CN113378700B - Grassland rat damage dynamic monitoring method based on unmanned aerial vehicle aerial image - Google Patents

Abstract

The invention provides a grassland rat damage dynamic monitoring method based on aerial images of an unmanned aerial vehicle, which comprises the following steps: acquiring aerial images of different time spaces of a grassland area to be monitored by using an unmanned aerial vehicle; preprocessing the acquired image, and then interpreting to obtain an interpreted image; the interpreting includes classifying the target object; performing classification precision evaluation on the interpreted images, and taking the images meeting the preset precision requirement as a result set; and comparing the images of the same plot in different time spaces in the result set, marking the change area of the target object according to the time flow, and drawing a rat damage development trend graph. The classification precision of the interpreted images is judged through the precision evaluation confusion matrix table, and the method is objective and convenient; a specific rat damage prevention and control planning map is formulated by combining a rat damage development trend map of an area to be monitored with rat damage prevention and control methods suitable for different damage degrees and referring to local terrain, slope and mechanization level, and data support is provided for scientific rat damage prevention and control work.

Description

A dynamic monitoring method for grassland rodents based on aerial images taken by UAV

technical field

The invention relates to the field of prevention and control of prairie pika diseases, in particular to a method for dynamic monitoring of prairie rodent pests based on unmanned aerial vehicle images.

Background technique

In recent years, UAV remote sensing has rapidly emerged in grassland resource monitoring and grassland ecology due to its advantages of high resolution, high timeliness, strong maneuverability, and low-altitude flight under clouds. Unmanned aircraft, referred to as unmanned aerial vehicle (UAV), is an unmanned aircraft that can carry a variety of equipment, perform multi-field tasks, and fly autonomously through remote control equipment. The combination of UAV and remote sensing technology, namely Unmanned Aerial Vehicle Remote Sensing (UAVRS), uses UAV as a carrier to carry various sensors such as cameras (including visible light cameras, multispectral cameras, thermal infrared cameras, etc.), laser radars, etc. , to obtain a platform for low-altitude high-resolution remote sensing data. Compared with the traditional satellite-based space remote sensing, UAV remote sensing has the advantages of low-altitude flight under the clouds and high mobility, which makes up for the defect that satellite remote sensing cannot obtain clear images due to cloud cover; at the same time, it is time-effective, The characteristics of high temporal and spatial resolution are also not available in traditional satellite remote sensing with a long revisit period and hundreds of kilometers above the ground.

Grassland ecosystem is one of the most important terrestrial ecosystems and plays a pivotal role in the ecological environment. As an integral part of the grassland ecosystem, grassland resources play an important role in the cycle of the grassland ecosystem system. In order to monitor grassland resources quickly, conveniently, accurately and effectively, scholars at home and abroad have begun to use satellite-based space remote sensing to monitor Monitoring grassland vegetation coverage, grassland rodent damage, grassland ungulate wildlife, and estimation of grassland biomass. However, traditional medium and low-resolution satellite remote sensing images have a long acquisition cycle and are easily affected by climate, so they cannot obtain effective information on the ground in local areas. Compared with satellite remote sensing, UAV remote sensing has the characteristics of high resolution and image acquisition under the cloud, which can significantly reduce the impact of mixing effects on monitoring accuracy, and effectively compensate for the low surface resolution and long revisit cycle of satellite space and air remote sensing systems. , being greatly affected by water vapor, etc., it provides a new method for the application research of remote sensing for monitoring grassland resources in small and medium scales.

In recent years, due to comprehensive factors such as global climate change and human overuse, grasslands around the world have been degraded to varying degrees and have become increasingly serious, and rodents have occurred frequently. Therefore, scientific understanding and quantitative evaluation of the spatiotemporal dynamics and distribution of grassland pests are urgently needed. It is a new research idea to apply UAVs to the monitoring and evaluation of rodent damage area, rodent damage distribution, plant community coverage and aboveground biomass in alpine grassland.

Existing technologies focus more on the processing and classification of images taken by drones. For example, the patent with the publication number CN11881728A discloses a method for monitoring grassland rodent damage based on low-altitude remote sensing, and specifically discloses the method for obtaining the optimal Surface characteristics of grasslands to be monitored for rodent infestation. However, after obtaining the characteristics of rodent damage, there is no method or inspiration for applying the characteristics to the actual problem of rodent damage control.

Contents of the invention

In order to solve the above technical problems, the present invention provides a method for dynamic monitoring of grassland rodent damage based on unmanned aerial vehicle images, which improves the accuracy of interpretation results by means of precision evaluation, and draws the development trend map of rodent damage according to the time flow. Provide data support for the actual prevention and control of rodents on grasslands.

The invention provides a method for dynamic monitoring of grassland rodent damage based on unmanned aerial vehicle images, which is characterized in that it comprises steps:

Step S1. Use the drone to obtain aerial images of the grassland area to be monitored at different times and spaces;

Step S2. Preprocessing the collected image, and then interpreting to obtain the interpreted image; the interpreting includes classifying the target object to obtain a result set;

Step S3. Comparing the images of different time and space in the result set, marking the changing area of the target object according to the time flow, and drawing a development trend diagram of the rodent infestation.

In some preferred embodiments, the step S2 further includes: adjusting the flight parameters of the drone according to the classification accuracy of the target object.

In some preferred embodiments, the preprocessing in step S2 includes: sequential screening, dense point cloud generation, digital surface model output, projection transformation and geometric correction, and clipping.

In some preferred embodiments, the target objects include: grassland, bare ground and rat holes.

In some preferred embodiments, the step S2 also includes that the method for classifying the target object includes:

Build a support vector machine model and train it;

Generate classification accuracy evaluation confusion matrix table according to classification accuracy; described classification accuracy evaluation items include: overall classification accuracy, kappa coefficient, misclassification error, omission error, mapping accuracy and user accuracy;

According to the classification accuracy, it is judged whether the training of the support vector machine model is completed.

In some preferred embodiments, step S3 also includes:

Divide the image of the same plot in the same time and space into several small grids, calculate the individual mousehole density in each grid, and calculate the overall average mousehole density in all grids;

The small grid with the individual mouse hole density higher than the overall average mouse hole density is marked as a high-risk area, otherwise it is marked as a low-risk area;

Draw a graph of the development trend of rodent damage from high-risk areas to low-risk areas under the spatio-temporal conditions;

Repeat the above steps for the images of the same plot in different time and space, and draw the development trend map of rodent damage in this plot;

Repeat the above steps for the images of different plots to obtain the development trend map of rodent damage in the entire area to be monitored.

In some preferred embodiments, the graph of the development trend of the rodent damage includes: a graph of the development direction and trend of the rodent damage and a graph of the development trend of the rodent damage.

In some preferred embodiments, the method for dynamic monitoring of the development trend of grassland rodent damage also includes:

Step S4. Determine the control method and labor cost applicable to the rodent damage in the area to be monitored, and draw a rodent damage prevention and control planning map in combination with the development trend map of the rodent damage.

Beneficial effect:

1. The present invention evaluates the classification accuracy of interpreted images through the accuracy evaluation confusion matrix table, which is objective and convenient. On the basis of higher classification accuracy, the entire process is simplified, and a result set of a suitable size is provided for subsequent steps ;

2. By comparing the images of the same plot in different time and space, the change area of the target object is marked according to the time flow, so that the target object in the area to be monitored has good continuity and predictability in time and space;

3. Through the development trend map of rodent damage in the area to be monitored, combined with the rodent damage prevention and control methods suitable for different damage levels, and referring to the local terrain, slope position, and mechanization level, a specific rodent damage prevention and control planning map is formulated, which is a scientific basis for rodent damage. Prevention and control work provides data support.

Description of drawings

Fig. 1 is a flow chart of the method for dynamic monitoring of grassland rodent damage based on unmanned aerial vehicle images in a preferred embodiment of the present invention;

Fig. 2 is a schematic diagram of an image interpretation result of another preferred embodiment in the present invention;

FIG. 3 is a flow chart of another preferred embodiment of the method for dynamic monitoring of rodent damage in grassland based on aerial images taken by drones in the present invention.

Detailed ways

In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further elaborated below in conjunction with the accompanying drawings. In describing the present invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", " The orientation or positional relationship indicated by "outside", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, so as to Specific orientation configurations and operations, therefore, are not to be construed as limitations on the invention.

As shown in Figure 1, the present invention provides a method for dynamic monitoring of grassland rodent damage based on unmanned aerial vehicle images, including steps:

Step S1. Use the UAV to obtain aerial images of different time and space in the grassland area to be monitored.

It should be understood that the present invention does not intend to limit the specific brand and model of the UAV, and has no requirements for its flight parameters. Those skilled in the art only need to ensure that the above parameters can realize clear shooting of images of the grassland area to be monitored. It should be noted that due to the large spatial variability of grasslands and the high heterogeneity of the ground, in order to improve the recognizability of UAV aerial images, it is necessary to set up sample plots in the area to be monitored, and set Appropriate scales and markers are used so that aerial images have reference objects for classification and size resolution, and, for the specific conditions of different sites, the setting methods of scales and markers will also be different. In some preferred embodiments, a general setting method for the sample plot is given: place a 50m*50m ruler in the middle of the sample plot, and make eye-catching marks at every 5m intervals of the ruler; The sides and corners of the plots are clearly marked.

Further, the different time and space specifically refers to different time and different space heights. Due to the large seasonal variation of the grassland, only collecting images of the grassland area to be monitored in a single season or within a short period of time has fewer samples for predicting the development trend of rodent damage, thereby improving the accuracy of image prediction. Therefore, in some preferred embodiments, it is preferable to collect images of the grassland area to be monitored in winter and summer. On the other hand, since the flight height of the UAV has a large impact on the image accuracy, a step of adjusting the flight parameters of the UAV according to the accuracy is also set up in the follow-up. In this way, while improving the image accuracy and classification accuracy, it also increases the richness of samples in the sample set. In some other preferred embodiments, the adjustment method of the flight parameters is: according to the obtained aerial image visualization accuracy, and the classification accuracy evaluation in the subsequent step S2, the flight parameters of the drone are adjusted, and the flight parameters include : Parameters such as flight area, route, altitude, speed, camera resolution and other parameters.

Step S2. Perform preprocessing on the collected images, and then perform interpretation to obtain an interpreted image; the interpretation includes classifying target objects to obtain a result set.

Those skilled in the art should know that preprocessing the aerial images of drones is an essential step, but the specific methods of preprocessing are different for different application objects. In the present invention, since the image needs to be input into the support vector machine model in the subsequent steps, and the trend image needs to be further drawn, special processing is required during preprocessing. In some preferred embodiments, the preprocessing includes: sequential screening, dense point cloud generation, digital surface model output, projection transformation and geometric correction, and clipping operations. Wherein, the surface point data collection obtained by the dense point cloud through image analysis is the basis of reverse modeling. In the present invention, when drawing the trend image, the essence is to operate the point cloud; the digital surface model (Digital Surface Model, DSM) refers to the ground elevation model that includes the height of surface buildings, bridges and trees, which can truly express the ground undulations. The setting of the projection conversion and geometric correction steps is because the spatial variability of the grassland is large, and the heterogeneity of the ground is also high, and projection conversion and geometric correction are required to obtain a real image reflecting the grassland topography.

Furthermore, as shown in Figure 2, the interpretation of the image refers to identifying the target from the image, qualitatively and quantitatively extracting the distribution, structure, function and other related information of the target, including the shape, size, shadow, and color of the target. , color, texture, pattern, position and layout information. The specific interpretation method is not the key point of the present invention, so various commonly used interpretation methods can be used. In some preferred embodiments, the target objects include: grassland, bare ground and rat holes. In some other preferred embodiments, since the resolution of UAV aerial images cannot be distinguished when identifying dead leaves and vegetation, only vegetation patches can be distinguished, which may cause the UAV aerial image interpretation The resulting vegetation coverage is higher than the actual coverage of the target area. Therefore, the relevant parameters can be assisted to be corrected by way of field investigation. The specific method is: manually survey the sample plots in the grassland area to be monitored, calculate the correlation between the survey results and the interpretation results, and use the correlation coefficient as the adjustment coefficient to correct the interpretation results.

The present invention also provides a better method for classifying target objects, including:

Build a support vector machine model and train it;

Generate classification accuracy evaluation confusion matrix table according to classification accuracy; described classification accuracy evaluation items include: overall classification accuracy, kappa coefficient, misclassification error, omission error, mapping accuracy and user accuracy;

According to the classification accuracy, it is judged whether the training of the support vector machine model is completed.

It should be understood that the present invention is not intended to limit the specific structure and related parameters of the support vector machine, and those skilled in the art can apply general support vector machine models to the present invention.

Step S3. Comparing the images of different time and space in the result set, marking the changing area of the target object according to the time flow, and drawing a development trend diagram of the rodent infestation.

Wherein, the method of marking the changing area of the target object can be determined on site by those skilled in the art. For example, a manual method can be used to draw the mousehole distribution outline of the drone image at the same time at different times using different lines, and the obtained Use computer software (such as Adobe Photoshop and AutoCAD, etc.) to mark out the changing part of the line representing the outline, and mark and fill in the new or reduced area of rodent damage surrounded by it, and proceed sequentially for other plots, and then according to This draws a graph of the development trend of rodent infestation. It should be understood that when drawing the trend graph manually, it is necessary for those skilled in the art to reasonably predict the development trend of the rodent damage based on the change of the rodent damage in the time flow, combined with work experience and theoretical analysis.

In some preferred embodiments, a kind of specific steps of drawing the development trend graph of rodent damage are provided, including:

Divide the image of the same space-time and the same plot into several small grids, calculate the individual mousehole density in each grid, and calculate the overall average mousehole density in all grids; wherein, the specific size of the small grids is determined by the field The technicians determine on-site according to the accuracy of the aerial images and the size of the area to be detected. A better choice is 50m*50m.

The small grid with the individual mouse hole density higher than the overall average mouse hole density is marked as a high-risk area, otherwise it is marked as a low-risk area;

Draw a graph of the development trend of rodent damage from high-risk areas to low-risk areas under the spatio-temporal conditions;

Repeat the above steps for the images of the same plot in different time and space, and draw the development trend map of rodent damage in this plot;

Repeat the above steps for the images of different plots to obtain the development trend map of rodent damage in the entire area to be monitored.

In some preferred embodiments, the graph of the development trend of the rodent damage includes: a graph of the development direction and trend of the rodent damage and a graph of the development trend of the rodent damage.

Further, in order to better realize the rodent damage development trend graph obtained by using the above steps, as shown in Figure 3, the present invention also provides subsequent steps, including:

Step S4. Determine the control method and labor cost applicable to the rodent damage in the area to be monitored, and draw a rodent damage prevention and control planning map in combination with the development trend map of the rodent damage. It should be understood that the control methods and costs of grasslands with different regions and rodent damage conditions are different. Therefore, when determining the above methods and costs, the statistics and opinions of relevant departments can be used, or obtained by visiting local herdsmen. The rodent pest control planning map includes a distribution map of suitable methods for rodent pest control, a distribution map of suitable human cost input for rodent pest control, a distribution map of suitable material cost input for rodent pest control, a distribution map of suitable financial cost input for rodent pest control, and the like. The above contents can provide data support for the actual rodent control work of the grassland management department, and have certain guiding significance.

Those skilled in the art will further appreciate that the embodiments of the present invention may be realized or implemented by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer readable memory. The methods can be implemented in a computer program using standard programming techniques - including a non-transitory computer-readable storage medium configured with a computer program, where the storage medium so configured causes the computer to operate in a specific and predefined manner - according to the specific Methods and Figures described in the Examples. Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system. However, the programs can be implemented in assembly or machine language, if desired. In any case, the language may be a compiled or interpreted language. Furthermore, the program can be run on an application specific integrated circuit programmed for this purpose. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of functions in the above description. Whether these functions are executed by hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the present invention.

In addition, operations of processes described herein may be performed in any suitable order unless otherwise indicated herein or otherwise clearly contradicted by context. The processes described herein (or variations and/or combinations thereof) can be performed under the control of one or more computer systems configured with executable instructions, and as code that collectively executes on one or more processors (e.g. , executable instructions, one or more computer programs or one or more applications), hardware or a combination thereof. The computer program comprises a plurality of instructions executable by one or more processors.

Further, the method can be implemented in any type of computing platform operably connected to a suitable one, including but not limited to personal computer, minicomputer, main frame, workstation, network or distributed computing environment, stand-alone or integrated computer platform, or communicate with charged particle tools or other imaging devices, etc. Aspects of the invention can be implemented as machine-readable code stored on a non-transitory storage medium or device, whether removable or integrated into a computing platform, such as a hard disk, optically read and/or written storage medium, RAM, ROM, etc., such that they are readable by a programmable computer, when the storage medium or device is read by the computer, can be used to configure and operate the computer to perform the processes described herein. Additionally, the machine-readable code, or portions thereof, may be transmitted over a wired or wireless network. The invention described herein includes these and other various types of non-transitory computer-readable storage media when such media include instructions or programs that implement the steps described above in conjunction with a microprocessor or other data processor. The invention also includes the computer itself when programmed according to the methods and techniques described herein.

Claims (7)

1. A grassland rat damage dynamic monitoring method based on unmanned aerial vehicle aerial images is characterized by comprising the following steps:

s1, acquiring aerial images of different time spaces of a grassland area to be monitored by using an unmanned aerial vehicle;

s2, preprocessing the acquired image, and then interpreting to obtain an interpreted image; the interpretation comprises the steps of classifying the target object to obtain a result set;

s3, comparing the images of different time and space in the result set, marking the change area of the target object according to the time flow, and drawing a rat damage development trend graph;

step S3 further includes:

dividing the images of the same space, time and place into a plurality of small grids, calculating the individual rat hole density in each grid, and calculating the overall average rat hole density in all grids;

marking the small grids with the individual rat hole density higher than the overall average rat hole density as high-risk areas, and otherwise, marking the small grids as low-risk areas;

drawing a mouse damage development trend graph from a high risk area to a low risk area under the space-time condition;

repeating the steps on the images of the same plot in different time and space, and drawing a mouse damage development trend graph of the plot;

and repeating the steps on the images of different land parcels to obtain a rat damage development trend graph of the whole area to be monitored.

2. The method for dynamically monitoring the rodent pests in the grassland based on the aerial image of the unmanned aerial vehicle as claimed in claim 1, wherein the step S2 further comprises: and adjusting the flight parameters of the unmanned aerial vehicle according to the classification precision of the target object.

3. The method for dynamically monitoring the rodent pests in the grassland based on the aerial image of the unmanned aerial vehicle as claimed in claim 1, wherein the preprocessing of the step S2 comprises: screening, dense point cloud generation, digital earth surface model output, projection conversion and geometric correction and cutting are sequentially carried out.

4. The method for dynamically monitoring the rodent damage in the grassland based on the aerial image of the unmanned aerial vehicle as claimed in claim 1, wherein the target object comprises: grass, bare land and rat holes.

5. The method for dynamically monitoring the rodent damage in the grassland based on the aerial image of the unmanned aerial vehicle as claimed in claim 1, wherein the step S2 further comprises the steps of:

constructing a support vector machine model and training;

generating a classification precision evaluation confusion matrix table according to the classification correctness; the classification accuracy evaluation item includes: overall classification accuracy, kappa coefficient, misclassification error, drawing accuracy and user accuracy;

and judging whether the support vector machine model completes training or not according to the classification precision.

6. The method for dynamically monitoring the rodent damage in the grassland based on the aerial image of the unmanned aerial vehicle as claimed in claim 1, wherein the development trend graph of the rodent damage comprises: a trend chart of the development direction of the mouse damage and a trend chart of the development degree of the mouse damage.

7. The dynamic grassland rodent damage monitoring method based on unmanned aerial vehicle aerial images as claimed in claim 1, wherein the dynamic grassland rodent damage development trend monitoring method further comprises:

and S4, determining a control method suitable for the mouse damage of the area to be monitored and the labor cost thereof, and drawing a mouse damage control planning map by combining the development trend map of the mouse damage.

Images