimated, recognizing the leek maggot region in the leek field image to be estimated, and then obtaining the total number P of target pixels corresponding to the leek maggot region in the whole leek field image to be estimated. all .

[0015] S8: according to a preset average pixel number of a single leek maggot and then calculating the number N of leek maggots in the leek field image to be estimated:

[0016]

[0017] S9: according to the actual area of the leek field to be estimated, calculating the number of leek maggots in a unit area, i.e. the leek maggot density of the leek field to be estimated.

[0018] As a further improvement of the present application, in step S2, a region of interest is set in the single frame image. The region of interest is used to extract the target image in the whole single frame image, and then the segmentation of the single frame image is realized.

[0019] As a further improvement of the present application, in step S2, a single frame image is collected by an image collection module. The image collection module is arranged directly above the center point of the target leek field. The image collection module is directed to the target leek field, and the image collection module is projected on the ground with a view area not less than the area of the target leek field.

[0020] As a further improvement of the present invention, in step S2, the fill light module and the image acquisition module are installed at the same fixed position, and the fill light direction of the fill light module is facing the target vegetable patch.

[0021] As a further improvement of the present invention, the image acquisition module adopts a CCD camera, and the fill light module adopts a soft box.

[0022] As a further improvement of the present invention, in step S3, the method for preprocessing the target image includes the following process:

[0023] Eliminate the interference of light, noise and other factors in the target image to obtain a clearer target image.

[0024] As a further improvement of the present invention, in step S4, other appearance training samples include leek training samples and soil training samples.

[0025] As a further improvement of the present invention, in step S5, the leek maggot recognition model extracts color features in the training sample by quantizing the color histogram, thereby realizing the recognition of various appearances in the image.

[0026] As a further improvement of the present invention, in step S5, during the training phase of the leek maggot recognition model, the pixel δ occupied by the leek maggots in each leek maggot training sample is further calculated. i Perform quantity statistics and then calculate the average number of pixels per unit after the training of the leek maggot recognition model is completed

[0027]

[0028] Where m is the number of leek maggot training samples.

[0029] The present invention also discloses a leek maggot density estimation system, which uses any of the above-mentioned leek maggot density estimation methods to estimate the density of leek maggots in a leek field. The leek maggot density estimation system includes: a model construction module, an image acquisition module, a fill light module, an image processing module, and a calculation module.

[0030] The model building module is used to build a leek maggot recognition model based on neural network.

[0031] The image acquisition module is used to acquire single-frame images of the leek field, and is also used to acquire images of the leek field to be density estimated.

[0032] The fill-light module is used to fill in the leek field with intermittent light under shading conditions, thereby capturing single-frame images. The fill-light module's fill-light frequency is consistent with the single-frame image acquisition frequency.

[0033] The image processing module is used to first segment the single-frame image to obtain a target image corresponding to the target vegetable patch area within the single-frame image. This target image is then preprocessed to obtain an enhanced target image. Multiple image features are then captured from the enhanced target image to obtain leek maggot training samples and other training features, thereby forming a sample set for training the leek maggot recognition model. The image processing module is also used to sequentially segment, preprocess, and enhance the leek field image to obtain the vegetable patch image to be estimated.

[0034] The constructed leek maggot recognition model requires network training before application. During the training process, the leek maggot recognition model is initialized, a loss function and a training function are set, and the leek maggot recognition model is iteratively trained using a sample set. The validated neural network model parameters are retained to complete the training of the leek maggot recognition model. The trained leek maggot recognition model is then used to perform regional recognition on the image of the vegetable patch to be estimated, identifying the leek maggot area in the image.

[0035] The calculation module is used to calculate the total number of target pixels P corresponding to the leek maggot area in the entire image of the vegetable patch to be estimated according to the leek maggot area in the vegetable patch image to be estimated. all Then, according to a preset average number of pixels per unit occupied by a single leek maggot, Then, the number N of leek maggots in the image of the vegetable patch to be estimated is calculated. Then, based on the actual area of ​​the vegetable patch to be estimated, the number of leek maggots per unit area is calculated, that is, the leek maggot density of the vegetable patch to be estimated.

[0036] Compared with the prior art, the technical solution disclosed in the present invention has the following beneficial effects:

[0037] This leek maggot density estimation method constructs a neural network-based leek maggot recognition model and trains it using a sample set consisting of multiple images of leek fields. The trained leek maggot recognition model is then used to perform regional recognition on the image of the vegetable patch to be estimated, identifying the leek maggot region within the image. This identifies the total number of target pixels corresponding to the leek maggot region within the entire image. The number of leek maggots in the image is then calculated based on the average number of pixels per unit pixel, δ, occupied by individual leek maggots. Based on the actual area of ​​the vegetable patch to be estimated, the number of leek maggots per unit area, i.e., the leek maggot density of the patch to be estimated, is then calculated.

[0038] The leek maggot density estimation method can estimate the density of leeks at night or early morning, interval light supplement is carried out on the leek field under shading conditions by using a light supplement module, then a single frame image is collected, and the light supplement frequency of the light supplement module is consistent with the collection frequency of the single frame image, high temperature light is avoided, so that the interference on the leek maggot is reduced, the leek maggot is prevented from drilling back to the ground due to light stimulation, the accuracy of the leek maggot density estimation is ensured. The image is synchronously lighted for a short time while the single frame image is collected, so that the light is dark, a large number of shadows and noise points exist in the single frame image, the image quality is improved, the accuracy of the leek maggot density estimation is further improved, so that the leek maggot density of each leek plot in the leek field is effectively estimated, and sufficient theoretical basis is provided for the later disease and pest control.

[0039] The leek maggot density estimation system has the same beneficial effects as the above method, and will not be described here. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 A flowchart of the leek maggot density estimation method in the embodiment 1 of the present application;

[0041] Figure 2 A perspective view of the range of the CCD camera and the region of interest in the embodiment 1 of the present application;

[0042] Figure 3 A segmentation diagram of the target image in the embodiment 1 of the present application;

[0043] Figure 4 A schematic diagram of the image collection module arranged on each leek plot in other embodiments of the present application;

[0044] Figure 5 A sample diagram of a plurality of appearances in the embodiment 1 of the present application;

[0045] Figure 6 A perspective structure schematic diagram of the spraying device arranged above the leek field in the greenhouse in the embodiment 3 of the present application;

[0046] Figure 7 A Figure 6 A partial enlarged perspective structure schematic diagram of the spraying mechanism and the moving mechanism in the embodiment 3 of the present application;

[0047] Figure 8 A Figure 7 A perspective structure schematic diagram of the driving assembly in the embodiment 3 of the present application;

[0048] Figure 9 A Figure 8 A front view of the carrier in the embodiment 3 of the present application;

[0049] Figure 10A flowchart of a digital pest control method for leek planting sheds in Embodiment 5 of the present application. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present application will be clearly and completely described in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0051] It should be noted that when a component is referred to as being "mounted on" another component, it can be directly on the other component or there can be a middle component. When a component is referred to as being "disposed on" another component, it can be directly disposed on the other component or there can be a middle component. When a component is referred to as being "fixed on" another component, it can be directly fixed on the other component or there can be a middle component.

[0052] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description of the application herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

[0053] Embodiment 1

[0054] Please refer to Figure 1 The present embodiment provides a leek maggot density estimation method for estimating the density of leek maggots in a leek field. The leek field can include multiple grid-shaped plots. Each plot is planted with multiple rows of leeks with a certain plant spacing. The leeks are harvested after a growth cycle, leaving short (about 2-3 cm in length) roots in the plots. During a period after the end of harvesting, an attractant is sprayed onto the land of the plots, thereby attracting leek maggots to the surface of the land. Some steps in the leek maggot density estimation method in the present embodiment are image acquisition and density estimation during the period when the leek maggots are attracted to the surface of the land. The leek maggot density estimation method includes the following steps:

[0055] S1: Construct a leek maggot recognition model based on a neural network.

[0056] In this embodiment, the neural network of the leek maggot recognition model can extract color features. Color features are the most widely used visual features in image retrieval. The main reason is that color is often closely related to the objects or scenes contained in the image. In addition, compared with other visual features, color features have less dependence on the size, direction and viewing angle of the image, thereby having higher robustness. In this embodiment, the leek maggot captured on the ground has a milky white color feature, the leek in the leek field has a greenish color feature, and the soil in the leek field has a yellowish-brown color feature. The color features of these appearances are quite different and easy for the leek maggot recognition model to learn and recognize.

[0057] Color histogram is a color feature widely used in many image retrieval systems. It describes the proportion of different colors in the entire image, without considering the spatial position of each color, i.e., it cannot describe the objects or objects in the image. Color histogram is particularly suitable for describing images that are difficult to automatically segment. Therefore, in this embodiment, a leek maggot recognition model based on neural network is constructed to realize color quantization by using neural network method. The neural network of the leek maggot recognition model can use existing convolutional neural network (CNN), or other neural networks such as deep neural network (DNN) and recurrent neural network (RNN). The specific principles are not repeated here.

[0058] S2: Collecting a single frame image of the leek field, and then performing segmentation processing on the single frame image to obtain a target image corresponding to the target vegetable plot area in the single frame image.

[0059] In the shading condition, a light supplement module is used to supplement light in the leek field at intervals, and then a single frame image is collected. The light supplement frequency of the light supplement module is consistent with the collection frequency of the single frame image.

[0060] Please refer to Figure 2 In this embodiment, the collection of the single frame image can be completed by an image collection module. The image collection module is arranged directly above the center point of the target vegetable plot. The image collection module is directed at the target vegetable plot, and the image collection module projects an area on the ground that is not less than the area of the target vegetable plot. The image collection module can use a high-resolution CCD camera. The light supplement module can use a soft light box.

[0061] In this embodiment, a region of interest can be set in the collected single frame image, so that the target image in the entire single frame image can be extracted, and the segmentation of the single frame image can be realized. In the figure, the region of interest is Figure 2The F1 part of the vegetable plot in the image needs to be removed from the single frame image, and the ridge R around the vegetable plot needs to be removed as well. In this embodiment, the shooting field of view of the CCD camera can be fixed, and since the vegetable plot is generally in the form of a square grid, the shooting field of view of the CCD camera can at least cover a single vegetable plot F1, and can also capture part of the ridge around the vegetable plot, but does not need to capture the vegetable plot F2 adjacent to the single vegetable plot F1 to avoid interference. The illumination range of the soft light box can also be set in the same way. The target image obtained after the segmentation process is shown in the rectangular dashed box in Figure 3 .

[0062] Referring to Figure 4 , in this embodiment, a guide rail (not shown in the figure) can be arranged above the multiple vegetable plots, and a set of image acquisition modules and a matching light supplement module are arranged on the guide rail. The light supplement module and the image acquisition module are installed at the same fixed position, and the light supplement direction of the light supplement module is directly opposite the target vegetable plot. In this way, the image acquisition module and the light supplement module can move along the extension direction of the guide rail, so as to be positioned directly above each vegetable plot to perform image acquisition and auxiliary light supplement. Of course, in other embodiments, the image acquisition module and the light supplement module can also be arranged as shown in Figure 4 . A plurality of image acquisition modules and light supplement modules corresponding to the number of vegetable plots can be arranged on each vegetable plot.

[0063] In this embodiment, the leek density can be estimated at night or early morning. The light supplement module is used to supplement light to the leek field at intervals under the condition of shading, and then a single frame image is acquired. The light supplement frequency of the light supplement module is consistent with the acquisition frequency of the single frame image, avoiding high-temperature light, thereby reducing the interference on the leek maggot, avoiding the leek maggot from drilling back into the ground due to light stimulation, and ensuring the accuracy of the leek maggot density estimation. The image is synchronously lighted for a short time while the single frame image is acquired, avoiding the existence of a large number of shadows and noise points in the single frame image due to dark light, improving the image quality, and further improving the accuracy of the leek maggot density estimation.

[0064] S3: Preprocessing the target image to obtain an enhanced target image.

[0065] In this embodiment, the method for preprocessing the target image can include the following processes:

[0066] Eliminate the interference of light, noise and other factors in the target image. Of course, in other embodiments, the preprocessing method can also include existing steps in image processing such as edge enhancement, bending correction, shadow processing, etc., which will not be described here. The purpose is to obtain a clearer target image.

[0067] S4: In the enhanced target image, the images of multiple appearances are intercepted respectively to obtain the leek maggot training samples and other appearance training samples, and then a sample set for training the leek maggot recognition model is constituted. In this embodiment, the other appearance training samples can include leek training samples and soil training samples, and of course can also include some field sundries (such as gravel, other insect samples). The richer the sample set is, the more accurate the leek maggot recognition model trained is in recognizing the leek maggot, but at the same time, the training process is more complex, so the appropriate number and type of samples need to be selected.

[0068] Please refer to Figure 5 , (a) is one of the intercepted leek appearance samples; (b) is one of the intercepted soil appearance samples; (c) is one of the intercepted leek maggot appearance samples.

[0069] S5: The initialization of the leek maggot recognition model is completed, the loss function and the training function are set, the leek maggot recognition model is iteratively trained by using the sample set, and the neural network model parameters verified are retained, and then the training of the leek maggot recognition model is completed.

[0070] In this embodiment, as described above, the leek maggot recognition model can extract the color features in the training samples through quantization of the color histogram, and then realize the recognition of various appearances in the image. After continuous training and optimization, the leek maggot recognition model that can accurately recognize the area covered by the leek maggot in the vegetable plot image can be obtained.

[0071] In addition, in the training stage of the leek maggot recognition model, the number of pixels δ i occupied by the leek maggot in each leek maggot training sample is counted, and then after the training of the leek maggot recognition model is completed, the average pixel number δ

[0072]

[0073] In the formula, m is the sample number of the leek maggot training sample.

[0074] S6: The leek field image to be estimated in density is collected according to steps S2 and S3, and the leek field image is sequentially segmented and preprocessed and enhanced to obtain the vegetable plot image to be estimated.

[0075] It should be noted that all the steps before step S6 are to obtain the leek maggot recognition model trained. However, the image collected in step S2 is also collected in a period of time after the leek is harvested and the trapping agent is sprayed. From step S6, it corresponds to the actual application process of the leek maggot density estimation. The vegetable plot image to be estimated is the vegetable plot that needs to be estimated in density for subsequent pest control.

[0076] S7: The leek maggot recognition model trained is used for region recognition on the to-be-estimated vegetable plot image, a leek maggot region in the to-be-estimated vegetable plot image is recognized, and then a target total number P of pixels corresponding to the leek maggot region in the whole to-be-estimated vegetable plot image is obtained all .

[0077] S8: According to the unit average number of pixels occupied by a single leek maggot and then the number N of leek maggots in the to-be-estimated vegetable plot image is calculated

[0078]

[0079] S9: According to the actual area of the to-be-estimated vegetable plot, the number of leek maggots in a unit area, that is, the leek maggot density of the to-be-estimated vegetable plot, is calculated.

[0080] Embodiment 2

[0081] The embodiment provides a leek maggot density estimation system, which can use the leek maggot density estimation method in embodiment 1 to estimate the density of leek maggots in a leek field. The leek maggot density estimation system comprises a model construction module, an image acquisition module, a light supplementing module, an image processing module and a calculation module.

[0082] The model construction module is used to construct a leek maggot recognition model based on a neural network.

[0083] The image acquisition module is used to acquire a single frame image of a leek field and is also used to acquire a leek field image to be estimated in density.

[0084] The light supplementing module is used to supplement light in a leek field at intervals under shading conditions, and then acquire a single frame image. The light supplementing frequency of the light supplementing module is consistent with the acquisition frequency of the single frame image.

[0085] The image processing module is used to first perform segmentation processing on the single frame image to obtain a target image corresponding to a target vegetable plot region in the single frame image. Then, the target image is preprocessed to obtain an enhanced target image. Then, images of multiple appearances are respectively intercepted in the enhanced target image to obtain leek maggot training samples and other appearance training samples, and then a sample set used for training the leek maggot recognition model is formed. The image processing module is also used to sequentially perform segmentation and preprocessing enhancement on the leek field image to obtain a to-be-estimated vegetable plot image.

[0086] The leek maggot recognition model is trained before application, and in the training process, the leek maggot recognition model is initialized, a loss function and a training function are set, the leek maggot recognition model is iteratively trained by using a sample set, and the neural network model parameters verified are retained, so that the leek maggot recognition model is trained. The leek maggot recognition model trained is used for region recognition of the to-be-estimated vegetable plot image, and a leek maggot region in the to-be-estimated vegetable plot image is recognized.

[0087] The calculation module is configured to calculate a target total number P of pixels of the leek maggot region in the to-be-estimated vegetable plot image. all According to a preset average number of unit pixels occupied by a single leek maggot The number N of leek maggots in the to-be-estimated vegetable plot image is calculated. Then, according to the actual area of the to-be-estimated vegetable plot, the number of leek maggots per unit area, i.e., the leek maggot density of the to-be-estimated vegetable plot, is calculated.

[0088] Embodiment 3

[0089] The embodiment provides a spraying device for leek maggot control, which is used for spraying trapping agents and insecticides on leek fields respectively. The leek field can include a plurality of grid-shaped vegetable plots. A plurality of rows of leeks with a certain plant spacing are planted in each vegetable plot. The leeks are harvested after a growth cycle, leaving short roots (about 2-3 cm in length) in the vegetable plot. At this time, the spraying device can be used to spray pesticides in each vegetable plot. The spraying device provided by the embodiment can be applied in a greenhouse, and of course, in some embodiments, it can also be applied in an open field.

[0090] Please refer to Figure 6 , the spraying device can include a spraying mechanism 1, a moving mechanism 2, a leek maggot density estimation system, and a controller. The leek maggot density estimation system can apply the leek maggot density estimation method in embodiment 1, or directly use the leek maggot density estimation system in embodiment 2.

[0091] Please refer to Figure 7 , the spraying mechanism 1 is used for spraying trapping agents or insecticides to a plurality of vegetable plots in a leek field. The spraying mechanism 1 includes a plurality of sprayers 11, and can also include two liquid storage tanks 12.

[0092] The plurality of sprayers 11 are located above the leek field and are arranged along the width direction of the leek field. The plurality of sprayers 11 can use existing electric sprayers similar to those used for unmanned aerial vehicle pesticide spraying, and of course, other types of sprayers can also be selected, as long as they can meet the requirement that the sprayed mist pesticide is uniformly spread on the soil surface of the vegetable plot. The plurality of sprayers 11 can be installed on the same straight pipe support to achieve stability.

[0093] Two liquid storage tanks 12 are used to store the trapping agent and the insecticide respectively. Each liquid storage tank 12 is connected with multiple sprayers 11 through multiple connecting pipes. In other words, each sprayer 11 is connected with two liquid storage tanks 12 through two pipes respectively. Therefore, an electric valve electrically connected with the controller can be arranged on each pipe. When the trapping agent needs to be sprayed, the electric valve on the pipe connected with the trapping agent storage tank 12 is opened, and the electric valve on the pipe connected with the insecticide storage tank 12 is closed. Conversely, when the insecticide needs to be sprayed, the electric valve on the pipe connected with the insecticide storage tank 12 is opened, and the electric valve on the pipe connected with the trapping agent storage tank 12 is closed. Thus, the different agents can be sprayed without interference.

[0094] Please refer to Figure 8 and Figure 9 , the conveying end of the moving mechanism 2 is used to drive the spraying mechanism 1 to move along the length direction of the leek field. The moving mechanism 2 can include a carrier 21, a track 22 and a driving assembly 23.

[0095] The track 22 can be fixedly arranged above the leek field and extend in parallel to the length direction of the leek field. In the embodiment, the two ends of the track 22 can be fixedly connected to the opposite ends of the greenhouse, and the middle section of the track 22 can be supported by multiple node supports.

[0096] The carrier 21 is slidingly installed on the track 22, and the bottom of the carrier 21 is fixedly connected with the spraying mechanism 1. The driving assembly 23 is installed on the carrier 21 and is used to drive the carrier 21 to generate relative motion with the track 22, thereby forming the conveying end of the moving mechanism 2. In the embodiment, a T-shaped through slot 211 can be formed in the carrier 21 and extend through the front and back of the carrier 21. The carrier 21 can be slidingly connected with the track 22 through the T-shaped through slot 211. A strip-shaped receiving groove 212 can also be formed in the top of the carrier 21.

[0097] The driving assembly 23 can include a rack 231, a gear 232 and a driving motor 233, and can also include two symmetrically arranged mounting plates for mounting the gear. The rack 231 penetrates through the receiving groove 212 and is fixedly connected to the upper surface of the track 22, and the length of the rack 231 matches the carrier 21. The gear 232 is rotatably installed on the top of the carrier 21, and the gear 232 is in meshing connection with the rack 231. The driving motor 233 is used to drive the gear 232 to rotate, thereby realizing the relative motion between the carrier 21 and the track 22. In addition, multiple pulleys 213 can be rotatably connected to the two sides of the carrier 21 in the T-shaped through slot 211. The rolling surface of each pulley 213 is in contact with the track 22, thereby reducing the friction between the carrier 21 and the track 22.

[0098] The leek maggot density estimation system is used to estimate the leek maggot density of each plot of the leek field. The specific estimation process and principle have been described in the foregoing embodiments, and will not be repeated here.

[0099] The controller is used to:

[0100] (a), first control the moving mechanism 2 to drive the spraying mechanism 1 to spray a certain amount of trapping agent to each vegetable plot. In this embodiment, the trapping agent can be wood vinegar of a certain concentration. Since leek maggot is a light-averse organism living in the soil, a certain amount of wood vinegar is sprayed under the premise of shading, and the stimulating odor of the juice spontaneously flowing from the wound of the just-harvested leek is used, so that a part of the leek maggot can be effectively attracted to the ground, thereby facilitating the estimation of the leek maggot density of the whole vegetable plot according to the part of the leek maggot attracted to the ground in the subsequent process.

[0101] (b), after a preset time period after spraying the trapping agent to each vegetable plot, the leek maggot density estimation system estimates the leek maggot density of each vegetable plot.

[0102] By setting the preset time period, the odor of the trapping agent can be more comprehensively volatilized into the soil. The length of the preset time period can be determined according to the amount of trapping agent sprayed and the thickness of the soil layer and other factors. In actual operation, the best dose is summarized through many experiments.

[0103] (c), according to the size of the leek maggot density estimation value of each vegetable plot, the required amount of insecticide for the corresponding vegetable plot is set, and the moving mechanism 2 is controlled again to drive the spraying mechanism 1 to spray the corresponding amount of insecticide to each vegetable plot in turn. The amount of insecticide is positively correlated with the size of the leek maggot density estimation value.

[0104] The insecticide can be mixed with garlic oil of a certain concentration, or other environmentally friendly and residue-free insecticides can be selected.

[0105] In this embodiment, the leek maggot density estimation value reflects the real distribution of leek maggot in the vegetable plot, and the required amount of insecticide corresponding to the estimated density can be obtained by referring to an estimation value-dose table obtained through many experiments. The required amount of insecticide for each vegetable plot is determined by referring to the estimation value.

[0106] The spraying device sprays trapping agent to each vegetable plot in the leek field under shading conditions by setting a linearly movable spraying mechanism on the vegetable plots in the leek field, estimates the leek maggot density of each vegetable plot after a period of time, and sets the amount of insecticide to be sprayed according to the estimation value. The spraying mode of the spraying device is self-adaptively adjusted according to the degree of insect damage, and is suitable for local conditions. It can increase the amount of insecticide for leek fields with serious insect damage, increase the penetration depth of the insecticide, and improve the effect of insect control and prevention. It can also appropriately reduce the amount of insecticide for leek fields with light insect damage, while ensuring the effect of insect control and prevention, and taking into account the effect of cost saving.

[0107] In addition, in the embodiment, the spraying rate of the spraying mechanism 1 can be kept constant, and the moving speed of the moving mechanism 2 can be adjusted. Alternatively, the moving speed of the moving mechanism 2 can be kept constant, and the spraying rate of the spraying mechanism 1 can be adjusted. In either way, the amount of pesticide sprayed by the spraying mechanism 1 in each vegetable plot can be adjusted.

[0108] Embodiment 4

[0109] The embodiment provides a spraying method for leek maggot control, which can be applied to the spraying device in embodiment 3. The spraying method comprises the following steps:

[0110] (I) Spraying a certain amount of trapping agent into each vegetable plot of the leek field in sequence.

[0111] (II) After a preset time period after spraying the trapping agent in each vegetable plot, the leek maggot density of the corresponding vegetable plot is estimated.

[0112] (III) According to the leek maggot density estimation value of each vegetable plot, the required amount of insecticide for the corresponding vegetable plot is set, and the corresponding amount of insecticide is sprayed into each vegetable plot in sequence. The amount of insecticide is positively correlated with the size of the leek maggot density estimation value, and the light shielding condition is kept during spraying the trapping agent and the insecticide.

[0113] Embodiment 5

[0114] The embodiment provides a digital pest control method for a leek planting shed, which can be applied to the leek planting shed. The leek planting shed can comprise one or more leek fields, and each leek field can comprise a plurality of vegetable plots arranged in a grid shape. A plurality of rows of leeks with a certain plant spacing are planted in each vegetable plot.

[0115] Referring to Figure 10 , the digital pest control method can comprise the following steps:

[0116] I. Pest control preparation stage

[0117] (1) A spraying device is erected above the leek field in the leek planting shed, and the spraying device can adopt the spraying device in embodiment 3.

[0118] (2) According to the actual area of each vegetable plot, the amount of trapping agent required to be sprayed in each vegetable plot is set.

[0119] (3) The sampling frequency of the image acquisition module is set according to the center point of each vegetable plot, and the light supplement frequency of the light supplement module is set to be consistent with the sampling frequency.

[0120] (4) The inside of the leek planting shed is kept in a closed and light shielding condition.

[0121] In the pest control preparation stage, leeks in the leek field can also be harvested. The average height of the harvested leeks is maintained at 3-8 cm. The juice flowing out of the harvested leeks can also emit a trapping odor for the leek maggots underground.

[0122] II. Trapping agent spraying stage

[0123] The moving mechanism 2 drives the spraying mechanism 1 to move in a targeted manner, and sprays a corresponding amount of trapping agent to each plot of land in turn, and waits for a while, thereby trapping part of the leek maggots underground to the ground.

[0124] III. Density estimation stage

[0125] (1) After a preset period of time after spraying the trapping agent in each plot of land, the moving mechanism 2 drives the image acquisition module to acquire a single frame of image of each plot of land in turn.

[0126] (2) According to the single frame of image of each plot of land, the leek maggot density of each plot of land is estimated respectively, and the corresponding leek maggot density estimation value of each plot of land is obtained.

[0127] IV. Insecticide spraying stage

[0128] (1) According to the leek maggot density estimation value of each plot of land, a preset estimation value-dose table is used to look up the required amount of insecticide to be sprayed for each plot of land.

[0129] (2) The moving mechanism 2 drives the spraying mechanism 1 to move in a targeted manner, and sprays a corresponding amount of insecticide to each plot of land in turn, thereby completing a round of insect control period.

[0130] In the insecticide spraying stage, after spraying insecticide to each plot of land, the soil between the leek rows can also be turned over, and a certain amount of wood ash can be spread on the surface of the turned soil. After spreading wood ash on each plot of land, the plot of land can also be ventilated and dried. The drying period is 5-7 days. Wood ash can dry the soil on the one hand, destroy the humid environment that leek maggots like, on the other hand, black wood ash can fully absorb the energy of sunlight, and properly warm the soil, further accelerating the drying of the soil. It should be noted that leeks are not resistant to high temperature, so ventilation is required during drying, and cooling is required if necessary to protect the growth of leeks.

[0131] In addition, after each round of insect killing period ends, the leek maggot density estimation value of each vegetable plot and the amount of insecticide sprayed can be counted, and the statistical data of each round of insect killing period can be generated and added to a biological control database for real-time updating. The statistical data of multiple rounds of insect killing period in the biological control database can be regularly analyzed, and the data change curve reflecting the insect control effect of each vegetable plot and the whole leek planting shed can be obtained. The controller can send the regularly obtained data change curve to the interactive terminal, such as a mobile phone, a computer or a monitoring center, by wireless transmission, thereby providing data reference for relevant agricultural technicians, and facilitating the summary, improvement and prediction evaluation of the stage biological control.

[0132] It should be noted that in the above bait spraying stage, bait spraying stage and insecticide spraying stage, the conveying end of the moving mechanism 2 needs to move from the first end of the planting shed to the end at least three times. The first movement corresponds to the spraying of the bait, and after spraying the bait in multiple plots, it can return to the starting point. After waiting for a preset period of time, the second movement is performed, which corresponds to the collection of images of multiple vegetable plots. During the collection, the leek maggot density estimation value of each vegetable plot is also calculated. After collecting the images of each vegetable plot, the conveying end can immediately return to the first end, and then the third movement is performed, which corresponds to the spraying of the insecticide.

[0133] The insect control method of the embodiment can be used to build a spraying mechanism that can reciprocate along the length direction of the leek field in the leek planting shed, and then the bait spraying, leek maggot density estimation and insecticide spraying are performed on the leek field in turn. By spraying a certain amount of bait on each vegetable plot, a part of the leek maggot under each vegetable plot is attracted to the ground, and then the image of each vegetable plot is collected, and the leek maggot density of each vegetable plot is estimated by image recognition technology. Finally, according to the leek maggot density estimation value of each vegetable plot, the amount of insecticide required for each vegetable plot is set, and a corresponding amount of insecticide is sprayed on multiple vegetable plots in the leek planting shed. The amount of insecticide can be increased for the leek field with serious insect infestation, the penetration depth of the insecticide can be increased, and the insect control effect can be improved. The amount of insecticide can also be appropriately reduced for the leek field with light insect infestation, which can save costs while ensuring the insect control effect. The method can balance the high spraying efficiency and good spraying effect.

[0134] The technical features of the above embodiments can be combined in any way. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not exist, they should be considered as the scope of the description.

[0135] The above embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but should not be understood as a limitation on the scope of the application. It should be noted that for ordinary skilled persons in the art, several modifications and improvements can be made without departing from the concept of the present application, which are within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for estimating the density of leek maggots, characterized in that: It includes the following steps: S1: Construct a leek maggot recognition model based on neural network; S2: collecting a single-frame image of the leek field, and performing segmentation processing on the single-frame image to obtain a target image corresponding to a target vegetable patch area in the single-frame image; Wherein, under the light-blocking condition, a fill-in light module is used to fill in the light of the leek field at intervals, thereby collecting the single-frame image; the fill-in light frequency of the fill-in light module is consistent with the collection frequency of the single-frame image; S3: Preprocessing the target image to obtain an enhanced target image; S4: intercepting images of various appearances from the enhanced target image to obtain leek maggot training samples and other appearance training samples, thereby forming a sample set for training the leek maggot recognition model; S5: completing the initialization of the leek maggot recognition model, setting a loss function and a training function, iteratively training the leek maggot recognition model using the sample set, and retaining the verified neural network model parameters, thereby completing the training of the leek maggot recognition model; S6: collecting the leek field image to be density estimated according to steps S2 and S3, and performing segmentation and preprocessing enhancement on the leek field image in sequence to obtain the vegetable bed image to be estimated; S7: Using the trained leek maggot recognition model to perform region recognition on the vegetable patch image to be estimated, identify the leek maggot region in the vegetable patch image to be estimated, and then obtain the total number of target pixels P corresponding to the leek maggot region in the entire vegetable patch image to be estimated. all ; S8: The average number of pixels per unit occupied by a single leek maggot according to a preset Then the number N of leek maggots in the vegetable patch image to be estimated is calculated: S9: Calculate the number of leek maggots per unit area according to the actual area of ​​the vegetable patch to be estimated, that is, the leek maggot density of the vegetable patch to be estimated.

2. The method for estimating the density of leek maggots according to claim 1, wherein In step S2, a region of interest is set in the single-frame image; the region of interest is used to extract the target image in the entire single-frame image, thereby achieving segmentation of the single-frame image.

3. The method for estimating the density of leek maggots according to claim 1, wherein In step S2, the single-frame image is captured by an image acquisition module; the image acquisition module is set directly above the center point of the target vegetable patch; the framing direction of the image acquisition module is facing the target vegetable patch, and the framing area projected by the image acquisition module on the ground is not less than the area of ​​the target vegetable patch.

4. The method for estimating the density of leek maggots according to claim 3, wherein: In step S2, the fill light module and the image acquisition module are installed at the same fixed position, and the fill light direction of the fill light module is facing the target vegetable patch.

5. The method for estimating the density of leek maggots according to claim 4, wherein: The image acquisition module adopts a CCD camera; the fill light module adopts a soft box.

6. The method for estimating the density of leek maggots according to claim 1, wherein: In step S3, the method of preprocessing the target image The following processes are included: Eliminate interference from light, noise and other factors in the target image to obtain a clearer target image.

7. The method for estimating the density of leek maggots according to claim 1, wherein: In step S4, the other appearance training samples include leek training samples and soil training samples.

8. The method for estimating the density of leek maggots according to claim 1, wherein: In step S5, the leek maggot recognition model extracts color features in the training sample by quantizing the color histogram, thereby realizing the recognition of various appearances in the image.

9. The method for estimating the density of leek maggots according to claim 1, wherein: In step S5, during the training phase of the leek maggot recognition model, the number of pixels occupied by leek maggots in each leek maggot training sample is further calculated. i Perform quantity statistics, and then calculate the average number of pixels per unit after the training of the leek maggot recognition model is completed. Wherein, m is the number of samples of the leek maggot training samples.

10. A leek maggot density estimation system, characterized in that: The leek maggot density estimation method according to any one of claims 1 to 9 is used to estimate the density of leek maggots in a leek field; the leek maggot density estimation system comprises: A model building module, which is used to build a leek maggot recognition model based on a neural network; An image acquisition module, which is used to acquire single-frame images of the leek field and also used to acquire images of the leek field to be density estimated; A fill light module is used to fill light to the leek field at intervals under shading conditions, thereby capturing the single-frame image; the fill light frequency of the fill light module is consistent with the capture frequency of the single-frame image; An image processing module is configured to first segment a single-frame image to obtain a target image corresponding to a target vegetable patch area in the single-frame image; then preprocess the target image to obtain an enhanced target image; then, images of various appearances are captured from the enhanced target image to obtain leek maggot training samples and other appearance training samples, thereby forming a sample set for training the leek maggot recognition model; the image processing module is further configured to sequentially segment, preprocess, and enhance the leek field image to obtain a vegetable patch image to be estimated; Among them, the constructed leek maggot recognition model needs to undergo network training before application. During the training process, the leek maggot recognition model is initialized, a loss function and a training function are set, the leek maggot recognition model is iteratively trained using the sample set, and the verified neural network model parameters are retained to complete the training of the leek maggot recognition model; the trained leek maggot recognition model is used to perform regional recognition on the vegetable patch image to be estimated, and the leek maggot area in the vegetable patch image to be estimated is identified; The leek maggot density estimation system also includes: A calculation module is used to calculate the total number of target pixels P corresponding to the leek maggot area in the entire vegetable patch image to be estimated based on the leek maggot area in the vegetable patch image to be estimated. all ; Then according to a preset average number of pixels per unit occupied by a single leek maggot Then, the number N of leek maggots in the image of the vegetable patch to be estimated is calculated; and then, based on the actual area of ​​the vegetable patch to be estimated, the number of leek maggots per unit area is calculated, that is, the leek maggot density of the vegetable patch to be estimated.

Images