Biomass inversion method and system

Abstract

The invention provides a biomass inversion method and system, and the method comprises the steps: obtaining the target remote sensing data of a target biomass inversion region, determining the input characteristics corresponding to each target pixel in the target biomass inversion region according to the target remote sensing data, and obtaining the input characteristics corresponding to each target pixel in the target biomass inversion region according to the input characteristics corresponding to each target pixel; target biomass values corresponding to the target pixels are determined, the input features of any target pixel in the target pixels comprise the features of any target pixel and the features of target adjacent pixels adjacent to any target pixel, and the target biomass values corresponding to the target pixels are determined according to the target biomass values corresponding to the target pixels. And generating a biomass map corresponding to the target biomass inversion region. The content of the input feature of each pixel can be richer, so that the accuracy and effectiveness of the generated biomass map can be improved.

Description

Technical Field

[0001] This specification relates to the field of artificial intelligence technology, and particularly to a biomass inversion method and system. Background Art

[0002] Aboveground biomass (AGB), simply referred to as biomass, is an important indicator for assessing the carbon storage and forest quality of forest ecosystems. It plays a dominant role in the carbon cycle between the land and the atmosphere, thus affecting climate change.

[0003] Therefore, quickly and accurately obtaining the spatial distribution data of forest aboveground biomass can not only help improve resource assessment, carbon accounting, and carbon models, but also the spatio-temporal dynamic monitoring of forest aboveground biomass is crucial for research on terrestrial ecosystem functions, carbon budgets, global climate change, etc.

[0004] However, most traditional biomass detections rely on on-site actual measurements and visual inspections at observation points, which are time-consuming, laborious, and difficult to work. The traditional methods are increasingly unable to meet the requirements. Therefore, there is an urgent need for a method that can meet the needs of large-scale, rapid, and accurate biomass inversion.

[0005] It should be noted that the content of the above related technologies is only the information known to the inventor personally, and does not represent that the above information has entered the public domain before the filing date of this specification, nor does it represent that it can become the prior art of this specification. Summary of the Invention

[0006] This specification provides a biomass inversion method and system to avoid at least one of the above technical problems.

[0007] In a first aspect, this specification provides a biomass inversion method, including:

[0008] Obtaining target remote sensing data of a target biomass inversion area;

[0009] Determining input features corresponding to each target pixel in the target biomass inversion area according to the target remote sensing data, and determining the target biomass value corresponding to each target pixel according to the input features corresponding to each target pixel, wherein the input features of any target pixel among the target pixels include the features of the any target pixel and the features of target adjacent pixels adjacent to the any target pixel; and

[0010] Generating a biomass map corresponding to the target biomass inversion area according to the target biomass values corresponding to each target pixel.

[0011] In a second aspect, this specification provides a biomass inversion system, including:

[0012] At least one storage medium storing at least one instruction set for biomass inversion;

[0013] At least one processor communicatively connected to the at least one storage medium, wherein when the at least one processor runs, it reads the at least one instruction set and executes the method described in the first aspect according to the instructions of the at least one instruction set.

[0014] In a third aspect, this specification provides a computer-readable non-transitory storage medium, wherein the computer-readable non-transitory storage medium stores at least one instruction set, and the at least one instruction set is executed by at least one processor to implement the method described in the first aspect.

[0015] As can be seen from the above technical solutions, the biomass inversion method and system provided in this specification determine the input features corresponding to each pixel in the remote sensing data, and the input feature of each pixel includes, in addition to its own features, the features of adjacent pixels, so that the content of the input feature of each pixel is richer. Therefore, when determining the biomass value based on the input features, the accuracy and reliability of the determined biomass value can be improved. Furthermore, the accuracy and effectiveness of the biomass map generated based on the biomass value can be improved.

[0016] Other functions of the biomass inversion method and system provided in this specification will be partially listed in the following description. The creative aspects of the biomass inversion method and system provided in this specification can be fully explained through practice or use of the methods, devices, and combinations described in the detailed examples below. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] To more clearly illustrate the technical solutions in the embodiments of this specification, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of this specification. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a schematic diagram of the application scenario of the biomass inversion method provided in the embodiments of this specification;

[0019] Figure 2 It is a schematic diagram of the structure of the biomass inversion system provided in the embodiments of this specification;

[0020] Figure 3 It is a schematic flowchart of the biomass inversion method provided in the embodiments of this specification;

[0021] Figure 4 It is a schematic diagram of the principle of the biomass inversion method provided in the embodiments of this specification;

[0022] Figure 5 This is a schematic flowchart for training a biomass value prediction model provided in this specification. Detailed implementation manners

[0023] Here, exemplary embodiments will be described in detail, and the examples are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation manners described in the following exemplary embodiments do not represent all implementation manners consistent with this specification. On the contrary, they are merely examples of devices and methods consistent with some aspects of this specification as detailed in the appended claims.

[0024] It should be understood that the terms "include" and "have" and any variations thereof in the embodiments of this specification are intended to cover but not exclude inclusion. For example, a product or device including a series of components does not necessarily have to be limited to those components clearly listed, but may include other components not clearly listed or inherent to these products or devices.

[0025] The term "and / or" in the embodiments of this specification describes the association relationship of associated objects and indicates that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally represents an "or" relationship between the associated objects before and after.

[0026] The term "plurality" in the embodiments of this specification means two or more, and other quantifiers are similar thereto.

[0027] The terms "first", "second", "third", "target", "sample", etc. in this specification are used to distinguish similar or like objects or entities, and do not necessarily mean to limit a specific order or sequence, unless otherwise indicated (Unless otherwise indicated). It should be understood that such terms can be interchanged under appropriate circumstances, for example, it is possible to implement in an order other than those given in the illustrations or descriptions of the embodiments of this specification.

[0028] The term "unit / module" used in this specification refers to any known or later-developed hardware, software, firmware, artificial intelligence, fuzzy logic, or a combination of hardware or / and software code that can perform functions related to that element.

[0029] To facilitate the reader's understanding of this specification, the application scenarios of this specification are now introduced.

[0030] The technical solution provided in this specification is applicable to scenarios where it is necessary to determine the biomass of a certain area. Exemplarily, the technical solution provided in this specification can be applied to large-scale biomass estimation in forest resource surveys, providing data support for fields such as local afforestation and carbon trading.

[0031] Among them, forest resource survey can be understood as a forestry survey targeting forest land, forest trees, and animals, plants, and their environmental conditions growing within the forest area. Biomass can be understood as the total amount of organic matter existing per unit area at a certain moment, usually expressed in dry weight. Relatively speaking, it reflects the productivity and carbon storage of the ecosystem.

[0032] For example, in a certain forest area, by implementing the technical solution provided in this specification, the value of the biomass of this forest area (abbreviated as biomass value) can be determined, and then the biomass map of this forest area can be determined.

[0033] Figure 1 It is a schematic diagram of the application scenario of the biomass inversion method of the embodiments of this specification. Among them, the biomass inversion method of this specification can be applied to Figure 1 the scenario 100 shown. As Figure 1 shown, the scenario 100 may include a target user 101, a client 102, a server 103, and a network 104.

[0034] The target user 101 may be a user who triggers the biomass inversion of the target biomass inversion area. For example, the target user 101 may perform a target operation on the client 102 to trigger the biomass inversion of the target biomass inversion area. In some embodiments, the target user 101 may upload the target remote sensing data of the target biomass inversion area on the client 102 to trigger the biomass inversion of the target biomass inversion area. In some embodiments, the target user 101 may input indication information on the client 102 to indicate the target remote sensing data to be subjected to biomass inversion. The above indication information may be the identifier of the target remote sensing data, etc.

[0035] Among them, biomass inversion refers to the process of extracting vegetation biomass information from remote sensing images through remote sensing technology or other indirect measurement means, combined with mathematical models or algorithms. Simply put, it is to estimate the total biomass of surface vegetation (including the mass of above-ground and underground parts) using remote sensing data, so as to provide important basis for ecosystem assessment, carbon storage monitoring, and resource management.

[0036] Remote sensing data can be understood as data obtained based on remote sensing technology. Exemplarily, remote sensing data includes data obtained by non-contact sensors (such as devices on platforms like satellites, airplanes, drones, etc.) to detect and acquire information about corresponding areas from a distance. Such data can contain information in multiple aspects, including surface features, atmospheric conditions, ocean dynamics, and the impacts of human activities. Remote sensing data includes remote sensing images, which can be understood as ground images obtained from satellite or aerial platforms and contain information about ground objects and the environment, etc.

[0037] Correspondingly, target remote sensing data can be understood as data obtained by non-contact sensors to detect and acquire information about the target biomass inversion area from a distance.

[0038] The client 102 can be an electronic device that provides an interaction function to the target user 101. For example, the client 102 can provide an interaction interface to the target user 101, and the target user 101 can perform interaction operations on the interaction page. In some embodiments, in response to detecting an operation of the target user 101 triggering biomass inversion, the client 102 executes the biomass inversion method described in this specification. At this time, the client 102 may store data or instructions for executing the biomass inversion method described in this specification and can execute or be used to execute the data or instructions. In some embodiments, the client 102 may include a hardware device with data information processing capabilities and the necessary programs to drive the hardware device to work, so as to execute the biomass inversion method described in this specification.

[0039] In some embodiments, the client 102 may include a mobile device, a tablet computer, a laptop computer, an in-vehicle device of a motor vehicle, or the like, or any combination thereof. In some embodiments, the mobile device may include a smart home device, a smart mobile device, a virtual reality device, an augmented reality device, or the like, or any combination thereof. In some embodiments, the smart home device may include a smart TV, a desktop computer, etc., or any combination. In some embodiments, the smart mobile device may include a smart phone, a personal digital assistant, a gaming device, a navigation device, etc., or any combination thereof. In some embodiments, the in-vehicle device in a motor vehicle may include an in-vehicle computer, an in-vehicle TV, etc. In some embodiments, the client 102 may include a collection device for collecting target remote sensing data.

[0040] In some embodiments, the client 102 may be installed with one or more applications (APPs). The APPs can provide the target user 101 with the ability to interact with the outside world through the network 104 and an interface. The APPs include, but are not limited to: web browser APPs, search APPs, chat APPs, shopping APPs, video APPs, financial management APPs, instant messaging tools, email clients, social platform software, and so on. In some embodiments, a target APP may be installed on the client 102. The target APP can collect target remote sensing data for the client 102.

[0041] As Figure 1 shown, the client 102 may be communicatively connected to the server 103. Among them, the server 103 may be communicatively connected to one client 102 or may be communicatively connected to multiple clients 102. In some embodiments, the client 102 may interact with the server 103 through the network 104 to receive or send messages, etc. For example, the client 102 may interact with the server 103 through the network 104 to send target remote sensing data to the server 103.

[0042] The server 103 may be a server that provides various services. For example, the server 103 may be a cloud server or a local server. The server 103 may be communicatively connected to one client 102 and receive the data sent by the client 102, or may be communicatively connected to multiple clients 102 and receive the data sent by each client 102 respectively.

[0043] In some embodiments, the biomass inversion method described in this specification may be executed on the server 103. At this time, the server 103 may store the data or instructions for executing the biomass inversion method described in this specification and may execute or be used to execute the data or instructions. The server 103 may include a hardware device with data information processing capabilities and the necessary programs for driving the hardware device to work.

[0044] The network 104 is a medium for providing a communication connection between the client 102 and the server 103. The network 104 can facilitate the exchange of information or data. As Figure 1 shown, the client 102 and the server 103 may be respectively connected to the network 104 and transmit information or data to each other through the network 104.

[0045] In some embodiments, network 104 can be any type of wired or wireless network, or a combination thereof. For example, network 104 can include a cable network, a wired network, an optical fiber network, a telecommunication network, an intranet, the Internet, a Local Area Network (LAN), a Wide Area Network (WAN), a Wireless Local Area Network (WLAN), a Metropolitan Area Network (MAN), a Public Switched Telephone Network (PSTN), a Bluetooth networkTM, a short-range wireless network (ZigBeeTM), a Near Field Communication (NFC) network, or a similar network.

[0046] In some embodiments, network 104 can include one or more network access points. For example, network 104 can include a wired or wireless network access point, such as a base station or an Internet exchange point, through which one or more components of client 102 and server 103 can be connected to network 104 to exchange data or information.

[0047] It is worth noting that Figure 1 the numbers of client 102, server 103, and network 104 in

[0048] are merely illustrative. According to the implementation requirements, there can be any number of client 102, server 103, and network 104. And the biomass inversion method provided in this specification can be executed entirely on client 102, entirely on server 103, or partially on client 102 and partially on server 103. Figure 1 That is to say, Figure 1 and the above description for

[0049] Figure 2The figure shows a hardware structure diagram of a biomass inversion system 200 provided according to an embodiment of this specification. The biomass inversion system 200 can execute the biomass inversion method described in this specification. The biomass inversion method is introduced in other parts of this specification. When the biomass inversion method is executed on the client 102, the biomass inversion system 200 can be the client 102. When the biomass inversion method is executed on the server 103, the biomass inversion system 200 can be the server 103. When the biomass inversion method is partially executed on the client 102 and partially executed on the server 103, the biomass inversion system 200 can be a system including the client 102 and the server 103.

[0050] As Figure 2 shown, the biomass inversion system 200 may include at least one storage medium 203 and at least one processor 202. In some embodiments, the biomass inversion system 200 may further include a communication port 204 and an internal communication bus 201. The biomass inversion system 200 may further include I / O components 205.

[0051] The internal communication bus 201 can connect different system components. For example, the internal communication bus 201 can connect the storage medium 203, the processor 202, the communication port 204, and the I / O components 205.

[0052] The I / O components 205 support input / output between the biomass inversion system 200 and other components.

[0053] The communication port 204 is used for data communication between the biomass inversion system 200 and the outside world. For example, the communication port 204 can be used for data communication between the biomass inversion system 200 and the network 104. The communication port 204 can be a wired communication port or a wireless communication port.

[0054] The storage medium 203 may include a data storage device. The data storage device can be a non-temporary storage medium or a temporary storage medium. For example, the data storage device can include one or more of a magnetic disk 2031, a read-only storage medium (ROM) 2032, or a random access storage medium (RAM) 2033. The storage medium 203 further includes at least one instruction set stored in the data storage device. The instruction set includes computer program code, and the computer program code can include programs, routines, objects, components, data structures, processes, modules, etc. for executing the biomass inversion method provided in this specification.

[0055] At least one processor 202 can be communicatively connected to at least one storage medium 203. The at least one processor 202 is configured to execute the above-mentioned at least one instruction set. When the biomass inversion system 200 is running, the at least one processor 202 reads the at least one instruction set and, according to the instructions of the at least one instruction set, executes the biomass inversion method provided in this specification. The processor 202 can execute all the steps included in the biomass inversion method. The processor 202 can be in the form of one or more processors. In some embodiments, the processor 202 can include one or more hardware processors, such as a microcontroller, a microprocessor, a reduced instruction set computer (RISC), an application-specific integrated circuit (ASIC), an application-specific instruction set processor (ASIP), a central processing unit (CPU), a graphics processing unit (GPU), a physics processing unit (PPU), a microcontroller unit, a digital signal processor (DSP), a field-programmable gate array (FPGA), an advanced RISC machine (ARM), a programmable logic device (PLD), any circuit or processor capable of executing one or more functions, etc., or any combination thereof.

[0056] For illustrative purposes only, only one processor 202 is shown in the biomass inversion system 200 in the drawings. However, it should be noted that the biomass inversion system 200 in this specification can also include multiple processors. Therefore, the operations and / or method steps disclosed in this specification can be executed by one processor or jointly executed by multiple processors. For example, if it is described in this specification that the processor 202 of the biomass inversion system 200 executes step A and step B, it should be understood that step A and step B can also be jointly or separately executed by two different processors 202 (for example, the first processor executes step A, the second processor executes step B, or the first and second processors jointly execute steps A and B).

[0057] Please refer to Figure 3 , Figure 3 which is a schematic flow diagram of the biomass inversion method provided in the embodiments of this specification. As Figure 3 shown, the method includes the following S301 to S303:

[0058] S301: Obtain the target remote sensing data of the target biomass inversion area.

[0059] Combined with the above description of the application scenario, the target biomass inversion area can be understood as the area where biomass inversion needs to be carried out, such as a certain forest area, etc.

[0060] Correspondingly, the target remote sensing data can be understood as the remote sensing data obtained by the biomass inversion system regarding the target biomass inversion area.

[0061] The following examples can be used to obtain the target remote sensing data:

[0062] In one example, the biomass inversion system can be connected to a data acquisition device and receive the target remote sensing data collected and sent by the data acquisition device.

[0063] In another example, the biomass inversion system can provide a tool for loading data, and the target user can transmit the target remote sensing data to the biomass inversion system through this data loading tool.

[0064] Among them, the data loading tool can be an interface for connecting to an external device, such as an interface for connecting to other storage devices, and the target remote sensing data transmitted by the external device can be obtained through this interface; the data loading tool can also be a display device. For example, the biomass inversion system can output an interface for the data loading function on the display device, and the target user can import the target remote sensing data into the biomass inversion system through this interface.

[0065] In some embodiments, the target remote sensing data is multi-source remote sensing data. The multi-source remote sensing data includes observation data obtained from different types of sensors, different platforms (such as satellites, airplanes, drones, etc.), and different observation conditions. By integrating data from multiple sources, various characteristics and dynamic changes of various organisms in the target biomass inversion area can be understood and analyzed more comprehensively and accurately.

[0066] Exemplarily, when the target remote sensing data is multi-source remote sensing data, the target remote sensing data may include hyperspectral data, multispectral data, and long strip synthetic aperture radar (L-SAR) remote sensing image data.

[0067] S302: Determine the input features corresponding to each target pixel in the target biomass inversion area according to the target remote sensing data, and determine the target biomass value corresponding to each target pixel according to the input features corresponding to each target pixel. Among them, the input features of any target pixel among the target pixels include the features of any target pixel and the features of the target adjacent pixels adjacent to any target pixel.

[0068] In some embodiments, the biomass inversion system can implement S302 based on a trained biomass value prediction model. Among them, the biomass value prediction model is trained to output a biomass value based on the input features.

[0069] It should be noted that the biomass value prediction model can be pre-trained by the biomass inversion system, or can be pre-trained by other systems (such as a training system), and this embodiment does not make a limitation.

[0070] For example, if the biomass value prediction model has been pre-trained by the training system, the training system can transmit the trained biomass value prediction model to the biomass inversion system via a communication link. Alternatively, the training system can externally provide a call interface for calling the biomass value prediction model. Correspondingly, when the biomass inversion system has a biomass inversion requirement, it can call the biomass value prediction model through this call interface.

[0071] The process of training the biomass value prediction model is described elsewhere in this specification.

[0072] In this embodiment, on the one hand, the biomass value prediction model can determine the input features corresponding to each target pixel based on the target remote sensing data; on the other hand, on this basis, the biomass value prediction model can predict the biomass value based on the input features.

[0073] It should be noted that in this embodiment, the input features of a target pixel include not only the features of the target pixel itself but also the features of other pixels adjacent to the target pixel (i.e., target adjacent pixels) (which can also be referred to as local spatial feature information).

[0074] Among them, the features of the target pixel itself can be understood as the relevant information corresponding to the target pixel, such as the biological-related information corresponding to the target pixel in the actual scene (specifically, vegetation distribution information, etc.), and building-related information (such as building distribution information, etc.); the features of the target adjacent pixels can be understood as the information-related information corresponding to the target adjacent pixels, such as the biological-related information corresponding to the target adjacent pixels in the actual scene (specifically, vegetation distribution information, etc.), and building-related information (such as building distribution information, etc.).

[0075] That is to say, in this embodiment, when the biomass inversion system predicts the biomass value corresponding to a certain target pixel, it is not only based on the biological-related information corresponding to the target pixel, but also combines the biological-related information corresponding to the pixels adjacent to the target pixel on this basis. Therefore, the accuracy and reliability of the determined biomass value corresponding to the target pixel can be improved.

[0076] In addition, the target adjacent pixels can be one adjacent pixel or multiple adjacent pixels, which is not limited in this embodiment.

[0077] In some embodiments, "determining the input features corresponding to each target pixel in the target biomass inversion area according to the target remote sensing data" in S302 includes the following steps 11 to 13:

[0078] Step 11: Determine the target remote sensing index features corresponding to the target remote sensing data.

[0079] Exemplarily, the biomass inversion system can determine the target remote sensing index features based on remote sensing indices. A remote sensing index can be understood as a technology based on remote sensing technology, which constructs and enhances spectral and radar characteristics by combining different bands of satellite spectra and radar images, so as to reflect the characteristics of a certain ground object.

[0080] Correspondingly, the target remote sensing index features can be understood as images including the ground object information in the target remote sensing data. Therefore, the target remote sensing index features can also be referred to as target remote sensing index images or target remote sensing index data.

[0081] In some embodiments, the remote sensing indices may include: Normalized Difference Vegetation Index (NDVI), Normalized Difference Index (NDI45), Ratio Leaf Area Index (RFDI), Ratio Vegetation Index (RVI), and Drought Stress Index (DPI), etc. Among them, different remote sensing indices may correspond to different fluctuations.

[0082] Correspondingly, in combination with the above example, when the target remote sensing data is multi-source remote sensing data, in step 11, the biomass inversion system can determine the corresponding remote sensing indices based on the fluctuations corresponding to different remote sensing data respectively, so as to obtain the target remote sensing index features.

[0083] For example, in combination with Figure 4 it can be known that when obtaining multi-source remote sensing data (that is, the target remote sensing data is multi-source remote sensing data), the biomass inversion system can calculate the remote sensing indices, so as to obtain the corresponding target remote sensing index features, specifically, it can be the target remote sensing index image as shown in Figure 4 ...

[0084] Step 12: Stack the target remote sensing index features and the target remote sensing data to obtain the target stacked data.

[0085] Continuing to refer to the above example and Figure 4 , the biomass inversion system can fuse the target remote sensing index features and the target remote sensing data again (such as the stacking shown in Figure 4 ...), so as to make the content in the target stacked data more abundant.

[0086] In combination with the above example, when the target remote sensing data is multi-source remote sensing data, such as hyperspectral data, multispectral data, and L-SAR remote sensing image data, the biomass inversion system can stack the hyperspectral data and the remote sensing index features, so as to avoid data redundancy and enrich the content in the target stacked data as much as possible.

[0087] Step 13: Determine the input features corresponding to each target pixel according to the target stacked data.

[0088] Exemplarily, the biomass inversion system can use a model to determine the input features corresponding to each target pixel in the target stacked data. That is, step 13 can be understood as that, when the biomass inversion system obtains the target stacked data, it can determine the input features corresponding to each target pixel based on the biomass value prediction model.

[0089] Combined with the above analysis of steps 11 to 13, it can be seen that in this embodiment, the biomass inversion system can improve the richness of the target stacked data through data stacking, so as to improve the input features determined based on the target stacked data to contain as much content as possible, and further improve the accuracy and reliability of biomass inversion.

[0090] In some embodiments, the basic network model of the biomass value prediction model is a Light-Saturation Function Based Biomass Inversion (LSFBI) model. The biomass inversion model based on the light saturation function includes a backbone network, a convolutional layer, and a classifier; the backbone network is ResNet18; the number of output channels of the first convolutional layer (conv1) in the convolutional layer is multiple (such as 256, etc.); the output dimension of the output fully connected layer of the classifier is 1, which is used to output the biomass value (i.e., the regression value of biomass).

[0091] Exemplarily, the convolutional layer includes a convolutional kernel, and the size of the convolutional kernel is N*N, where N is an integer greater than 1; the above step 13 may include the following steps 21 and 22:

[0092] Step 21: For any target pixel in the target stacked data, the target stacked data is cropped into N*N input data with the target pixel as the center point.

[0093] It should be noted that the size of N in this specification is not limited, and it can be determined by the biomass inversion system based on requirements, historical records, experiments, etc. For example, N can be 3, or a larger or smaller value.

[0094] Taking N equal to 3 as an example, the biomass inversion system can crop the target stacked data with a size of 3*3 to generate an image with a size of 3*3*Y as the input data.

[0095] Among them, Y is the number of channels, and Y is an integer greater than 1. Exemplarily, the target stacked data can be stacked data of multiple channels. For example, Y can be 37.

[0096] Continue to refer to the above example and Figure 4, for each pixel in the target stacked data, the pixel can be used as the center point, and the target stacked data can be cropped with a size of N*N to obtain the cropped stacked data of N*N, and the cropped stacked data is used as the input data.

[0097] Step 22: Read the data of each target pixel in the input data based on the convolution kernel to obtain the input features corresponding to each target pixel.

[0098] Exemplarily, the biomass inversion system can read each target pixel in the input data, take values with a 3*3 sliding window as the corresponding input features.

[0099] Continuing to refer to the above example and Figure 4 , the biomass inversion system can input the input data obtained in Step 21 into the biomass value prediction model, so that the convolution kernel in the biomass value prediction model determines the corresponding input features of the input data, and the corresponding biomass value can be determined on this basis.

[0100] Based on the above analysis of Step 21 and Step 22, in this embodiment, the input data is the cropped 3*3 data, which is equivalent to a 3*3 cube data. Compared with the 1*1 input data, the features of the spatial dimension are increased, that is, the features of adjacent pixels are increased. Therefore, the determined input features are richer, thereby improving the accuracy and reliability of the later prediction.

[0101] Combined with the above analysis, the target stacked data can be the stacked data of multiple channels. In some embodiments, after obtaining the stacked data of multiple channels, the biomass inversion system normalizes the stacked data of each channel (Normalization), so as to improve the prediction efficiency and accuracy of the later biomass value prediction model by scaling the features to a similar scale.

[0102] Exemplarily, combined with the above example, if the number of channels is 32 and the remote sensing index is a 5-dimensional remote sensing index, the biomass inversion system can stack the hyperspectral data of 32 channels and the target remote sensing index images corresponding to the 5-dimensional remote sensing index, so as to obtain the stacked data with 37 channels. On this basis, the biomass inversion system normalizes the stacked data of each of the 37 channels.

[0103] In some embodiments, the normalization process includes the following Steps 31 and 32:

[0104] Step 31: For the stacked data of any channel among multiple channels, obtain the mean and standard deviation of the stacked data before cropping of any channel.

[0105] Step 32: Regularize the cropped input data for any channel according to the mean and standard deviation of any channel.

[0106] Exemplarily, for each channel, the biomass inversion system may first calculate the mean and standard deviation of the stacked data of the channel before cropping. Then, the biomass inversion system may regularize the stacked data (i.e., the input data) of the channel after cropping based on the calculated mean and standard deviation.

[0107] Based on the above analysis of Step 31 and Step 32, in this embodiment, the biomass inversion system regularizes by combining the mean and standard deviation, without compressing or stretching the distribution range of the data, so that the regularized data still retains the main characteristics of the original distribution. In addition, in the case where the target remote sensing data is multi-source remote sensing data, the problem of scale differences between multi-source remote sensing data can be effectively solved.

[0108] S303: Generate a biomass map corresponding to the target biomass inversion area according to the target biomass values corresponding to the respective target pixels.

[0109] Exemplarily, after obtaining the respective target biomass values, the biomass inversion system may redefine the length and width according to the size of the target stacked data to obtain a two-dimensional biomass prediction array corresponding to the respective target biomass values, and generate a biomass distribution map (i.e., the biomass map) through the geographic projection mode of the target stacked data.

[0110] Continuing to refer to the above example and Figure 4 , after the biomass value prediction model determines the biomass values corresponding to the respective target pixels in the stacked data, it may generate and output a corresponding biomass map based on the respective biomass values.

[0111] Combining the above analysis of S301 to S303, in this embodiment, the biomass inversion system can determine the input features corresponding to the respective pixels in the remote sensing data, and the input feature of each pixel includes, in addition to its own feature, the features of adjacent pixels, so that the content of the input feature of each pixel is more abundant. Therefore, when the biomass inversion system determines the biomass value according to the input feature, the accuracy and reliability of the determined biomass value can be improved. Furthermore, the accuracy and effectiveness of the biomass map generated based on the biomass value can be improved.

[0112] Combining the above example, the biomass value prediction model may be trained by the biomass inversion system or by the training system. This specification mainly takes the biomass inversion system training the biomass value prediction model as an example to elaborate on the training process of the biomass value prediction model.

[0113] Exemplarily, the biomass value prediction model is obtained by the biomass inversion system training a basic network model based on the obtained training set; the training set includes the sample remote sensing data of the sample biomass inversion area and the corresponding biomass values; the biomass value prediction model is a model that uses the corresponding biomass value as the training true value and learns the ability of the basic network model to predict the training true value based on the sample remote sensing data.

[0114] Among them, the sample remote sensing data is used to determine the input features corresponding to each sample pixel in the sample biomass inversion area, and the input features of any sample pixel among the sample pixels include the features of any sample pixel and the features of the sample adjacent pixels adjacent to any sample pixel.

[0115] For example, the biomass inversion system can obtain a training set for training the basic network model. The training set can include two parts of the sample biomass inversion area: sample remote sensing data and the corresponding biomass values. The sample biomass inversion area can be understood as the biomass inversion area corresponding to the sample remote sensing data.

[0116] Relatively speaking, the biomass inversion system can use the corresponding biomass value as the training label (i.e., the training true value), so that the biomass value predicted by the basic network model based on the sample remote sensing data is as close as possible to the training label, thereby obtaining a biomass value prediction model that can accurately output the biomass value based on the input data.

[0117] Similarly, in the training stage, the input features of each sample pixel include not only its own features but also the features of the pixels adjacent to it.

[0118] For the convenience of the reader's understanding, the principle of the biomass inversion system training the biomass value prediction model is elaborated in detail below.

[0119] Please refer to Figure 5 , Figure 5 which is the schematic flowchart of training the biomass value prediction model provided in this specification. As Figure 5 shown, the training process includes the following S501 to S508:

[0120] S501: Obtain sample remote sensing data.

[0121] Similarly, the sample remote sensing data can be multi-source remote sensing data, such as hyperspectral data, multispectral data, and L-SAR remote sensing image data.

[0122] In some embodiments, the hyperspectral data parameters may include spatial resolution, spectral resolution, and wavelength range. For example, the spatial resolution of hyperspectral data is 10 m, the spectral resolution is 2.5 nm, and the wavelength range is 400 - 1000 nm. The spatial resolution of multispectral data can range from 10 m to 60 m, and the wavelength range can be 443 - 2190 nm. The resolution of L-SAR remote sensing image data is better than 10 m * 10 m.

[0123] S502: Obtain the diameter at breast height data of each tree in the sample biomass inversion area. Among them, the time of the sample remote sensing data is as close as possible (e.g., close) to the diameter at breast height data.

[0124] The diameter at breast height data can be understood as the data obtained by measuring each tree in the sample biomass inversion area one by one and recording its various parameters (such as diameter at breast height, tree height, tree species information, where for shrubs, ground diameter, tree height, tree species information, etc. are recorded).

[0125] The sample biomass inversion area may include multiple sample plots, and the diameter at breast height data can be the diameter at breast height data corresponding to each sample plot.

[0126] In some embodiments, the biomass inversion system can obtain the diameter at breast height data based on methods such as stratified random sampling method, kilometer grid system sampling method, and manual selection sampling method.

[0127] Taking the stratified random sampling method as an example:

[0128] The biomass inversion system can first determine the number of sample plots n based on Equation 1. Equation 1:

[0129]

[0130] where t α is the reliability index, E is the allowable relative error of the biomass estimate, and its value is 1 minus the required accuracy; L is the total number of tree species participating in stratified sampling, such as the total number of tree species in the sample plot; W i is the proportion of the planting area of the i-th tree species in the total planting area of the participating tree species; S t represents the standard deviation of the NDVI value of the sample plot; represents the mean NDVI of the i-th tree species.

[0131] Then, the biomass inversion system can calculate the sample plot sampling quantity corresponding to each tree species based on Equation 2. For example, the sample plot sampling quantity n i of the i-th tree species. Equation 2:

[0132]

[0133] where S i is the standard deviation of NDVI within the distribution range of the i-th tree species.

[0134] Correspondingly, the biomass inversion system can obtain the corresponding diameter at breast height data based on the sample sampling quantity corresponding to each tree species.

[0135] Taking the kilometer grid system sampling method as an example:

[0136] The biomass inversion system can preset kilometer grids with a sample preset width and obtain corresponding multiple kilometer grid intersection points, so as to obtain the corresponding diameter at breast height data based on the obtained multiple kilometer grid intersection points.

[0137] Taking the manual selection sampling method as an example:

[0138] The biomass inversion system can obtain the tree species distribution dense area selected by the target user, the sample plot coordinates covering as many tree species as possible, and obtain the corresponding diameter at breast height data based on the sample plot coordinates.

[0139] Among them, the diameter at breast height data can be realized based on a Real-time kinematic (RTK) device.

[0140] S503: Calculate the corresponding biomass value according to the diameter at breast height data.

[0141] Exemplarily, the biomass inversion system can first calculate the biomass value of each tree based on the diameter at breast height data, and then accumulate the biomass values of all trees to obtain the total biomass value of the sample plot (i.e., the corresponding biomass value).

[0142] For example, the biomass inversion system can calculate the biomass value of individual trees (i.e., each tree) of different tree species in a certain sample plot based on a binary biomass calculation formula. The biomass value is the sum of the above-ground biomass value and the below-ground biomass value. After calculating the biomass value of individual trees, accumulate and calculate the biomass values of all individual trees in the sample plot, which is the biomass value of the sample plot.

[0143] S504: Determine the sample remote sensing index characteristics corresponding to the sample remote sensing data.

[0144] To avoid cumbersome statements, the same or similar content already described in the above examples will not be repeated. For example, regarding the implementation principle of S504, reference can be made to the implementation principle of determining the target remote sensing index characteristics corresponding to the target remote sensing data in the above examples, which will not be elaborated here.

[0145] S505: Stack the sample remote sensing index characteristics and the sample remote sensing data to obtain sample stacked data.

[0146] Similarly, regarding the implementation principle of S505, reference can be made to the implementation principle of stacking to obtain the target stacked data in the above examples, which will not be elaborated here.

[0147] In addition, during the training phase, the biomass inversion system can also regularize the sample stacked data. The specific implementation principle can refer to the implementation principle of regularizing the target stacked data in the above example, which will not be elaborated here.

[0148] S506: According to the plot numbers and plot center coordinates in the sample biomass inversion area, crop the sample stacked data with a size of 3 * 3 to obtain the cropped stacked data. Among them, the sample biomass inversion area includes multiple plots, and one plot corresponds to one plot number and plot center coordinate.

[0149] Exemplarily, combining the above analysis, it can be known that the sample biomass inversion area usually includes multiple plots. In order to distinguish different plots, the biomass inversion system can label each plot with a number by means of annotation. In addition, the biomass inversion system can also obtain the center coordinates of each plot.

[0150] Correspondingly, for each plot, the biomass inversion system can crop from the sample stacked data based on the plot number and the center coordinate to obtain the stacked data corresponding to the plot.

[0151] S507: Input the cropped stacked data into the basic network model to obtain the predicted biomass value.

[0152] Combined with the above example, for the cropped stacked data of each plot, the biomass inversion system can generate an image with a size of 3 * 3 * 37. The biomass inversion system can name the respective corresponding images based on the sample numbers of each plot. The biomass inversion system can match each plot sequence with the total biomass value of the corresponding plot as the training set. That is, the training set includes the biomass value corresponding to each plot and the cropped stacked data.

[0153] Correspondingly, the biomass inversion system can use the cropped stacked data as the input data of the basic network model, and the basic network model can predict the biomass value based on the input data.

[0154] S508: With the goal of minimizing the difference between the predicted biomass value and the corresponding biomass value, train to obtain a biomass value prediction model.

[0155] Exemplarily, the biomass inversion system can set training parameters, such as the number of iterations (such as 50 times), the learning rate (such as 0.01), the number of training batches each time (such as 32 batches), etc., and train to obtain a biomass value prediction model that meets the prediction requirements with the goal of minimizing the difference between the predicted biomass value and the corresponding biomass value.

[0156] In some embodiments, the biomass inversion system may set the coefficient of determination, root mean square error, etc. as performance evaluation indicators to adjust the training parameters according to the performance evaluation indicators.

[0157] It should be noted that the above examples are only used to exemplarily illustrate the possible implementation manners of the biomass inversion method in this specification, and should not be construed as a limitation on the implementation manners of the biomass inversion method in this specification. Exemplarily, on the basis of the above technical concept, some of the above technical features can be combined to obtain a new embodiment; new technical features can also be added on the basis of the above examples to obtain a new embodiment; some technical features can also be reduced on the basis of the above examples to obtain a new embodiment; some of the technical features in the above examples can be replaced with other technical features; some of the technical features and orders in the above examples can also be adjusted to obtain a new embodiment, and so on, which will not be listed one by one here.

[0158] According to the above technical concept, this specification also provides a computer-readable non-transitory storage medium, in which at least one instruction set is stored. When the at least one instruction set is executed by a processor, the steps of the biomass inversion method described in this specification are implemented.

[0159] In some possible embodiments, various aspects of this specification can also be implemented in the form of a program product, which includes program code. When the program product runs on the biomass inversion system 200, the program code is used to cause the biomass inversion system 200 to execute the steps of the biomass inversion method described in this specification. The program product for implementing the above method can be a portable compact disc read-only memory (CD-ROM) including program code and can run on the biomass inversion system 200. However, the program product of this specification is not limited to this. In this specification, the readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system. The program product can adopt any combination of one or more readable media. The readable media can be a readable signal medium or a readable storage medium. The readable storage medium can, for example, but not be limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the readable storage medium include: an electrical connection having one or more wires, a portable disc, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. The computer-readable storage medium can include a data signal propagated in a baseband or as part of a carrier wave, in which the readable program code is carried. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The readable storage medium can also be any readable medium other than the readable storage medium, and this readable medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium can be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above. The program code for performing the operations of this specification can be written in any combination of one or more programming languages, including object-oriented programming languages - such as Java, C++, etc., and also including conventional procedural programming languages - such as the "C" language or similar programming languages. The program code can be executed entirely on the biomass inversion system 200, partially on the biomass inversion system 200, executed as an independent software package, partially on the biomass inversion system 200 and partially on a remote biomass inversion system, or entirely on a remote biomass inversion system 200.

[0160] The above describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve the desired results. Additionally, the processes depicted in the figures do not necessarily require a particular order or a sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0161] In summary, after reading this detailed disclosure, those skilled in the art will understand that the foregoing detailed disclosure may be presented only by way of example and is not necessarily limiting. Although not explicitly stated herein, those skilled in the art can understand that this specification is intended to encompass various reasonable changes, improvements, and modifications to the embodiments. These changes, improvements, and modifications are intended to be proposed by this specification and are within the spirit and scope of the exemplary embodiments of this specification.

[0162] Furthermore, certain terms in this specification have been used to describe embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" mean that the specific features, structures, or characteristics described in connection with that embodiment may be included in at least one embodiment of this specification. Thus, it should be emphasized and understood that two or more references to "an embodiment" or "one embodiment" or "alternative embodiments" in various parts of this specification do not necessarily all refer to the same embodiment. Additionally, the specific features, structures, or characteristics may be appropriately combined in one or more embodiments of this specification.

[0163] It should be understood that in the foregoing description of the embodiments of this specification, for the purpose of helping to understand a feature and for the purpose of simplifying this specification, this specification combines various features in a single embodiment, drawing, or its description. However, this does not mean that the combination of these features is necessary. Those skilled in the art, when reading this specification, may well mark out some of these devices as separate embodiments for understanding. That is to say, the embodiments in this specification can also be understood as the integration of multiple sub - embodiments. And the content of each sub - embodiment is also valid when it has fewer features than all the features of a single foregoing disclosed embodiment.

[0164] Every patent, patent application, publication of patent applications, and other materials cited herein, such as articles, books, specifications, publications, documents, references, etc. (excluding any historical prosecution files associated therewith), are hereby incorporated by reference for all purposes relevant hereto, e.g., in the specification and claims of this application. However, in the event of any inconsistency or conflict between the description, definition, and / or terminology of such materials and the description, definition, and / or terminology used herein, the description, definition, and / or terminology used herein shall control.

[0165] Finally, it should be understood that the embodiments of the application disclosed herein are illustrative of the principles of the embodiments of this specification. Other modified embodiments are also within the scope of this specification. Accordingly, the embodiments disclosed in this specification are presented by way of example only and not by way of limitation. Those skilled in the art may implement the application in this specification by taking alternative configurations based on the embodiments in this specification. Therefore, the embodiments of this specification are not limited to the embodiments precisely described in the application.

Claims

1. A biomass inversion method, comprising: Obtain target remote sensing data of target biomass inversion area; Determine input features corresponding to each target pixel in the target biomass inversion area according to the target remote sensing data, and determine target biomass values ​​corresponding to each target pixel according to the input features corresponding to each target pixel, wherein the input features of any target pixel among the target pixels include features of the arbitrary target pixel and features of target adjacent pixels adjacent to the arbitrary target pixel; as well as A biomass map corresponding to the target biomass inversion area is generated according to the target biomass values ​​corresponding to the target pixels.

2. The method according to claim 1, wherein: The step of determining the input features corresponding to each target pixel in the target biomass inversion area according to the target remote sensing data includes: Determining a target remote sensing index feature corresponding to the target remote sensing data; Stacking the target remote sensing index feature and the target remote sensing data to obtain target stacking data; and An input feature corresponding to each of the target pixels is determined according to the target stacking data.

3. The method according to claim 2, wherein: The input features corresponding to the target pixels are determined based on a pre-trained biomass value prediction model, and the biomass value prediction model is trained to output a biomass value based on the input features; The biomass value prediction model includes an N*N convolution kernel, where N is an integer greater than 1; Determining input features corresponding to each of the target pixels based on the biomass value prediction model includes: For the arbitrary target pixel in the target stacked data, the target stacked data is cropped into the N*N input data with the arbitrary target pixel as the center point; as well as The data of each target pixel in the input data is read based on the convolution kernel to obtain the input features corresponding to each target pixel.

4. The method according to claim 3, wherein: The target stacking data includes stacking data of a plurality of channels, and the method further includes: For stacked data of any channel among the multiple channels, obtaining a mean and a standard deviation of the stacked data of any channel before clipping; and The cropped input data of the arbitrary channel is regularized according to the mean and standard deviation of the arbitrary channel.

5. The method according to claim 3, wherein: The basic network model of the biomass value prediction model is a biomass inversion model based on a light saturation function, and the biomass inversion model based on a light saturation function includes a backbone network, a convolutional layer, and a classifier; the backbone network is ResNet18; the number of output channels of the first convolutional layer in the convolutional layer is multiple; the output dimension of the output fully connected layer of the classifier is 1, which is used to output the biomass value.

6. The method according to any one of claims 3 to 5, wherein: The biomass value prediction model is obtained by training the basic network model based on the obtained training set; the training set includes sample remote sensing data of the sample biomass inversion area and the corresponding biomass value; the biomass value prediction model is a model in which the corresponding biomass value is used as the training true value and the basic network model learns to predict the training true value based on the sample remote sensing data; The sample remote sensing data is used to determine the input features corresponding to each sample pixel in the sample biomass inversion area, and the input features of any sample pixel among the sample pixels include the features of the arbitrary sample pixel and the features of sample adjacent pixels adjacent to the arbitrary sample pixel.

7. The method according to claim 6, wherein: Obtaining the training set includes: obtaining the sample remote sensing data; Obtaining the data of each tree length in the sample biomass inversion area; and The corresponding biomass value is calculated according to the data of each tree gauge to obtain the training set including the sample remote sensing data and the corresponding biomass value.

8. The method according to claim 6, wherein: The basic network model is trained based on the training set to obtain the biomass value prediction model, including: Determining a sample remote sensing index feature corresponding to the sample remote sensing data; Stacking the sample remote sensing index feature and the sample remote sensing data to obtain sample stacking data; Determining, based on the sample stacking data, input features corresponding to each of the sample pixels in the basic network model; and Based on the input features corresponding to each of the sample pixels, the basic network model is trained to predict the ability of the corresponding biomass value to obtain the biomass value prediction model.

9. The method according to claim 6, wherein: The sample remote sensing data is multi-source remote sensing data, and the multi-source remote sensing data includes hyperspectral data, multispectral data and L-SAR remote sensing image data.

10. A biomass inversion system, comprising: at least one storage medium storing at least one set of instructions for performing biomass inversion; At least one processor is communicatively connected to the at least one storage medium, wherein when the at least one processor is running, the at least one instruction set is read, and the method as described in any one of claims 1 to 9 is executed according to the instructions of the at least one instruction set.

Images