Biological diversity evaluation method for wild ecological environment

Abstract

The invention discloses a biological diversity evaluation method for a wild ecological environment, and relates to the technical field of public services, and the method comprises the steps: dividing a remote sensing image into a plurality of ecological system regions; setting a plurality of sampling survey selection boxes on the remote sensing image; establishing a target function by taking the minimum sum of the distances between any two sampling survey selection boxes as a target; solving the objective function by adopting a genetic algorithm to obtain an optimal selection box; determining a sampling investigation area on the remote sensing image; controlling acquisition equipment to go to a sampling investigation area to acquire a wild organism image; and analyzing the wild organism image to obtain a biodiversity evaluation result. The optimal selection box obtained by the invention not only can cover all ecological systems, but also is small enough in distance between the sampling investigation areas, so that the time spent on moving the acquisition equipment among the plurality of sampling investigation areas is shortened, and the efficiency of evaluation operation is improved.

Description

Technical Field

[0001] The present application relates to the field of public service technology, and in particular to a method for assessing biodiversity in a wild ecological environment. Background Art

[0002] Wildlife is an important part of the Earth's ecosystem. Wild life includes wild plants, wild animals and wild microorganisms. The living environment of these organisms is not affected by humans, and their distribution area and number are huge. Assessing the diversity of wildlife is an effective means to understand its current status and can provide important information for ecological and environmental governance.

[0003] The traditional biodiversity assessment process relies heavily on manual work, requiring operators to go to the area to be assessed and use survey methods such as sample line method, sample point method and direct counting method to count the number of various types of organisms. This assessment method is not suitable for large areas, and there is also serious subjectivity in the selection of survey points, which leads to limited application. In order to meet the assessment needs of large areas and improve the accuracy of assessment results, assessment methods based on information technology have been proposed. For example, CN114399415A proposes a biodiversity survey and assessment system, which conducts ecological analysis and distribution simulation on the area to be investigated, and then selects the area that actually needs sampling survey, and then collects images in the sampling survey area. The number of organisms can be obtained by analyzing and processing the images.

[0004] Although the above method does not rely on manual work and is applicable to large areas, and has greatly improved the accuracy of the evaluation results, the operating efficiency of the shooting system is not considered when selecting the sampling survey area. If the selected sampling survey areas are distributed in multiple distant locations in a large area, the shooting system will need to spend a lot of time to collect images, resulting in an adverse effect on the efficiency of the entire evaluation process. Summary of the invention

[0005] The embodiment of the present application provides a method for assessing biodiversity in a wild ecological environment, which is used to solve the problem in the prior art that the operating efficiency of the shooting system is not considered when selecting a sampling survey area.

[0006] The present application embodiment provides a method for assessing biodiversity of a wild ecological environment, including:

[0007] Obtain remote sensing images of the assessment area;

[0008] Divide remote sensing images into multiple ecosystem regions;

[0009] A plurality of sampling survey selection boxes are set on the remote sensing image, and each sampling survey selection box covers a part of at least one ecosystem area;

[0010] Determine the distance between any two sampling survey selection boxes and establish an objective function with the minimum sum of distances as the goal;

[0011] Genetic algorithm is used to solve the objective function and obtain the best selection box. In the solving process, the number and area of ​​the sampling survey selection box are subject to the preset number constraint conditions, area constraint conditions, regional category constraint conditions and proportion constraint conditions.

[0012] Determine the sampling survey area corresponding to the best selection frame on the remote sensing image;

[0013] Control the collection equipment corresponding to the assessment area to go to the sampling survey area, and use the collection equipment to collect images of wildlife in the sampling survey area;

[0014] Analyze wildlife images to generate biodiversity assessments.

[0015] The biodiversity assessment method of a wild ecological environment in this application has the following advantages:

[0016] Under the constraints including regional category constraints, a genetic algorithm is used to solve the objective function of minimizing the sum of distances between multiple sampling survey selection boxes. The optimal selection box obtained can not only cover all ecosystems, but also make the distances between the corresponding sampling survey areas small enough, which greatly reduces the time spent on moving the collection equipment between multiple sampling survey areas and improves the efficiency of the assessment work. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 A flowchart of a method for assessing biodiversity in a wild ecological environment provided in an embodiment of the present application.

[0019] Figure 2 A schematic diagram of the distribution of the sampling survey areas in the evaluation area provided in the embodiment of the present application. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0021] Figure 1 A flowchart of a method for assessing biodiversity in a wild ecological environment provided in an embodiment of the present application. The present application embodiment provides a method for assessing biodiversity in a wild ecological environment, including:

[0022] S100, obtaining a remote sensing image of an assessment area.

[0023] Exemplarily, the assessment area is a geographical area where biodiversity assessment in a wild ecological environment is required. When acquiring remote sensing images, the most recent remote sensing images acquired by satellites or aircraft may be intercepted to obtain remote sensing images of the assessment area.

[0024] Specifically, before starting the assessment operation, the operator needs to determine the location of the assessment area in advance, and the assessment area can be determined by inputting the longitude and latitude of the edge points of the assessment area. After inputting the longitude and latitude of the edge, the computer can convert these longitude and latitude into multiple coordinate points on the map, and connect these coordinate points in sequence to obtain the assessment area in the form of a closed area.

[0025] S110, divide the remote sensing image into multiple ecosystem regions.

[0026] For example, Figure 2 As shown in the figure, the assessment area is a rectangular area outside. The assessment area contains multiple ecosystems, and these ecosystems correspond to multiple ecosystem areas on the remote sensing image. When dividing the ecosystem area, the remote sensing image can be divided using a trained ecological classification model, and the ecological classification model can be established based on a convolutional neural network. The use of an ecological classification model can automatically process remote sensing images without human intervention. During the classification process, the ecological classification model will determine the probability that each pixel in the remote sensing image belongs to an ecosystem. If the probability exceeds the set probability threshold, the ecosystem with the highest probability is determined as the ecosystem to which the pixel belongs. After obtaining the ecosystem to which each pixel belongs, the pixels belonging to the same ecosystem will be connected into regions to obtain the ecosystem area.

[0027] Since the area where the ecosystem is located changes very little, after obtaining the remote sensing image, it can also be divided into multiple ecosystem areas by manual labeling. Although this division method requires human intervention, since the area where the ecosystem is located will only undergo major changes in months or even years, it takes a long time for human intervention, and the human occupation is almost negligible.

[0028] S120, setting a plurality of sampling survey selection boxes on the remote sensing image, each sampling survey selection box covering a portion of at least one ecosystem area.

[0029] For example, in Figure 2 In the figure, the dotted rectangular boxes are sampling survey selection boxes. The purpose of these sampling survey selection boxes is to determine the best sampling survey areas so that the collection equipment can achieve the highest possible efficiency when operating in these sampling survey areas.

[0030] S130, determining the distance between any two sampling survey selection boxes, and establishing an objective function with the minimum sum of the distances as the goal.

[0031] Exemplarily, the objective function is expressed as:

[0032]

[0033] Where F represents the objective function, i and j represent the labels of the sampling survey selection boxes, N is the number of sampling survey selection boxes, and d ij represents the distance between the i-th sampling survey selection box and the j-th sampling survey selection box, i≠j and d ij and d ji Only one is involved in the calculation.

[0034] When calculating the distance between the sampling survey selection boxes, the distance is the minimum value of the distance between any points on the two sampling survey selection boxes. That is, select a point from each of the two sampling survey selection boxes, and then calculate the distance between the two points. After the distances of all points are calculated, select the minimum distance as the distance between the two sampling survey selection boxes.

[0035] S140, using a genetic algorithm to solve the objective function and obtain the best selection box. During the solving process, the number and area of ​​the sampling survey selection boxes are subject to preset quantity constraints, area constraints, area category constraints and proportion constraints.

[0036] Exemplarily, the quantity constraint is expressed as:

[0037] N≤N max

[0038] The area constraint is expressed as:

[0039]

[0040] The regional category constraints are expressed as:

[0041]

[0042] The proportional constraint is expressed as:

[0043]

[0044] Among them, N max Indicates the preset number of sampling survey selection boxes, s i represents the area of ​​the i-th sampling survey selection box, S max Indicates the preset value of the total area of ​​multiple sampling survey selection boxes, n iu Indicates whether the i-th sampling survey selection box covers the u-th ecosystem area. If so, then n iu The value is 1, otherwise it is 0, U is the number of ecosystem areas, s iu represents the area of ​​the uth ecosystem region covered by the i-th sampling survey selection box, s u represents the area of ​​the u-th ecosystem region, and a represents the preset ratio.

[0045] The quantity constraint limits the maximum number of sampling survey selection boxes, which can be determined according to the number of acquisition devices corresponding to the assessment area. For example, when there are three acquisition devices that can be used at the same time, the maximum number of sampling survey selection boxes can be set to three. However, there is no absolute correspondence between the number of acquisition devices and the number of sampling survey selection boxes. For example, when there is only one acquisition device that can be used, the sampling survey selection box cannot be set to one, otherwise it is very likely that the best selection box cannot be solved. The area constraint limits the size of the sampling survey selection box. Usually, the total area preset value can be set to a certain proportion of the area corresponding to the remote sensing image, that is, the area of ​​the assessment area, such as 30% or 25%. This area ratio setting can ensure that the sampling survey selection box will not be too large, causing the assessment to take too much time, nor will it be too small, causing inaccurate assessment results. The regional category constraint enables the sampling survey selection box to cover all types of ecosystem areas, avoiding distortion of assessment results due to missing ecosystem areas. The proportion constraint ensures that the sampling survey selection box covers enough area when covering the ecosystem area, avoiding inaccurate assessment results due to too small coverage area. Since remote sensing images may only cover a part of an ecosystem, when calculating the area of ​​an ecosystem region, only the area belonging to the ecosystem region in the remote sensing image can be calculated without considering the area outside the remote sensing image.

[0046] The genetic algorithm solution process mainly includes five steps: initialization, determination of fitness value, selection, crossover and mutation. During initialization, it is necessary to randomly generate multiple groups of initial sampling survey selection boxes that meet the constraints according to the set initialization parameters. Each group of initial sampling survey selection boxes will be regarded as a population, and each initial sampling survey selection box in the population will be regarded as a chromosome, and each chromosome will be represented in a digitally encoded manner. When determining the fitness value, the fitness value of each population determined based on the objective function will be calculated, and then when selecting, all populations will be arranged in order from high to low according to the fitness value, and multiple populations with the highest fitness values ​​will be selected to enter the crossover step. During crossover, the chromosomes in the two populations are exchanged with a certain probability. After the exchange, it is necessary to determine whether the newly generated population meets all the constraints. If it meets the constraints, it will be retained, otherwise it will be discarded. The population that has completed the crossover will continue to enter the mutation step. During mutation, the chromosomes in each population will be mutated according to a certain probability, and then after obtaining the new population, it is also necessary to judge the constraints, and only the new population that meets all the constraints will be retained. After the mutation process is completed, the fitness value of each new population will be calculated, and the population with the highest fitness value will be recorded. Then the next round of iteration will be carried out. After multiple iterations, the fitness value of the new population remains basically unchanged or the number of iterations is reached. The population with the highest fitness value will be output as the best selection box. The best selection box includes multiple selection boxes set on the remote sensing image. The number and area of ​​these selection boxes meet the requirements of the constraint conditions and can also achieve the shortest sum of distances.

[0047] S150, determining a sampling survey area corresponding to the best selection frame on the remote sensing image.

[0048] Exemplarily, the best selection box is an area with a specific shape, such as a rectangle, set on the remote sensing image, and the area enclosed by the points on the best selection box is the sampling survey area on the remote sensing image. The location of the sampling survey area can be represented by the latitude and longitude of each point on the best selection box.

[0049] S160, controlling the collection device corresponding to the evaluation area to go to the sampling survey area, and having the collection device collect images of wildlife in the sampling survey area.

[0050] For example, since wild ecological environments are usually in primitive environments far away from towns, there are no conditions for deploying a large number of collection equipment. In a large area, multiple representative assessment areas can be delineated, and these assessment areas with similar distances can share the same equipment station, which is equipped with at least one collection device. Each collection device corresponds to at least one assessment area, and the collection equipment includes drones, unmanned vehicles, and unmanned ships.

[0051] Specifically, the equipment station can be set up in a location close to the assessment area and with vehicle access to facilitate the construction and maintenance of the equipment station. When setting up the assessment area, the ecosystems contained in the assessment area are already known, so it is necessary to deploy collection equipment that matches all ecosystems in the equipment station.

[0052] Furthermore, ecosystems include forest ecosystems, grassland ecosystems, freshwater ecosystems, marine ecosystems, wetland ecosystems and urban ecosystems. When the assessment area includes grassland ecosystems and wetland ecosystems, wildlife images are collected by drones; when the assessment area includes forest ecosystems and urban ecosystems, wildlife images are collected by drones and unmanned vehicles; when the assessment area includes freshwater ecosystems and marine ecosystems, wildlife images are collected by drones and unmanned ships.

[0053] Specifically, grassland ecosystems and wetland ecosystems do not have ground and water surface equipment, that is, the conditions for unmanned vehicles and unmanned boats to pass through, so they can only collect ground and aerial images through drones. The situation of forest ecosystems and urban ecosystems is more complicated, including both aerial and ground creatures, and drones cannot collect images of ground creatures very well, so drones and unmanned vehicles are needed to cooperate, with drones collecting images of aerial creatures and unmanned vehicles collecting images of ground creatures. Freshwater ecosystems and marine ecosystems require the cooperation of drones and unmanned boats, with drones collecting images of aerial and surface creatures, and unmanned boats collecting images of underwater creatures.

[0054] Furthermore, in forest ecosystems and urban ecosystems, the unmanned vehicle determines the ground moving path based on the ground environmental information, and the drone is synchronously located directly above the unmanned vehicle when the unmanned vehicle moves along the ground moving path; in freshwater ecosystems and marine ecosystems, the unmanned boat determines the water surface moving path based on the water surface environmental information, and the drone is synchronously located directly above the unmanned boat when the unmanned boat moves along the water surface moving path.

[0055] Specifically, in a wild ecological environment, it is very likely that the conditions for the unmanned vehicle to reach any location in the forest or city are not met, so there is a situation where the unmanned vehicle can only collect images of a part of the forest or urban ecosystem area. At this time, the unmanned vehicle will calculate the area ratio of the area where the image has been collected to the entire forest or urban ecosystem area where the image needs to be collected. If the set ratio has been reached, such as 70% or higher, it is considered that the drone has also collected enough images, so the drone will no longer continue to collect images of the remaining area. If the set ratio is not reached, when the unmanned vehicle cannot move forward, the drone will go to the remaining area alone to collect images, so that all aerial creatures at all locations in the forest or urban ecosystem area are collected. The same situation also exists in freshwater ecosystems and marine ecosystems, so when the unmanned ship cannot move forward, it will calculate the area ratio of the area where the image has been collected to the entire area where the image needs to be collected. If the set ratio has been reached, the drone will no longer continue to collect images of the remaining area. If it is not reached, the drone will go to the remaining area alone to collect images.

[0056] When a drone and an unmanned vehicle or a drone and an unmanned ship cooperate to collect images, the images collected by the drone and the unmanned vehicle / unmanned ship may contain the same organism. In order to avoid increased difficulty in organism identification due to different positions, the drone will be controlled to always fly directly above the unmanned vehicle or unmanned ship. When the same organism exists in the images collected at the same time, high-accuracy organism matching can be performed, and the number of the same organism can be corrected to avoid the same organism being counted multiple times.

[0057] Furthermore, since different collection devices have different moving speeds in the same ecosystem area, the moving speed of each collection device can also be taken into consideration when solving the objective function using the genetic algorithm. If there are multiple possibilities for the optimal selection box obtained by the genetic algorithm, for example Figure 2 The sampling survey selection box in the middle is on the line connecting the two sampling survey selection boxes on the left and right. At this time, no matter which sampling survey selection box in the middle is close to the sampling survey selection box on the side, the sum of the distances remains unchanged, but there are multiple possible positions. At this time, the total time required for the sampling device to perform the task at each position of the sampling survey selection box can be further calculated, and the best selection box with the shortest time can be selected.

[0058] S170, analyzing the wildlife image to obtain a biodiversity assessment result.

[0059] Exemplarily, when any organism is included in a wildlife image, the species of the organism can be identified using a biological recognition model to obtain the species and quantity of each organism in the wildlife image. After the quantity correction of the same organism in S160, the data on the species and quantity of organisms in all wildlife images are summarized to obtain the species and total quantity of all organisms in a task. Based on these data, a variety of different biodiversity indices can be calculated, including the Shannon-Wiener diversity index, the Margalef richness index, the Simpson index, the Pielou evenness index and the comprehensive biodiversity index. Using these biodiversity indices, an assessment report in a preset format can be formed, and the assessment report can be considered as a biodiversity assessment result.

[0060] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0061] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for assessing biodiversity in a wild ecological environment, characterized in that: include: Obtain remote sensing images of the assessment area; dividing the remote sensing image into a plurality of ecosystem regions; Setting a plurality of sampling survey selection boxes on the remote sensing image, each of the sampling survey selection boxes covering a portion of at least one ecosystem area; Determine the distance between any two of the sampling survey selection boxes, and establish an objective function with the goal of minimizing the sum of the distances; A genetic algorithm is used to solve the objective function to obtain an optimal selection box. During the solving process, the number and area of ​​the sampling survey selection boxes are subject to preset quantity constraints, area constraints, area category constraints and proportion constraints. Determine a sampling survey area corresponding to the optimal selection frame on the remote sensing image; Controlling a collection device corresponding to the assessment area to go to the sampling survey area, and having the collection device collect images of wildlife in the sampling survey area; The wildlife images are analyzed to obtain biodiversity assessment results.

2. The method for assessing biodiversity of a wild ecological environment according to claim 1, characterized in that: The remote sensing image is divided into a plurality of ecosystem regions using a trained ecological classification model.

3. The method for assessing biodiversity of a wild ecological environment according to claim 2, characterized in that: The ecological classification model is established based on a convolutional neural network.

4. The method for assessing biodiversity of a wild ecological environment according to claim 1, characterized in that: The objective function is expressed as: Wherein, F represents the objective function, i and j represent the labels of the sampling survey selection boxes, N is the number of the sampling survey selection boxes, and d ij represents the distance between the i-th sampling survey selection box and the j-th sampling survey selection box, i≠j and d ij and d ji Only one is involved in the calculation.

5. The method for assessing biodiversity of a wild ecological environment according to claim 4, characterized in that: The distance between the two sampling survey selection boxes is the minimum value of the distance between any points on the two sampling survey selection boxes.

6. The method for assessing biodiversity of a wild ecological environment according to claim 4, characterized in that: The quantity constraint condition is expressed as: N≤N max The area constraint is expressed as: The area category constraint is expressed as: The proportional constraint is expressed as: Among them, N max Indicates the preset number of sampling survey selection boxes, s i represents the area of ​​the i-th sampling survey selection box, S max Indicates the total area preset value of multiple sampling survey selection boxes, n iu Indicates whether the i-th sampling survey selection box covers the u-th ecosystem area. If yes, then n iu The value is 1, otherwise it is 0, U is the number of ecosystem areas, s iu represents the area of ​​the uth ecosystem region covered by the i-th sampling survey selection box, s u represents the area of ​​the u-th ecosystem region, and a represents a preset ratio.

7. The method for assessing biodiversity of a wild ecological environment according to claim 1, characterized in that: A device station is set in each of the multiple evaluation areas, and at least one of the collection devices is set in the device station. Each of the collection devices corresponds to at least one of the evaluation areas, and the collection devices include unmanned aerial vehicles, unmanned vehicles and unmanned ships.

8. The method for assessing biodiversity of a wild ecological environment according to claim 7, characterized in that: Ecosystems include forest ecosystems, grassland ecosystems, freshwater ecosystems, marine ecosystems, wetland ecosystems and urban ecosystems. When the assessment area includes grassland ecosystems and wetland ecosystems, the wildlife images are collected by the drone; when the assessment area includes forest ecosystems and urban ecosystems, the wildlife images are collected by the drone and the unmanned vehicle; when the assessment area includes freshwater ecosystems and marine ecosystems, the wildlife images are collected by the drone and the unmanned boat.

9. A method for assessing biodiversity of a wild ecological environment according to claim 8, characterized in that: In the forest ecosystem and the urban ecosystem, the unmanned vehicle determines a ground moving path according to ground environment information, and when the unmanned vehicle moves along the ground moving path, the drone is synchronously located directly above the unmanned vehicle; In freshwater ecosystems and marine ecosystems, the unmanned ship determines a surface movement path based on water surface environment information, and when the unmanned ship moves along the surface movement path, the drone is synchronously located directly above the unmanned ship.

Images