A hyperspectral soil information unmixing monitoring method and device
By constructing a perturbation matrix and a spatial-spectral constraint unmixing model, and combining spectral similarity and spatial smoothness constraints, the problem of low efficiency and accuracy of hyperspectral remote sensing pixel unmixing was solved, and high efficiency and accuracy of soil information monitoring were achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANDONG NORMAL UNIV
- Filing Date
- 2026-06-23
- Publication Date
- 2026-07-21
AI Technical Summary
In existing technologies, the unmixing efficiency and accuracy of hyperspectral remote sensing pixels are not high, resulting in insufficient accuracy in soil information monitoring.
By acquiring a mixed spectral training dataset, a perturbation matrix is constructed and a spatial spectral constraint unmixing model is established. Combining spectral similarity and spatial smoothness constraints, soil spectral variation information is determined, and finally, soil information inversion is performed.
It improved the efficiency and accuracy of hyperspectral remote sensing pixel unmixing, thereby enhancing the accuracy of soil information monitoring.
Smart Images
Figure CN122430264A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of soil monitoring technology, specifically relating to a method and device for monitoring the unmixing of hyperspectral soil information. Background Technology
[0002] Efficient monitoring of soil salinization levels is of great significance for ensuring arable land and food security. Currently, the monitoring of soil properties in non-bare land is mainly carried out by unmixing hyperspectral remote sensing pixels to obtain soil information. However, existing methods cannot quickly and effectively extract spectral details when unmixing hyperspectral remote sensing pixels, resulting in low unmixing efficiency and accuracy.
[0003] Therefore, how to improve the unmixing efficiency and accuracy of sub-pixel soil in hyperspectral remote sensing, thereby improving the monitoring accuracy of soil salinization level, is a technical problem that needs to be solved by those skilled in the art. Summary of the Invention
[0004] The purpose of this invention is to solve the technical problem that the unmixing efficiency and accuracy of hyperspectral remote sensing pixels in the prior art are not high, which leads to low accuracy in soil information monitoring.
[0005] To achieve the above technical objectives, on the one hand, the present invention provides a method for monitoring the unmixing of hyperspectral soil information, the method comprising:
[0006] Obtain the mixed spectrum training dataset;
[0007] A perturbation matrix is constructed based on the endmember feature bands of the hybrid spectral training dataset;
[0008] Construct a spatial spectrum constrained unmixing model based on the perturbation matrix;
[0009] Soil spectral variation information in the target area is determined based on the spatial-spectral constrained unmixing model.
[0010] Soil information inversion is performed on the target area based on the soil spectral variation information.
[0011] Furthermore, the acquisition of the mixed spectral training dataset specifically involves obtaining the in-situ measurement value of the corresponding pixel at the ground location corresponding to each pixel in the remote sensing image.
[0012] Furthermore, the endmember feature band is specifically the spectral feature band of the endmember information response in the mixed spectrum, and the perturbation matrix is specifically the additional matrix of the response after applying soil information to the endmember matrix A initialized with standard endmember features.
[0013] Furthermore, the loss function of the spatial spectrum-constrained unmixing model is:
[0014]
[0015]
[0016] In the formula, For endmember matrices, Here is the perturbation matrix. For abundance matrix, For pixels, , and For regularization penalty terms, Due to spatial constraints, For inter-spectral constraints, This represents the soil variation disturbance term. The number of endmembers, Let be the projection of the endmember matrix into (r-1) dimensional space.
[0017] Furthermore, the method also includes constructing constraints for the spatial spectral constraint unmixing model, specifically including spectral similarity constraints and spatial smoothness constraints.
[0018] Furthermore, the spectral similarity constraint The specific formula is as follows:
[0019]
[0020]
[0021] In the formula, The spectral vector of the land cover. The data are standard spectral data from the measured spectral library. With abundance and a constraint of 1, It is the Frobenius norm.
[0022] Furthermore, the spatial smoothness constraint The specific formula is as follows:
[0023]
[0024]
[0025] In the formula, For data points Corresponding abundance, For data points Corresponding abundance, The mean squared error of the spatial nearest neighbors. For data points Approaching data point The spatial weights are given by N, where N is the total number of samples.
[0026] On the other hand, the present invention also provides a hyperspectral soil information unmixing monitoring device, the device comprising:
[0027] The acquisition module is used to acquire the mixed spectrum training dataset;
[0028] The perturbation matrix construction module is used to construct a perturbation matrix based on the endmember feature bands of the mixed spectral training dataset.
[0029] The model building module is used to construct a spatial spectrum constrained unmixing model based on the perturbation matrix;
[0030] The solution module is used to determine the soil spectral variation information of the target area based on the spatial spectral constraint unmixing model.
[0031] The monitoring module is used to perform soil information inversion on the target area based on the soil spectral variation information.
[0032] This invention provides a method and apparatus for unmixing and monitoring hyperspectral soil information. Compared with existing technologies, this method first acquires a mixed spectral training dataset; then constructs a perturbation matrix based on the endmember feature bands of the mixed spectral training dataset; next, it constructs a spatial-spectral constrained unmixing model based on the perturbation matrix; then, it determines the soil spectral variation information of the target area based on the spatial-spectral constrained unmixing model; finally, it performs soil information inversion on the target area based on the soil spectral variation information. This method can improve the efficiency and accuracy of hyperspectral remote sensing pixel unmixing, thereby improving the accuracy of soil information monitoring. Attached Figure Description
[0033] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0034] Figure 1 The diagram shown is a flowchart illustrating the hyperspectral soil information unmixing monitoring method provided in the embodiments of this specification.
[0035] Figure 2 The diagram shown is a structural schematic of the hyperspectral soil information unmixing monitoring device provided in the embodiments of this specification. Detailed Implementation
[0036] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0037] like Figure 1 The diagram shown is a flowchart of the hyperspectral soil information unmixing monitoring method provided in the embodiments of this specification. Although this specification provides the method operation steps or device structure shown in the following embodiments or figures, based on convention or without creative effort, the method or device may include more or fewer operation steps or module units after partial merging. In steps or structures where there is no necessary causal relationship in logic, the execution order of these steps or the module structure of the device are not limited to the execution order or module structure shown in the embodiments or figures of this specification. When the method or module structure is applied in actual devices, servers, or terminal products, it can be executed sequentially or in parallel according to the method or module structure shown in the embodiments or figures (e.g., in a parallel processor or multi-threaded processing environment, or even in a distributed processing or server cluster implementation environment).
[0038] The hyperspectral soil information unmixing monitoring method provided in the embodiments of this specification can be applied to terminal devices such as client and server devices, such as... Figure 1 As shown, the method specifically includes the following steps:
[0039] Step S101: Obtain the mixed spectrum training dataset.
[0040] Specifically, a grid-based sampling method will be used to deploy sampling points, taking into account the transit time of ZY_02D satellite imagery and the requirements of mixed scenes, and fine-tuning the sampling positions based on the actual locations of the sampling points. To reduce errors caused by uneven spatial distribution, five subsamples of topsoil (0-20cm) with a radius of 20m will be collected, thoroughly mixed to obtain approximately 1kg of sample, and discarding root tissue, grass, and leaves. Simultaneously, a rapid soil moisture meter will be used to quickly determine the soil moisture content of the sampling points. A GPS receiver will record the geographical location of the sampling points.
[0041] Soil information was analyzed from soil samples separately, and soil property information analysis was conducted in the laboratory according to national standards.
[0042] The acquisition of the mixed spectral training dataset specifically involves obtaining the in-situ measurement value of the corresponding pixel at the ground location corresponding to each pixel in the remote sensing image.
[0043] Step S102: Construct a perturbation matrix based on the endmember feature bands of the hybrid spectral training dataset.
[0044] Specifically, most traditional hyperspectral unmixing algorithms treat variation information as noise to be eliminated, and rarely effectively represent and utilize the variation information of endmember spectral features. This application expresses the endmembers in a mixed pixel as a combination of "pure endmembers" and spectral variations, where a hyperspectral remote sensing image pixel X, composed of r endmembers, ...
[0045] Represents a cell end element in The variability perturbation, that is, the soil spectral variability characteristics specifically refer to the variability perturbation of endmembers in pixels, This represents the abundance of each pixel. This perturbation representation is pixel-by-pixel, thus each type of material endmember can be represented as a pure spectrum plus varying degrees of spectral variability. For remote sensing monitoring of soil salinity, Soil spectral variation information based on "pure soil end-members" is key information for soil salinity monitoring and retrieval. Other endmember variation information in the data expresses the spectral variation information of interfering ground features. In the unmixing model, the endmember spectral characteristic bands will be used to... Perform initialization.
[0046] Step S103: Construct a spatial spectrum constrained unmixing model based on the perturbation matrix.
[0047] Specifically, when constructing the spatial-spectral constraint unmixing model, constraints are first constructed for the spatial-spectral constraint unmixing model. These constraints include spectral similarity constraints and spatial smoothness constraints.
[0048] Distance-based spectral similarity metrics are grounded in vector norm theory. They assume spectral features are high-dimensional vectors in Euclidean space, with the bit depth equal to the number of spectral bands. Therefore, measuring spectral similarity is transformed into measuring the distance between these high-dimensional vectors. The greater the geometric distance between two high-dimensional vectors, the greater the difference between their spectra; conversely, the smaller the geometric distance, the higher the similarity between their spectra. Similarity constraints are imposed using Euclidean distance and standard vegetation spectra. The loss function for these spectral similarity constraints is shown in the following equation:
[0049]
[0050]
[0051] In the formula, The spectral vector of the land cover. The data are standard spectral data from the measured spectral library. With the constraint that the abundance sum is 1, It is the Frobenius norm. To adjust the influence weight of similarity and to better reflect the practical application of soil salinity estimation while reducing model complexity, the number of endmembers in the unmixing model was set to K=3, and then the model's iteration rules were derived.
[0052] Most constraints in existing hyperspectral unmixing methods are geared towards remote sensing land cover classification and identification, lacking constraints specifically addressing the spatial variability of salinized areas. We adopt the popular regular expression to smooth the spatial constraints, interpreting it as: the more similar the spectra of neighboring pixels, the more similar their abundance; that is, the more similar the data points... The closer to another data point When the abundance of the two is closer, the similarity of the expression space is calculated using the Heat Kernel weighting method. The spatial smoothness constraint is specifically shown in the following formula:
[0053]
[0054]
[0055] In the formula, For data points Corresponding abundance, For data points Corresponding abundance, The mean squared error of the spatial nearest neighbors. For data points Approaching data point The spatial weights are given by N, where N is the total number of samples.
[0056] In the above formula, Let $\mathbf{a}$ be the mean squared error of the spatial nearest neighbor. For the non-optimal nearest neighbors of the spatial nearest neighbor, their similarity to the center pixel is not high enough, but they are still close in space. Therefore, the spatial nearest neighbors are divided into two groups, and their mean squared errors are used as the scales affecting the weights, thus obtaining the above regularization expression.
[0057] The unmixing mentioned in this application refers to finding the optimal solution for the endmember matrix A, the abundance matrix S, and the perturbation matrix dA. To obtain better unmixing results, three constraint terms concerning them are introduced into the objective function to construct a new constrained optimization problem. The endmember and perturbation terms are regularized using the L2 norm to penalize spectral variability and limit the size of the solution space, as expressed below:
[0058]
[0059]
[0060] In the formula, For endmember matrices, Here is the perturbation matrix. For abundance matrix, For pixels, , and For regularization penalty terms, Due to spatial constraints, For inter-spectral constraints, This represents the soil variation disturbance term. for, The projection of the endmember matrix into (r-1) dimensional space. d is the basic mathematical symbol.
[0061] Step S104: Determine the soil spectral variation information of the target area based on the spatial-spectral constraint unmixing model.
[0062] Specifically, the soil spectral variation information of the target region is obtained by unmixing the target region based on the spatial-spectral constraint unmixing model. The specific solution process is to use the alternating direction multiplier method to solve each variable alternately to obtain the optimal endmember matrix A, the optimal abundance matrix S, and the optimal soil variation perturbation matrix in the target region. The soil variation perturbation matrix The column information is soil spectral variation information.
[0063] Step S105: Perform soil information inversion on the target area based on the soil spectral variation information.
[0064] Specifically, the soil spectral variation information of the target area is input into the inversion model to obtain the soil attribute information of the target area. This application does not impose specific restrictions on the inversion model. The inversion model and its training process are both existing technologies. For example, parameters such as input layer, hidden layer, and output layer are set for the inversion model, and training is gradually optimized.
[0065] Based on the above-described hyperspectral soil information unmixing monitoring method, one or more embodiments of this specification also provide a platform or terminal for hyperspectral soil information unmixing monitoring. This platform or terminal may include devices, software, modules, plug-ins, servers, clients, etc., using the methods described in the embodiments of this specification, combined with necessary hardware implementation. Based on the same innovative concept, the systems in one or more embodiments provided in this specification are as described in the following embodiments. Since the implementation schemes and methods for solving the system problem are similar, the specific system implementation in the embodiments of this specification can refer to the implementation of the aforementioned methods. Repeated details will not be repeated. The terms "unit" or "module" used below can refer to a combination of software and / or hardware that achieves a predetermined function. Although the systems described in the following embodiments are preferably implemented in software, hardware implementation, and a combination of software and hardware, are also possible and contemplated.
[0066] Specifically, Figure 2 This is a schematic diagram of the module structure of one embodiment of the hyperspectral soil information unmixing monitoring device provided in this specification, as shown below. Figure 2 As shown, the hyperspectral soil information unmixing monitoring device provided in this specification includes:
[0067] Module 201 is used to acquire the mixed spectrum training dataset;
[0068] The perturbation matrix construction module 202 is used to construct a perturbation matrix based on the endmember feature bands of the mixed spectral training dataset;
[0069] Model building module 203 is used to build a spatial spectrum constrained unmixing model based on the perturbation matrix;
[0070] Solver module 204 is used to determine the soil spectral variation information of the target area based on the spatial spectral constraint unmixing model;
[0071] The monitoring module 205 is used to perform soil information inversion on the target area based on the soil spectral variation information.
[0072] It should be noted that the system described above may include other implementation methods based on the description of the corresponding method embodiments. The specific implementation methods can be referred to the description of the corresponding method embodiments above, and will not be elaborated here.
[0073] This application also provides an electronic device, including:
[0074] processor;
[0075] Memory used to store the processor's executable instructions;
[0076] The processor is configured to perform the methods provided in the embodiments described above.
[0077] The electronic device provided in this application stores executable instructions for a processor in its memory. When the processor executes these instructions, it first acquires a mixed spectral training dataset; then, it constructs a perturbation matrix based on the endmember feature bands of the mixed spectral training dataset; next, it constructs a spatial-spectral constrained unmixing model based on the perturbation matrix; then, it determines the soil spectral variation information of the target area based on the spatial-spectral constrained unmixing model; and finally, it performs soil information inversion on the target area based on the soil spectral variation information. This improves the efficiency and accuracy of hyperspectral remote sensing pixel unmixing, thereby enhancing the accuracy of soil information monitoring.
[0078] The foregoing has described 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 may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.
[0079] The methods or apparatus described in the above embodiments of this specification can implement business logic through a computer program and record it on a storage medium. The storage medium can be read and executed by a computer to achieve the effects of the solutions described in the embodiments of this specification, such as:
[0080] Obtain the mixed spectrum training dataset;
[0081] A perturbation matrix is constructed based on the endmember feature bands of the hybrid spectral training dataset;
[0082] Construct a spatial spectrum constrained unmixing model based on the perturbation matrix;
[0083] Soil spectral variation information in the target area is determined based on the spatial-spectral constrained unmixing model.
[0084] Soil information inversion is performed on the target area based on the soil spectral variation information.
[0085] The storage medium can include physical devices for storing information, typically digitizing the information and then storing it using electrical, magnetic, or optical methods. The storage medium can include: devices that store information using electrical energy, such as various types of memory, like RAM and ROM; devices that store information using magnetic energy, such as hard disks, floppy disks, magnetic tapes, magnetic core memory, bubble memory, and USB flash drives; and devices that store information using optical methods, such as CDs or DVDs. Of course, there are other readable storage media, such as quantum memories and graphene memories.
[0086] The embodiments in this specification are not limited to conforming to industry communication standards, standard computer resource data update and data storage rules, or the situations described in one or more embodiments of this specification. Slightly modified implementations based on certain industry standards or custom methods or embodiments can also achieve the same, equivalent, or similar, or predictable, implementation effects as described above. Embodiments that utilize these modified or modified methods for data acquisition, storage, judgment, and processing still fall within the scope of optional implementations of the embodiments in this specification.
[0087] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, ASICs, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.
[0088] The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or plug-ins may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between devices or units, and may be electrical, mechanical, or other forms.
[0089] These computer program instructions can also be loaded onto a computer or other programmable resource data updating device, causing a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable device for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0090] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, system embodiments are basically similar to method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. In the description of this specification, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this specification. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification and the features of different embodiments or examples.
[0091] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.
Claims
1. A method for monitoring the unmixing of hyperspectral soil information, characterized in that, The method includes: Obtain the mixed spectrum training dataset; A perturbation matrix is constructed based on the endmember feature bands of the hybrid spectral training dataset; Construct a spatial spectrum constrained unmixing model based on the perturbation matrix; Soil spectral variation information in the target area is determined based on the spatial-spectral constrained unmixing model. Soil information inversion is performed on the target area based on the soil spectral variation information.
2. The method for unmixing and monitoring hyperspectral soil information as described in claim 1, characterized in that, The acquisition of the mixed spectral training dataset specifically involves obtaining the in-situ measurement value of the corresponding pixel at the ground location corresponding to each pixel in the remote sensing image.
3. The hyperspectral soil information unmixing monitoring method as described in claim 2, characterized in that, The endmember feature bands are specifically the spectral feature bands of the endmember information response in the mixed spectrum, and the perturbation matrix is specifically the additional matrix of the response after applying soil information to the endmember matrix A initialized with standard endmember features.
4. The method for unmixing and monitoring hyperspectral soil information as described in claim 1, characterized in that, The loss function of the spatial spectrum-constrained unmixing model is shown in the following equation: ； ； In the formula, For endmember matrices, Here is the perturbation matrix. For abundance matrix, For pixels, , and For regularization penalty terms, Due to spatial constraints, For inter-spectral constraints, This represents the soil variation disturbance term. The number of endmembers, Let be the projection of the endmember matrix into (r-1) dimensional space.
5. The method for unmixing and monitoring hyperspectral soil information as described in claim 4, characterized in that, The method also includes constructing constraints for the spatial spectral constraint unmixing model, specifically including spectral similarity constraints and spatial smoothness constraints.
6. The method for unmixing and monitoring hyperspectral soil information as described in claim 5, characterized in that, The spectral similarity constraint The specific formula is as follows: ； ； In the formula, The spectral vector of the land cover. The data are standard spectral data from the measured spectral library. With abundance and a constraint of 1, It is the Frobenius norm.
7. The method for unmixing and monitoring hyperspectral soil information as described in claim 6, characterized in that, The spatial smoothness constraint The specific formula is as follows: ； ； In the formula, For data points Corresponding abundance, For data points Corresponding abundance, The mean squared error of the spatial nearest neighbors. For data points Approaching data point The spatial weights are given by N, where N is the total number of samples.
8. A hyperspectral soil information unmixing monitoring device, characterized in that, The device includes: The acquisition module is used to acquire the mixed spectrum training dataset; The perturbation matrix construction module is used to construct a perturbation matrix based on the endmember feature bands of the mixed spectral training dataset. The model building module is used to construct a spatial spectrum constrained unmixing model based on the perturbation matrix; The solution module is used to determine the soil spectral variation information of the target area based on the spatial spectral constraint unmixing model. The monitoring module is used to perform soil information inversion on the target area based on the soil spectral variation information.