Locust monitoring and identifying system and method

Abstract

The invention relates to the technical field of locusts, in particular to a locust monitoring and identifying system and method. The three-dimensional model reconstruction module constructs a model in a three-dimensional space according to the characteristic postures and sizes of the locusts in a ratio of 1: 1, then transmits the constructed model in the three-dimensional space to the target comparison module, and compares the sizes, colors and characteristics of various locusts in a locust database in the target comparison module and the locust recognition module. And after comparison is completed, the locust database compares the state of the locust being an adult or larva according to the obtained information of the locust, compares the habits and prevention modes of the locust, and uploads the comparison result to a display in a monitoring room through a data transmission module to remind a worker to make a treatment mode corresponding to the locust.

Description

Technical Field

[0001] The invention relates to the technical field of locusts, in particular to a locust monitoring and identification system and method. Background Art

[0002] The meadow locust is a devastating pest that harms grasslands. It has a omnivorous diet, a wide distribution, and migrates quickly. It is known as one of the four major disasters in agricultural and animal husbandry production. In recent years, the damage caused by meadow locusts to grasslands has become increasingly serious, with the annual damage area exceeding 150 million mu, resulting in large-scale grassland degradation and a decrease in grass production, which has seriously restricted the development of my country's animal husbandry. Timely and accurate monitoring and early warning of meadow locust occurrences are of great significance to the healthy development of animal husbandry in pastoral areas, the sustainable development of the regional economy, and social stability.

[0003] With the development of machine learning and deep learning, image recognition and target detection have received widespread attention. Locust detection and recognition, as the basis for studying the mechanism of locust outbreaks and establishing prediction models, is an important means to establish a dynamic monitoring and early warning system for locust disasters. Currently, there are few studies in the field of locust recognition and they are relatively backward. There are obvious differences in recognition accuracy and speed between traditional recognition methods and emerging target detection methods. In addition, traditional locust recognition methods cannot provide different warning information according to the different stages of locust growth during real-time monitoring of locusts.

[0004] In summary, in order to solve the technical problem raised in this article, the present invention proposes a locust monitoring and identification system and method. Summary of the invention

[0005] The present invention proposes a locust monitoring and identification system, which includes:

[0006] An image acquisition module, wherein the image acquisition module is used to acquire images of a target area, and the image acquisition module acquires target images at multiple angles and in different spatial areas;

[0007] An image preprocessing module, which processes the image information collected by the image acquisition module; the image preprocessing module includes target recognition and target segmentation;

[0008] A locust identification module, wherein the locust identification module uses a convolutional neural network to identify and analyze locust individuals identified by the target in the image preprocessing module; and the locust identification module has a locust database inside, and the database contains body features of known locust species;

[0009] A three-dimensional model reconstruction module, wherein the three-dimensional model reconstruction module reconstructs a three-dimensional model of the locust segmented by the target in the image preprocessing module;

[0010] A target comparison module, wherein the target comparison module is used to analyze the structural characteristics of locust individuals in the three-dimensional model reconstruction module. After the analysis is completed, the target comparison module compares the locust database in the target recognition module to determine the type and characteristics of the locust;

[0011] The data transmission module is used to transmit the data determined in the target comparison module to the monitoring room. An early warning analysis module is arranged inside the monitoring room. The early warning analysis module generates different warning information according to different data.

[0012] Further; the early warning analysis module is based on;

[0013] S=xy

[0014]

[0015] S is the size of the locust, x is the length of the locust in the image, and y is the width of the locust in the image; when S>5th instar nymph size, it is recorded as T; when S<5th instar nymph size, it is recorded as U; M is the contrast coefficient.

[0016] Furthermore, the target segmentation in the image preprocessing module includes segmenting the sky and grass in the target area into two different areas using the Otsu inter-class difference method; and then extracting the segmentation line between the sky and grass using the Sobel edge detection operator.

[0017] Furthermore, the image acquisition module also includes a remote sensing monitoring module; the remote sensing monitoring module performs remote sensing monitoring of vegetation, altitude, temperature, precipitation and soil in the target area through a geographic information system, a global satellite positioning system and a satellite remote sensing system to obtain environmental characteristics of the target area;

[0018] The priority determination module is used to classify the locusts segmented from the target area, and the priority determination module cooperates with the environmental features in the remote sensing monitoring module to make warning information in the monitoring room according to the priority.

[0019] Furthermore, the data between the priority determination module and the locust database is interoperable, and the locust species alarm priority corresponding to the priority determination module can be modified accordingly from within the locust database.

[0020] Further; the priority determination module includes a data analysis module, and the data analysis module independently selects the locust priority determination based on the data collected in the target area and the data in previous years; the data analysis module passes;

[0021] formula:

[0022] Where R xy It is the correlation coefficient between element x and element y, which is used to indicate the degree of correlation between the two elements, where x represents the vegetation element index and y represents the surface temperature index.

[0023] Furthermore, the data analysis module analyzes the data acquired by the remote sensing detection module, and the data analysis module analyzes the habits of the locust species in the locust database to analyze the destructiveness of the corresponding locust habits in the target area.

[0024] The beneficial effects of the present invention are as follows:

[0025] The three-dimensional model reconstruction module constructs a 1:1 model of the three-dimensional space based on the characteristic body shape and size of the locust, and then passes the constructed three-dimensional space model to the target comparison module. The target comparison module compares the size, color and characteristics of various locusts in the locust database in the locust identification module. After the comparison is completed, the locust database compares the adult or larval state based on the information of the locust, and compares the habits and prevention methods of this type of locust. The comparison results are uploaded to the display in the monitoring room through the data transmission module to remind the staff to deal with the locusts accordingly. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a system flow chart of the present invention; DETAILED DESCRIPTION

[0027] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0028] Embodiment 1:

[0029] like Figure 1 As shown; A locust monitoring and identification system, the monitoring and identification system comprises:

[0030] An image acquisition module, wherein the image acquisition module is used to acquire images of a target area, and the image acquisition module acquires target images at multiple angles and in different spatial areas;

[0031] An image preprocessing module, which processes the image information collected by the image acquisition module; the image preprocessing module includes target recognition and target segmentation;

[0032] A locust identification module, wherein the locust identification module uses a convolutional neural network to identify and analyze locust individuals identified by the target in the image preprocessing module; and the locust identification module has a locust database inside, and the database contains body features of known locust species;

[0033] A three-dimensional model reconstruction module, wherein the three-dimensional model reconstruction module reconstructs a three-dimensional model of the locust segmented by the target in the image preprocessing module;

[0034] A target comparison module, wherein the target comparison module is used to analyze the structural characteristics of locust individuals in the three-dimensional model reconstruction module. After the analysis is completed, the target comparison module compares the locust database in the target recognition module to determine the type and characteristics of the locust;

[0035] A data transmission module, which is used to transmit the data determined in the target comparison module to the monitoring room. The monitoring room is provided with an early warning analysis module, which generates different warning information according to different data;

[0036] The early warning analysis module is based on;

[0037] S=xy

[0038]

[0039] S is the size of the locust, x is the length of the locust in the image, and y is the width of the locust in the image; when S>5th instar nymph size, it is recorded as T; when S<5th instar nymph size, it is recorded as U; M is the contrast coefficient;

[0040] The target segmentation in the image preprocessing module includes segmenting the sky and the grass in the target area into two different areas using the Otsu inter-class difference method; then extracting the segmentation line between the sky and the grass using the Sobel edge detection operator;

[0041] The specific process is as follows;

[0042] The monitoring area is monitored and identified by an image acquisition module, and real-time video is taken. The image acquisition module may be a high-definition camera, an unmanned aerial vehicle or other shooting means. The captured image is then transmitted to an image preprocessing module. The image preprocessing module processes the image information collected by the image acquisition module. The main processing method is target recognition, which mainly recognizes individual locusts or locust groups captured in the image. The specific recognition method is to extract features of the locusts, such as edges, textures, shapes, etc. of the locusts in the collected image, through a HOG feature extraction algorithm.

[0043] Since the color of locusts may be different from the background of the monitoring area (such as grass and sand), for example, when locusts are in a swarming state, their color mostly changes to yellow and brown, while when they are in a non-swarming state, their color is mostly green. Therefore, after identifying the locust targets, segmenting the locusts and the image background in the images taken at various angles can reduce the impact on the monitoring of locusts after the color of the locust individuals and the background appear the same or have color differences. The specific segmentation method is to set a threshold to divide the image into different areas according to the grayscale value or color information of the pixels in the collected image; then use the Canny edge detection algorithm to detect the edge of the target locust in the collected image, first use a Gaussian filter to smooth the original image to remove noise; then usually use the Sobel edge detection operator to calculate the gradient strength and direction of the locust pixels in the collected image; after obtaining the gradient, apply the non-maximum suppression technique to eliminate the stray response caused by edge detection, and only The pixel points corresponding to the local gradient maximum are retained to obtain the refined edge of the locust; then the strong edge and the weak edge are distinguished by setting the high threshold and the low threshold; the pixel points above the high threshold are considered to be strong edges, and the pixel points below the low threshold are discarded; the target edge in the collected image can be effectively detected by the above Canny algorithm; after the feature of the locust is collected, it can be obtained; the characteristic posture and size of the locust are constructed 1:1 by the three-dimensional model reconstruction module. The model inside the three-dimensional space is then passed to the target comparison module. The target comparison module compares the size, color and characteristics of various locusts in the locust database in the locust identification module. After the comparison is completed, the locust database is compared with the adult or larvae state according to the information of the locust, and the habits and prevention methods of this type of locust are compared, and the comparison results are uploaded to the display in the monitoring room through the data transmission module to remind the staff to deal with the locusts accordingly;

[0044] When monitoring locusts, the early warning analysis module passes after analyzing the locust size information;

[0045] S=xy

[0046]

[0047] S is the size of the locust, x is the length of the locust in the image, and y is the width of the locust in the image; when S>5th instar nymph size, it is recorded as T; when S<5th instar nymph size, it is recorded as U; M is the contrast coefficient;

[0048] The size S of locusts in the target area is calculated through the early warning analysis module, and then the size of all locusts is compared with the size of the 5th instar nymphs. The size of the 5th instar nymphs varies according to the species of locusts. Here, the Asian migratory locust is taken as an example. The locust size S greater than the 5th instar nymph is recorded as T, and the locust size less than the 5th instar nymph is recorded as U. M is the ratio of all locusts with a size greater than the 5th instar nymph to those with a size less than the 5th instar nymph.

[0049] In the above calculation, when M>1, it means that the number of adults is greater than that of nymphs, and the conditions for a locust plague are met. At this time, the early warning analysis module issues an emergency warning to remind the staff to take prevention measures as soon as possible; when M<1, it means that the number of adults is less than that of nymphs. At this time, the staff is reminded to take advance defense measures to control the locusts within the range before the locust plague.

[0050] And on the basis of the above, by including the Ostu inter-class difference method in the image preprocessing module, the sky and grass of the target area collected by the image acquisition module are divided into two different areas. Due to the different colors of the sky and the grass, when the locusts jump into the air, the shape and background color of the locusts on the grass are different. Therefore, after the sky and the grass are segmented, the Sobel edge detection operator is used to extract the dividing line between the sky and the grass. It is achieved that the locusts jumping in the sky and the locusts on the grass can be identified more accurately, thereby improving the monitoring effect of the locusts.

[0051] Embodiment 2:

[0052] The image acquisition module also includes a remote sensing monitoring module; the remote sensing monitoring module performs remote sensing monitoring of vegetation, altitude, temperature, precipitation and soil in the target area through a geographic information system, a global satellite positioning system and a satellite remote sensing system to obtain environmental characteristics of the target area;

[0053] A priority determination module, wherein the priority determination module is used to classify locusts segmented from the target area, and the priority determination module cooperates with environmental features in the remote sensing monitoring module to generate warning information in the monitoring room according to the priority;

[0054] The data between the priority determination module and the locust database are intercommunication, and the locust species alarm priority corresponding to the priority determination module can be modified correspondingly from inside the locust database;

[0055] The specific process is as follows;

[0056] On the basis of the above embodiment, the image acquisition module also includes a remote sensing monitoring module, so that the remote sensing monitoring module can monitor weather information. The specific means are, for example, remote sensing satellites. When the remote sensing monitoring module monitors the target area, it can monitor the temperature, light and precipitation of the target area in real time. According to the living habits of locusts, locusts mostly like high temperature and drought environments. In addition, the temperature rises from May to August, and the locusts have also gone through a buffered breeding period, which will cause locusts to migrate on a large scale. At this time, the remote sensing monitoring module monitors the weather conditions of the target area in real time. If the remote sensing monitoring module monitors that the temperature continues to rise or the precipitation gradually decreases during the period between May and August, the remote sensing monitoring module transmits the monitoring information to the inside of the monitoring room through the data transmission module, and the computer in the monitoring room receives the feedback information, and according to the priority determination module (the priority determination module is an adaptive alarm system in the prior art), the adaptive alarm system performs simulation learning through a convolutional neural network, controls the alarm information level for the monitoring results of different locusts, and reminds the staff to be ready to prevent locusts at all times, so that the staff can better monitor and monitor the locusts.

[0057] Embodiment three:

[0058] The specific process is as follows;

[0059] The remote sensing monitoring module is connected to the locust identification module in the first embodiment and the locust database in the locust identification module. The remote sensing monitoring module monitors the vegetation, altitude and soil of the target area, monitors the vegetation, altitude and soil of the target area at all times, and focuses on monitoring the corresponding locusts according to the vegetation information of the target area. For example, the habit of the Asian migratory locust is to like plants such as seedlings (wheat, rice and corn), and the cotton locust likes plants such as cotton and soybean. If the vegetation in the monitoring target area is a seedling plant, when the locust identification module identifies that the locust in the image collected by the image acquisition module is the Asian migratory locust, the priority determination module locates the Asian migratory locust as the alarm information for priority processing, and reminds the staff to take precautions against the Asian migratory locust. If the locust identification module identifies that the locust in the image collected by the image acquisition module is the cotton locust, the priority determination module sets the alarm information priority of the cotton locust lower than that of the Asian migratory locust. On the contrary, if the vegetation in the target area is plants such as cotton and soybean, the priority determination module raises the alarm level of the cotton locust to above the level of the Asian migratory locust, so that the staff can take corresponding preventive measures for different locust types according to different vegetation.

[0060] Embodiment 4:

[0061] The priority determination module includes a data analysis module, and the data analysis module autonomously selects locust priority determination based on data collected in the target area and data from previous years;

[0062] The data analysis module analyzes the data acquired by the remote sensing detection module, and the data analysis module analyzes the habits of locust species in the locust database to analyze the destructiveness of the corresponding locust habits in the target area;

[0063] The specific principles are as follows;

[0064] By including a data analysis module in the priority determination module, the data analysis module autonomously selects the priority of locusts according to the data collected in the target area and the vegetation elements and surface temperature index of previous years. The selection method is to continuously learn the locust data of previous years through a convolutional neural network, and then autonomously select and adjust the priority of locusts: the data analysis module uses a formula;

[0065]

[0066] Where R xy is the correlation coefficient between element x and element y, which is used to indicate the degree of correlation between the two elements, where x represents the vegetation element index and y represents the surface temperature index. xy Between [-1, 1], when R xy >0, it means that there is a positive correlation between factor x and factor y. xy When <0, it means that there is a negative correlation between factor x and factor y, and when R xy The closer the absolute value of is to 1, the more factors x and y are in line with the living environment of locusts. xy The closer the absolute value is to 0, the less suitable the factors x and y are for the living environment of locusts.

[0067] According to the above formula, in the current year, R xy When the absolute value of R is close to 1, the locust alarm is raised to the highest level to remind staff to take urgent measures to prevent locusts. xy When the absolute value of is close to 0, the alarm to locusts will be adaptively decreased;

[0068] The data analysis module can be selected from the data in the remote sensing monitoring module to analyze the destructiveness of different types of locusts to the target area according to the habits of different locust species in the locust database; for example, when R xy When the absolute value is close to 1, the corresponding locust food type can be analyzed, which is used to analyze the degree of damage caused by locusts to vegetation in the target area, allowing staff to further judge the destructiveness of locusts.

Claims

1. A locust monitoring and identification system, characterized in that: The monitoring and identification system includes: An image acquisition module, wherein the image acquisition module is used to acquire images of a target area, and the image acquisition module acquires target images at multiple angles and in different spatial areas; An image preprocessing module, which processes the image information collected by the image acquisition module; the image preprocessing module includes target recognition and target segmentation; A locust identification module, wherein the locust identification module uses a convolutional neural network to identify and analyze locust individuals identified by the target in the image preprocessing module; and the locust identification module has a locust database inside, and the database contains body features of known locust species; A three-dimensional model reconstruction module, wherein the three-dimensional model reconstruction module reconstructs a three-dimensional model of the locust segmented by the target in the image preprocessing module; A target comparison module, wherein the target comparison module is used to analyze the structural characteristics of locust individuals in the three-dimensional model reconstruction module. After the analysis is completed, the target comparison module is compared with the locust database in the locust identification module to determine the type and characteristics of the locust; The data transmission module is used to transmit the data determined in the target comparison module to the monitoring room. An early warning analysis module is arranged inside the monitoring room. The early warning analysis module generates different warning information according to different data.

2. A locust monitoring and identification system as claimed in claim 1, characterized in that: The early warning analysis module is based on; S=xy S is the size of the locust, x is the length of the locust in the image, and y is the width of the locust in the image; when S>5th instar nymph size, it is recorded as T; when S<5th instar nymph size, it is recorded as U; M is the contrast coefficient.

3. A locust monitoring and identification system as claimed in claim 1, characterized in that: The target segmentation in the image preprocessing module includes segmenting the sky and the grass in the target area into two different areas using the Otsu inter-class difference method; and then extracting the segmentation line between the sky and the grass using the Sobel edge detection operator.

4. A locust monitoring and identification system as claimed in claim 1, characterized in that: The image acquisition module also includes a remote sensing monitoring module; the remote sensing monitoring module performs remote sensing monitoring of vegetation, altitude, temperature, precipitation and soil in the target area through a geographic information system, a global satellite positioning system and a satellite remote sensing system to obtain environmental characteristics of the target area; The priority determination module is used to classify the locusts segmented from the target area, and the priority determination module cooperates with the environmental features in the remote sensing monitoring module to make warning information in the monitoring room according to the priority.

5. A locust monitoring and identification system as claimed in claim 4, characterized in that: The data between the priority determination module and the locust database are interconnected, and the locust type alarm priority corresponding to the priority determination module can be modified accordingly from within the locust database.

6. A locust monitoring and identification system as claimed in claim 5, characterized in that: The priority determination module includes a data analysis module, which autonomously selects locust priority determination based on data collected in the target area and data from previous years; the data analysis module determines the priority of locusts through; official: Where Rxy is the correlation coefficient between element x and element y, i and i Corresponding to x1 and y1, it is used to indicate the correlation between two elements, where x represents the vegetation element index and y represents the surface temperature index.

7. A locust monitoring and identification system as claimed in claim 6, characterized in that: The data analysis module analyzes the data acquired by the remote sensing detection module, and the data analysis module analyzes the habits of locust species in the locust database to analyze the destructiveness of the corresponding locust habits in the target area.

Images