Abundance value determination method, device and computer readable storage medium

Abstract

The present application discloses a method, device and computer-readable storage medium for determining abundance values. A method for determining abundance values ​​includes: generating a correlation matrix of each pixel according to the correlation of reflection information between each pixel and the corresponding neighborhood of a satellite remote sensing reflection image; performing dimensionality reduction processing on the correlation matrix using a principal cause analysis algorithm to generate a principal component matrix; and processing the principal component matrix using a common space pattern algorithm to determine the abundance value of the end member corresponding to the ground object type in each pixel.

Description

Technical Field

[0001] The present application relates to the field of satellite remote sensing technology, and in particular to a method and device for determining an abundance value, and a computer-readable storage medium. Background Art

[0002] Satellite remote sensing technology has been widely used in monitoring the state of the earth. For example, remote sensing images can be used to determine the vegetation type and abundance value on the corresponding ground.

[0003] The publication number is: CN116469016A, and the name of the invention is a vegetation status monitoring method, device and storage medium based on abundance value, wherein the method includes: determining a plurality of first pixels corresponding to a predetermined area on the ground according to a satellite remote sensing reflection image generated based on a ground reflection spectrum signal; determining the abundance value of an end element corresponding to a ground object type in the first pixel according to a spectral feature corresponding to the first pixel; determining a second pixel corresponding to an edge of a vegetation area in the predetermined area according to the determined abundance value; determining a vegetation area according to the second pixel; and determining vegetation change status information in the predetermined area according to the determined vegetation area and pre-set reference area information, wherein the reference area is an area pre-determined in the predetermined area for planting vegetation.

[0004] In the above-mentioned prior art, the endmember abundance value in a pixel is usually determined by analyzing the reflection information of the pixel, thereby determining the vegetation classification value of the endmember in the pixel. However, a pixel may be a mixed pixel containing multiple endmembers, and the endmember corresponding to a type of land object may occupy multiple pixels continuously. Therefore, the method of determining the abundance value of the endmember in a pixel based solely on the reflection information of a pixel is inaccurate.

[0005] With respect to the technical problem of inaccurate abundance value calculation in the above-mentioned prior art, no effective solution has been proposed yet. Summary of the invention

[0006] The embodiments of the present application provide a method and device for determining an abundance value and a computer-readable storage medium, so as to at least solve the technical problem of inaccurate abundance value calculation in the prior art.

[0007] According to one aspect of an embodiment of the present application, a method for determining an abundance value is provided, including: generating a correlation matrix for each pixel based on the correlation of reflection information between each pixel of a satellite remote sensing reflection image and a corresponding neighborhood; performing dimensionality reduction processing on the correlation matrix using a principal cause analysis algorithm to generate a principal component matrix; and processing the principal component matrix using a common space pattern algorithm to determine the abundance value of the end member corresponding to the type of land object in each pixel.

[0008] According to another aspect of an embodiment of the present application, a device for determining an abundance value is also provided, including: a correlation matrix generation module, used to generate a correlation matrix for each pixel based on the reflection information correlation between each pixel of the satellite remote sensing reflection image and the corresponding neighborhood; a principal component matrix generation module, used to perform dimensionality reduction processing on the correlation matrix using a principal cause analysis algorithm to generate a principal component matrix; and an abundance value determination module, used to process the principal component matrix using a common space pattern algorithm to determine the abundance value of the end member corresponding to the ground object type in each pixel.

[0009] According to another aspect of an embodiment of the present application, a device for determining an abundance value is also provided, including: a processor; and a memory connected to the processor, for providing the processor with instructions for processing the following processing steps: generating a correlation matrix for each pixel based on the reflection information correlation between each pixel of the satellite remote sensing reflection image and the corresponding neighborhood; performing dimensionality reduction processing on the correlation matrix using a principal cause analysis algorithm to generate a principal component matrix; and processing the principal component matrix using a common space pattern algorithm to determine the abundance value of the end member corresponding to the ground object type in each pixel.

[0010] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0011] According to another aspect of the embodiments of the present application, a computer program product is also provided, including a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0012] In an embodiment of the present application, the computing device calculates the correlation between the pixel and its neighborhood based on the reflection information of each pixel, and generates a corresponding correlation matrix. After that, the computing device performs dimensionality reduction processing on the correlation matrix, and determines the abundance value of the end member corresponding to the ground object type in the pixel based on the correlation matrix generated by the reflection information of the pixel and its neighborhood after dimensionality reduction processing. Compared with the method in the prior art that determines the abundance value of the end member corresponding to the pixel only based on the reflection information of one pixel, the present technical solution associates the reflection information of multiple pixels, thereby calculating the abundance value of the end member corresponding to the most central pixel, so that the calculation of the abundance value is more accurate. And the correlation matrix is ​​sequentially reduced in dimensionality through the principal component analysis algorithm and the common space pattern algorithm, thereby reducing the amount of calculation and improving the calculation speed and efficiency. Thereby solving the technical problem of inaccurate calculation of abundance value existing in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0014] Figure 1 is a hardware structure block diagram of a computing device for implementing the method according to Embodiment 1 of the present application;

[0015] Figure 2 is a flow chart of the method for determining the abundance value according to the first aspect of Example 1 of the present application;

[0016] Figure 3 is a schematic diagram of a portion of a satellite remote sensing reflection image according to Example 1 of the present application;

[0017] Figure 4 is a module schematic diagram of the common space mode algorithm according to Example 1 of the present application;

[0018] Figure 5 is a sequential flow chart of the method for determining the abundance value according to Example 1 of the present application;

[0019] Figure 6 is a schematic diagram of a device for determining an abundance value according to Example 2 of the present application; and

[0020] Figure 7 It is a schematic diagram of the device for determining the abundance value according to Example 3 of the present application. DETAILED DESCRIPTION

[0021] In order to enable those skilled in the art to better understand the technical solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only embodiments of a part of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present application.

[0022] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0023] Example 1

[0024] According to this embodiment, a method embodiment of a method for determining an abundance value is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0025] The method embodiment provided in this embodiment can be executed in a mobile terminal, a computer terminal, a server or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computing device for implementing a method for determining an abundance value. Figure 1 As shown, the computing device may include one or more processors (the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory for storing data, and a transmission device for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0026] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry". The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuitry may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computing device. As described in the embodiments of the present application, the data processing circuitry serves as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0027] The memory can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the method for determining the abundance value in the embodiment of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, the method for determining the abundance value of the above-mentioned application is realized. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computing device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0028] The transmission device is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computing device. In one example, the transmission device includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device can be a radio frequency (Radio Frequency, RF) module, which is used to communicate with the Internet wirelessly.

[0029] The display may be, for example, a touch screen liquid crystal display (LCD) that may enable a user to interact with a user interface of the computing device.

[0030] It should be noted that, in some optional embodiments, the above Figure 1 The computing device shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of both hardware elements and software elements. Figure 1 This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the computing devices described above.

[0031] Under the above operating environment, according to the first aspect of this embodiment, a method for determining an abundance value is provided. The method comprises: Figure 1 The computing device implementation shown in . Figure 2 A schematic diagram showing the process of the method is shown in FIG. Figure 2 As shown, the method includes:

[0032] S202: generating a correlation matrix of each pixel according to the correlation of reflection information between each pixel of the satellite remote sensing reflection image and the corresponding neighborhood;

[0033] S204: Using a principal cause analysis algorithm to perform dimensionality reduction processing on the correlation matrix to generate a principal component matrix; and

[0034] S206: Processing the principal component matrix using a common spatial pattern algorithm to determine the abundance value of the end member corresponding to the ground feature type in each pixel.

[0035] Specifically, Figure 3 The schematic diagram of a partial satellite remote sensing reflection image corresponding to a ground area obtained from a satellite is shown, wherein the partial satellite remote sensing reflection image includes a plurality of pixels, namely P1 to P9.

[0036] First, the computing device determines the neighborhoods of P1 to P9 (pixels) respectively. For example, the neighborhood of P1 is P2, P4 and P5; the neighborhood of P2 is P1, P3, P4, P5 and P6; the neighborhood of P5 is P1 to P4 and P6 to P9. Then the computing device collects the reflection information of P1 to P9, and calculates the reflection information correlation between P1 to P9 and the corresponding neighborhoods according to the reflection information of P1 to P9. Then the computing device generates a correlation matrix corresponding to P1 to P9 respectively based on the reflection information correlation of P1 to P9. The method for calculating the reflection information correlation can be to calculate the similarity.

[0037] Furthermore, the computing device extracts principal components from the correlation matrices of P1 to P9 respectively through the principal cause analysis method, and generates principal component matrices corresponding to P1 to P9 respectively according to the extracted principal components, thereby achieving dimensionality reduction processing of the correlation matrices of P1 to P9.

[0038] Furthermore, the computing device processes the principal component matrix of P1 to P9 by a common spatial pattern algorithm to generate feature information corresponding to P1 to P9. Then, the computing device determines the abundance value of the end member corresponding to the feature type in P1 to P9 according to the feature information corresponding to P1 to P9. The feature type may include, for example, green vegetation, buildings, vacant land, agricultural land, and water bodies. For example, the computing device determines the abundance value of the green vegetation end member in P1 to P9.

[0039] As described in the background technology, the endmember abundance value in a pixel is usually determined by analyzing the reflection information of the pixel, thereby determining the vegetation classification value of the endmember in the pixel. However, a pixel may be a mixed pixel containing multiple endmembers, and the endmember corresponding to a type of land feature may occupy multiple pixels continuously. Therefore, the method of determining the abundance value of the endmember in a pixel based solely on the reflection information of a pixel is inaccurate.

[0040] In view of the technical problems described above, through the technical solution of the embodiment of the present application, the computing device calculates the correlation between the pixel and its neighborhood according to the reflection information of each pixel, and generates a corresponding correlation matrix. After that, the computing device performs dimensionality reduction processing on the correlation matrix, and determines the abundance value of the end member corresponding to the ground object type in the pixel according to the correlation matrix generated by the reflection information of the pixel and its neighborhood after dimensionality reduction processing. Compared with the method in the prior art that determines the abundance value of the end member corresponding to the pixel only according to the reflection information of one pixel, the technical solution associates the reflection information of multiple pixels, thereby calculating the abundance value of the end member corresponding to the most central pixel, so that the calculation of the abundance value is more accurate. And the correlation matrix is ​​sequentially reduced in dimensionality through the principal component analysis algorithm and the common space pattern algorithm, thereby reducing the amount of calculation and improving the calculation speed and efficiency. Thereby solving the technical problem of inaccurate calculation of abundance value existing in the prior art.

[0041] Optionally, the operation of generating a correlation matrix of each pixel based on the reflection information correlation between each pixel of the satellite remote sensing reflection image and the corresponding neighborhood includes: selecting a target pixel from each pixel of the satellite remote sensing reflection image; calculating the reflection information correlation between the target pixel and the corresponding neighborhood based on the reflection information of the target pixel and the reflection information of the corresponding neighborhood; and generating a correlation matrix of the target pixel based on the reflection information of the pixel and the corresponding reflection information correlation.

[0042] Specifically, refer to Figure 3 As shown, the computing device selects target pixels from among the pixels of the satellite remote sensing reflection image, ie, P1 to P9, that is, P1 to P9 are taken as target pixels in sequence.

[0043] Taking P5 as the target pixel, the computing device first determines the neighborhood of the target pixel P5. The neighborhood of the target pixel P5 is P1 to P4 and P6 to P9. The neighborhood of the pixel can also be expanded outward, for example, the neighborhood of P5 has g-1 pixels. Then the computing device collects the target pixel and its neighborhood, that is, P1 to P g Reflection information X i =[x 1,i ,x 2,i ,...,x m,i ] TWhere i=1-g (number of pixels), m is the number of bands used to represent the reflection information. The m bands may be, for example, infrared information, ultraviolet information, and the like.

[0044] For example, the reflection information of P1: X1 = [x 1,1 ,x 2,1 ,...,x m,1 ] T ;

[0045] Reflection information of P2: X2 = [x 1,2 ,x 2,2 ,...,x m,2 ] T ;

[0046] …

[0047] P g Reflection information: X g =[x 1,g ,x 2,g ,...,x m,g ] T .

[0048] Furthermore, the computing device calculates the correlation according to P1~P g The reflection information of P5 and P i (i.e. P1~P g )'s reflection information correlation r 5,i For example, to calculate P a With P b The reflection information correlation r a,b The correlation calculation formula is:

[0049]

[0050] For example, the reflection information correlation r between target pixels P5 and P1 is 5,1 for:

[0051]

[0052] Correlation r between the reflection information of target pixels P5 and P2 5,2 for:

[0053]

[0054] …

[0055] The target pixel P5 and P g The reflection information correlation r 5,g for:

[0056]

[0057] Furthermore, the computing device uses a preset formula Yt i =X i *r 5,i Generate the correlation matrix Y5 of the target pixel P5 = [Yt1, Yt2, ..., Yt g ] T ,in

[0058] Yt1=[y 1,1 ,y 1,2 ,...,y 1,m ];

[0059] Yt2=[y 2,1 ,y 2,2 ,...,y 2,m ];

[0060] …

[0061] Y g =[y g,1 ,y g,2 ,...,y g,m ];

[0062] thereby,

[0063]

[0064] Furthermore, the computing device calculates the correlation matrices of other target pixels according to the above-mentioned method of calculating the correlation matrix Y5 of the target pixel P5, which will not be described in detail here.

[0065] Therefore, the technical solution improves the correlation between the target pixel and the neighborhood by associating the reflection information between the target pixel and the neighborhood and calculating the correlation between the target pixel and the neighborhood.

[0066] Optionally, the operations of reducing the dimension of the correlation matrix using a principal cause analysis algorithm to generate a principal component matrix include: normalizing the correlation matrix to generate a normalized matrix; generating a correlation matrix based on the normalized matrix; and generating a principal component matrix based on a preset number of principal components.

[0067] Specifically, the computing device normalizes the correlation matrices of P1 to P9 to generate corresponding normalized matrices. Taking P5 as an example, the computing device determines the correlation matrix of P5 Y5 = [Yt1, Yt2, ..., Yt g ] T The computing device then uses the normalization formula to calculate the correlation matrix Y5 = [Yt1, Yt2, ..., Yt g ] TPerform normalization processing to generate the corresponding normalized matrix Z5=[Zt1,Zt2,...,Zt g ] T . The normalization formula is:

[0068]

[0069] Where i=1~g, j=1~m.

[0070] Where Zt1=[z 1,1 ,z 1,2 ,...,z 1,m ];

[0071] Zt2=[z 2,1 ,z 2,2 ,...,z 2,m ];

[0072] …

[0073] Z g =[z g,1 ,z g,2 ,...,z g,m ].

[0074] Afterwards, the computing device calculates the matrix Z5=[Zt1, Zt2, ..., Zt g ] T Generate correlation matrix

[0075]

[0076] Where D1 = [d 1,1 ,d 1,2 ,...,d 1,g ] T ;

[0077] D2=[d 2,1 ,d 2,2 ,...,d 2,g ] T ;

[0078] …

[0079] D g =[d g,1 ,d g,2 ,...,d g,g ] T .

[0080] Furthermore, the computing device is pre-set with the number of principal components k for generating the unit eigenvector, and the eigenvalues ​​λ1 to λ kThe unit eigenvector of g ] T , generate the principal component matrix corresponding to P5.

[0081] In addition, the computing device is pre-set with the number of principal components k, wherein the method for determining the number of principal components is as follows:

[0082] First, the computing device obtains the sample reflection information Sx of the target pixel P5 and g pixels in the neighborhood. i =[sx 1,i ,sx 2,i ,...,sx m,i ] T . Where i = 1 ~ g (number of pixels), m is the number of bands used to represent the corresponding reflection information. And

[0083] For example, the sample reflection information of P1: Sx1 = [sx 1,1 ,sx 2,1 ,...,sx m,1 ] T ;

[0084] Sample reflection information of P2: Sx2 = [sx 1,2 ,sx 2,2 ,...,sx m,2 ] T ;

[0085] …

[0086] P g Sample reflection information of: Sx g =[sx 1,g ,sx 2,g ,...,sx m,g ] T .

[0087] Furthermore, the computing device generates a sample correlation matrix corresponding to the sample reflection information in accordance with the above-mentioned method of generating a correlation matrix corresponding to the reflection information. Afterwards, the computing device uses |SR5-λI|=0 to obtain g eigenvalues ​​of SR5: λ1≥λ2≥...≥λ g The computing device then calculates each eigenvalue λ according to the variance contribution rate calculation formula i The variance contribution rate α i The variance contribution rate calculation formula is:

[0088]

[0089] For example, the variance contribution rate α1 of eigenvalue λ1 is:

[0090]

[0091] The variance contribution rate α2 of the eigenvalue λ2 is:

[0092]

[0093] …

[0094] Eigenvalue λ g The variance contribution rate α g :

[0095]

[0096] Further, the computing device calculates the sum of variance contribution rates of multiple eigenvalues ​​starting from eigenvalue λ1 as the cumulative variance contribution rate until the cumulative variance contribution rate obtained by the kth eigenvalue reaches a preset threshold value, for example, the preset threshold value is 70%, where k≤g.

[0097] For example, α1+α2+...+α k It just reaches more than 70%, thus k is determined as the number of principal components.

[0098] Thus, the computing device determines the principal component matrices of other pixels according to the above-mentioned method of generating the principal component matrix corresponding to P5, which will not be repeated here.

[0099] Therefore, this technical solution can normalize the data with more prominent values ​​into a reasonable range by normalizing the correlation matrix, thereby reducing the impact on the results.

[0100] Optionally, the operation of generating a principal component matrix according to a preset number of principal components includes: processing a correlation matrix according to the number of principal components to determine a unit eigenvector; and generating a principal component matrix according to the unit eigenvector and the correlation matrix.

[0101] Specifically, taking the target pixel P5 as an example, the computing device processes the correlation matrix R5 according to the number of principal components k, and determines k unit eigenvectors D1 to D k . Where D L =[d L,1 ,d L,2 ,...,d L,g ], L = 1~k. Thus, the computing device generates the corresponding unit characteristic matrix A5 according to the unit characteristic vector:

[0102]

[0103] Furthermore, the computing device converts the unit feature vector A5 and the correlation matrix Y5 of the target pixel P5 into [Yt1, Yt2, ..., Yt g ] T Calculate and generate the principal component matrix V5 corresponding to the target pixel P5 = [Vt1, Vt2, ..., Vt k ] T :

[0104]

[0105] Where Vt1 = [v 1,1 ,v 1,2 ,...,v 1,m ];

[0106] Vt2=[v 2,1 ,v 2,2 ,...,v 2,m ];

[0107] …

[0108] Vt k =[v k,1 ,v k,2 ,...,v k,m ].

[0109] Therefore, this technical solution will extract the principal components of the correlation matrix, thereby simplifying the complexity of the data and facilitating subsequent data processing and analysis.

[0110] Optionally, the operation of processing the principal component matrix using a common spatial pattern algorithm to determine the abundance value of the endmember corresponding to the land feature type in each pixel includes: processing the principal component matrix using a common spatial pattern algorithm corresponding to the land feature type to generate multiple feature information; classifying the multiple feature information respectively through a preset binary classification model to determine the land feature type probability of the endmember corresponding to the land feature type; and determining the abundance value of the endmember corresponding to the land feature type in each pixel according to the land feature type probability through a softmax classifier.

[0111] Specifically, refer to Figure 4 As shown, the computing device is pre-set with multiple common space mode algorithms corresponding to multiple ground object types. For example, if there are n ground object types, then ground object types 1 to n correspond to common space mode algorithms 1 to n respectively. Then the computing device uses common space mode algorithms 1 to n to respectively correspond to pixel P i The corresponding principal component matrix V i Processing is performed to generate corresponding n feature information, namely feature information 1 to n.

[0112] Furthermore, the computing device inputs the feature information 1 to n generated by the common spatial pattern algorithms 1 to n into the corresponding binary classification models 1 to n, respectively, and classifies the feature information 1 to n by the binary classification models 1 to n, and determines the feature type probability of the end member corresponding to the feature type. The binary classification models 1 to n are respectively used to classify the end members of the feature types 1 to n. The feature types include n types such as green vegetation, buildings, vacant land, agricultural land, and water bodies. For example, the binary classification model 1 is used to classify the green vegetation end member, and the computing device inputs the feature information 1 into the binary classification model 1. Then, the computing device determines the probability that the feature information 1 is a green vegetation end member and the probability that it is not a green vegetation end member through the binary classification model 1. Thus, the computing device can determine whether the feature information 1 is a green vegetation end member according to the output probability value.

[0113] For another example, the binary classification model 2 is used to classify building end-members, and the computing device inputs the feature information 2 into the binary classification model 2. Then, the computing device determines the probability that the feature information 2 is a building end-member and the probability that it is not a building end-member through the binary classification model 2. Thus, the computing device can determine whether the feature information 2 is a building end-member according to the output probability value.

[0114] Furthermore, the computing device simultaneously inputs the probabilities output by the binary classification models 1 to n into a preset softmax classifier, thereby outputting the end members corresponding to the ground feature types 1 to n in the pixel P through the softmax classifier. i The abundance value Af 1,i ~Af n,i .

[0115] Therefore, in this technical solution, by using multiple binary classification models, the corresponding feature representation can be determined for the end members of different types of land objects, and then the output of all binary classification models is integrated through the softmax classifier, which helps to capture the possible information associations between different models. In this way, the final abundance value can take into account information from multiple aspects, thereby improving the overall accuracy and stability.

[0116] Optionally, the operation of processing the principal component matrix using a common space pattern algorithm to generate characteristic information includes: generating a correlation matrix based on the multi-channel information in the principal component matrix, wherein the correlation matrix is ​​used to indicate the correlation between the multi-channel information; and extracting features from the correlation matrix through a projection matrix to generate characteristic information.

[0117] Specifically, taking the processing of the component matrix V5 of the target pixel P5 by the common spatial pattern algorithm 1 to generate feature information as an example, the computing device obtains the principal component matrix V5 corresponding to the target pixel P5 = [Vt1, Vt2, ..., Vt k ] T, Vt L As multi-channel information, where L=1~k.

[0118] Further, the computing device constructs a covariance matrix according to the multi-channel information as a correlation matrix, wherein the covariance matrix is ​​constructed according to the following formula:

[0119]

[0120] in, represents the matrix corresponding to multi-channel information, represents the field of real numbers of dimension k×m, tr[] represents the trace of the matrix, and is the covariance matrix corresponding to multi-channel information, represents the field of real numbers of dimension k×k.

[0121] Furthermore, the computing device extracts features from the correlation matrix using the pre-trained projection matrix H corresponding to the common space pattern algorithm 1 to generate feature information. The computing device calculates the feature information according to the following formula:

[0122] F = diag (HTEH),

[0123] in is the feature information, where q is the preset number of features, H is the projection matrix for feature extraction of the correlation matrix E, and diag() represents taking the diagonal elements.

[0124] The projection matrix H corresponding to the common space mode algorithm 1 is obtained by pre-training, and the training process is as follows:

[0125] For example, the common space mode algorithm 1 corresponds to the green vegetation end member, so the computing device obtains the sample reflection information of the green vegetation end member, thereby constructing the first reflection information sample set W1. And the computing device obtains the sample reflection information of the non-green vegetation end member, thereby constructing the second reflection information sample set W2. Wherein the reflection information sample set W1 = {w 1,1 ,w 1,2 ,...,w 1,u}, that is, the first reflection information sample set W1 includes u groups of sample reflection information w 1,1 ~w 1,u . The sample reflection information is:

[0126] w 1,1 =[wt 1,1,1 ,wt 1,2,1 ,...,wt 1,m,1 ] T ;

[0127] w 1,2 =[wt 1,1,2,wt 1,2,2 ,...,wt 1,m,2 ]; ...

[0128] w 1,u =[wt 1,1,u ,wt 1,2,u ,...,wt 1,m,u ] T .

[0129] WS 1,j,h It represents the j-th sample reflection information of the h-th group in the first reflection information sample set W1, where h=1~u, j=1~m.

[0130] And the reflection information sample set W2 = {w 2,1 ,w 2,2 ,...,w 2,u}, that is, the second reflection information sample set W2 includes u groups of sample reflection information w 2,1 ~w 2,u . The sample reflection information is:

[0131] w 2,1 =[wt 2,1,1 ,wt 2,2,1 ,...,wt 2,m,1 ] T ;

[0132] w 2,2 =[wt 2,1,2 ,wt 2,2,2 ,...,wt 2,m,2 ]; ...

[0133] w 2,u =[wt 2,1,u ,wt 2,2,u ,...,wt 2,m,u ] T .

[0134] WS 2,j,h It represents the j-th sample reflection information of the h-th group in the second reflection information sample set W2, where h=1~u, j=1~m.

[0135] Furthermore, the computing device establishes u sample association matrices corresponding to the reflection information sample set W1 (i.e., first sample association matrices) and u sample association matrices corresponding to the reflection information sample set W2 (i.e., second sample association matrices) according to the above-mentioned method for constructing association matrices. After that, the computing device calculates the average values ​​corresponding to each first sample association matrix respectively, and generates a sample average matrix Wa1 (i.e., first sample average matrix) according to the obtained multiple average values, and the computing device calculates the average values ​​corresponding to each second sample association matrix respectively, and generates a sample average matrix Wa2 (i.e., second sample average matrix) according to the obtained multiple average values.

[0136] Furthermore, the computing device determines the projection matrix H according to the principle of maximizing the difference between the signals of the first sample average matrix Wa1 and the second sample average matrix Wa2:

[0137]

[0138] The projection matrix represents a real number field of dimension m×q, where q is a predefined number.

[0139] In addition, the types of land features include green vegetation, buildings, open space, agricultural land, and water bodies, etc. Therefore, the computing device can correspond the common space mode algorithm 2 to the building end member, thereby training the projection matrix corresponding to the common space mode 2; the computing device corresponds the common space mode algorithm 3 to the open space end member, thereby training the projection matrix corresponding to the common space mode 3, and so on, thereby training the projection matrix corresponding to the common space mode n.

[0140] Therefore, by constructing a common space pattern algorithm corresponding to each type of land feature, the technical solution can process the correlation matrix with emphasis through each different common space pattern algorithm to achieve multi-dimensional and multi-faceted feature extraction.

[0141] In summary, reference Figure 5 As shown, the sequential steps of the computing device determining the abundance value of the end member corresponding to the ground feature type in each pixel are as follows:

[0142] S501: The computing device selects a target pixel from each pixel of the satellite remote sensing reflection image;

[0143] S502: The computing device calculates the reflection information correlation between the target pixel and the corresponding neighborhood according to the reflection information of the target pixel and the reflection information of the corresponding neighborhood;

[0144] S503: The computing device generates a correlation matrix of the target pixel according to the reflection information of the pixel and the corresponding reflection information correlation;

[0145] S504: The computing device normalizes the correlation matrix to generate a normalized matrix;

[0146] S505: The computing device generates a correlation matrix according to the specification matrix;

[0147] S506: The computing device processes the correlation matrix according to a preset number of principal components to determine a unit eigenvector;

[0148] S507: The computing device generates a principal component matrix according to the unit eigenvector and the correlation matrix;

[0149] S508: The computing device generates a correlation matrix according to the multi-channel information in the principal component matrix;

[0150] S509: The computing device extracts features from the correlation matrix through the projection matrix to generate feature information;

[0151] S510: The computing device classifies the plurality of feature information respectively through a preset binary classification model, and determines the ground feature type probability of the end member corresponding to the ground feature type;

[0152] S511: The computing device determines the abundance value of the end member corresponding to the ground object type in each pixel according to the ground object type probability through the softmax classifier.

[0153] In addition, according to a second aspect of this embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0154] In addition, according to a third aspect of this embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0155] Therefore, according to this embodiment, the computing device calculates the correlation between the pixel and its neighborhood according to the reflection information of each pixel, and generates a corresponding correlation matrix. After that, the computing device performs dimensionality reduction processing on the correlation matrix, and determines the abundance value of the end member corresponding to the ground object type in the pixel according to the correlation matrix generated by the reflection information of the pixel and its neighborhood after dimensionality reduction processing. Compared with the method in the prior art that determines the abundance value of the end member corresponding to the pixel only according to the reflection information of one pixel, the present technical solution associates the reflection information of multiple pixels, thereby calculating the abundance value of the end member corresponding to the most central pixel, so that the calculation of the abundance value is more accurate. And the correlation matrix is ​​sequentially reduced in dimensionality through the principal component analysis algorithm and the common space pattern algorithm, thereby reducing the amount of calculation and improving the calculation speed and efficiency. Thereby solving the technical problem of inaccurate calculation of abundance value existing in the prior art.

[0156] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0157] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0158] Example 2

[0159] Figure 6 The apparatus 600 for determining the abundance value according to this embodiment is shown, and the apparatus 600 corresponds to the method according to the first aspect of embodiment 1. Figure 6 As shown, the device 600 includes: a correlation matrix generation module 610, which is used to generate a correlation matrix for each pixel according to the reflection information correlation between each pixel and the corresponding neighborhood of the satellite remote sensing reflection image; a principal component matrix generation module 620, which is used to perform dimensionality reduction processing on the correlation matrix using a principal cause analysis algorithm to generate a principal component matrix; and an abundance value determination module 630, which is used to process the principal component matrix using a common space pattern algorithm to determine the abundance value of the end member corresponding to the ground object type in each pixel.

[0160] Optionally, the correlation matrix generation module 610 includes: a pixel selection submodule, used to select a target pixel from each pixel of the satellite remote sensing reflection image; a correlation calculation submodule, used to calculate the reflection information correlation between the target pixel and the corresponding neighborhood based on the reflection information of the target pixel and the reflection information of the corresponding neighborhood; and a first generation submodule, used to generate a correlation matrix of the target pixel based on the reflection information of the pixel and the corresponding reflection information correlation.

[0161] Optionally, the principal component matrix generation module 620 includes: a second generation submodule, used to normalize the correlation matrix to generate a normalized matrix; a third generation submodule, used to generate a correlation matrix based on the normalized matrix; and a fourth generation submodule, used to generate a principal component matrix based on a preset number of principal components.

[0162] Optionally, the fourth generating submodule includes: a determining unit, used to process the correlation matrix according to a preset number of principal components to determine the unit eigenvector; and a generating unit, used to generate the principal component matrix according to the unit eigenvector and the correlation matrix.

[0163] Optionally, the abundance value determination module 630 includes: a fifth generation submodule, used to process the principal component matrix using a common spatial pattern algorithm corresponding to the land feature type to generate corresponding feature information; a probability determination submodule, used to classify the corresponding feature information respectively through a preset binary classification model, and determine the land feature type probability of the end member corresponding to the land feature type; and an abundance value determination submodule, used to determine the abundance value of the end member corresponding to the land feature type in each pixel according to the land feature type probability through a softmax classifier.

[0164] Optionally, the fifth generation submodule includes: a matrix generation unit, used to generate a correlation matrix based on the multi-channel information in the principal component matrix, wherein the correlation matrix is ​​used to indicate the correlation between the multi-channel information; and an information generation unit, used to extract features from the correlation matrix through a projection matrix to generate feature information.

[0165] Therefore, according to this embodiment, the computing device calculates the correlation between the pixel and its neighborhood according to the reflection information of each pixel, and generates a corresponding correlation matrix. After that, the computing device performs dimensionality reduction processing on the correlation matrix, and determines the abundance value of the end member corresponding to the ground object type in the pixel according to the correlation matrix generated by the reflection information of the pixel and its neighborhood after dimensionality reduction processing. Compared with the method in the prior art that determines the abundance value of the end member corresponding to the pixel only according to the reflection information of one pixel, the present technical solution associates the reflection information of multiple pixels, thereby calculating the abundance value of the end member corresponding to the most central pixel, so that the calculation of the abundance value is more accurate. And the correlation matrix is ​​sequentially reduced in dimensionality through the principal component analysis algorithm and the common space pattern algorithm, thereby reducing the amount of calculation and improving the calculation speed and efficiency. Thereby solving the technical problem of inaccurate calculation of abundance value existing in the prior art.

[0166] Example 3

[0167] Figure 7 FIG. 7 shows an apparatus 700 for determining an abundance value according to this embodiment, and the apparatus 700 corresponds to the method according to the first aspect of embodiment 1. Figure 7 As shown, the device 700 includes: a processor 710; and a memory 720, which is connected to the processor 710 and is used to provide the processor 710 with instructions for processing the following processing steps: generating a correlation matrix for each pixel based on the reflection information correlation between each pixel of the satellite remote sensing reflection image and the corresponding neighborhood; performing dimensionality reduction processing on the correlation matrix using a principal cause analysis algorithm to generate a principal component matrix; and processing the principal component matrix using a common space pattern algorithm to determine the abundance value of the end member corresponding to the ground object type in each pixel.

[0168] Optionally, the operation of generating a correlation matrix of each pixel based on the reflection information correlation between each pixel of the satellite remote sensing reflection image and the corresponding neighborhood includes: selecting a target pixel from each pixel of the satellite remote sensing reflection image; calculating the reflection information correlation between the target pixel and the corresponding neighborhood based on the reflection information of the target pixel and the reflection information of the corresponding neighborhood; and generating a correlation matrix of the target pixel based on the reflection information of the pixel and the corresponding reflection information correlation.

[0169] Optionally, the operations of reducing the dimension of the correlation matrix using a principal cause analysis algorithm to generate a principal component matrix include: normalizing the correlation matrix to generate a normalized matrix; generating a correlation matrix based on the normalized matrix; and generating a principal component matrix based on a preset number of principal components.

[0170] Optionally, the operation of generating a principal component matrix according to a preset number of principal components includes: processing a correlation matrix according to the number of principal components to determine a unit eigenvector; and generating a principal component matrix according to the unit eigenvector and the correlation matrix.

[0171] Optionally, the operation of processing the principal component matrix using a common spatial pattern algorithm to determine the abundance value of the endmember corresponding to the land feature type in each pixel includes: processing the principal component matrix using a common spatial pattern algorithm corresponding to the land feature type to generate corresponding feature information; classifying the corresponding feature information respectively through a preset binary classification model to determine the land feature type probability of the endmember corresponding to the land feature type; and determining the abundance value of the endmember corresponding to the land feature type in each pixel according to the land feature type probability through a softmax classifier.

[0172] Optionally, the operation of processing the principal component matrix using a common space pattern algorithm to generate characteristic information includes: generating a correlation matrix based on the multi-channel information in the principal component matrix, wherein the correlation matrix is ​​used to indicate the correlation between the multi-channel information; and extracting features from the correlation matrix through a projection matrix to generate characteristic information.

[0173] Therefore, according to this embodiment, the computing device calculates the correlation between the pixel and its neighborhood according to the reflection information of each pixel, and generates a corresponding correlation matrix. After that, the computing device performs dimensionality reduction processing on the correlation matrix, and determines the abundance value of the end member corresponding to the ground object type in the pixel according to the correlation matrix generated by the reflection information of the pixel and its neighborhood after dimensionality reduction processing. Compared with the method in the prior art that determines the abundance value of the end member corresponding to the pixel only according to the reflection information of one pixel, the present technical solution associates the reflection information of multiple pixels, thereby calculating the abundance value of the end member corresponding to the most central pixel, so that the calculation of the abundance value is more accurate. And the correlation matrix is ​​sequentially reduced in dimensionality through the principal component analysis algorithm and the common space pattern algorithm, thereby reducing the amount of calculation and improving the calculation speed and efficiency. Thereby solving the technical problem of inaccurate calculation of abundance value existing in the prior art.

[0174] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0175] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0176] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0177] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0178] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0179] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0180] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for determining an abundance value, characterized in that: include: Generate a correlation matrix of each pixel according to the correlation of reflection information between each pixel of the satellite remote sensing reflection image and the corresponding neighborhood; Using a principal cause analysis algorithm to perform dimensionality reduction processing on the correlation matrix to generate a principal component matrix; as well as The principal component matrix is ​​processed using a common spatial pattern algorithm to determine the abundance value of the end member corresponding to the ground feature type in each pixel, where The operation of performing dimensionality reduction processing on the correlation matrix using the principal cause analysis algorithm to generate a principal component matrix includes: normalizing the correlation matrix to generate a standard matrix; generating a correlation matrix according to the standard matrix; and generating a principal component matrix according to a preset number of principal components, and wherein The operation of generating a principal component matrix according to a preset number of principal components includes: processing the correlation matrix according to the number of principal components to determine a unit eigenvector; and generating the principal component matrix according to the unit eigenvector and the correlation matrix.

2. The method according to claim 1, characterized in that The operation of generating a correlation matrix of each pixel according to the correlation of reflection information between each pixel of the satellite remote sensing reflection image and the corresponding neighborhood includes: Selecting a target pixel from each pixel of the satellite remote sensing reflection image; Calculating the reflection information correlation between the target pixel and the corresponding neighborhood according to the reflection information of the target pixel and the reflection information of the corresponding neighborhood; and A correlation matrix of the target pixel is generated according to the reflection information of the pixel and the corresponding reflection information correlation.

3. The method according to claim 1, characterized in that The operation of processing the principal component matrix by using a common spatial pattern algorithm to determine the abundance value of the end member corresponding to the ground feature type in each pixel includes: Processing the principal component matrix using the common space pattern algorithm corresponding to the ground feature type to generate corresponding feature information; Classifying the corresponding feature information respectively through a preset binary classification model to determine the ground feature type probability of the end member corresponding to the ground feature type; and The abundance value of the end member corresponding to the ground feature type in each pixel is determined according to the ground feature type probability by a softmax classifier.

4. The method according to claim 1, characterized in that: The operation of processing the principal component matrix by using the common space pattern algorithm to generate characteristic information includes: generating a correlation matrix according to the multi-channel information in the principal component matrix, wherein the correlation matrix is ​​used to indicate the correlation between the multi-channel information; and The feature information is generated by performing feature extraction on the correlation matrix through a projection matrix.

5. A device for determining abundance value, characterized in that: include: A correlation matrix generation module is used to generate a correlation matrix of each pixel according to the reflection information correlation between each pixel of the satellite remote sensing reflection image and the corresponding neighborhood; A principal component matrix generation module is used to perform dimensionality reduction processing on the correlation matrix using a principal cause analysis algorithm to generate a principal component matrix; as well as The abundance value determination module is used to process the principal component matrix using a common spatial pattern algorithm to determine the abundance value of the end member corresponding to the ground feature type in each pixel, wherein The operation of performing dimensionality reduction processing on the correlation matrix using the principal cause analysis algorithm to generate a principal component matrix includes: normalizing the correlation matrix to generate a standard matrix; generating a correlation matrix according to the standard matrix; and generating a principal component matrix according to a preset number of principal components, and wherein The operation of generating a principal component matrix according to a preset number of principal components includes: processing the correlation matrix according to the number of principal components to determine a unit eigenvector; and generating the principal component matrix according to the unit eigenvector and the correlation matrix.

6. A device for determining abundance value, characterized in that: include: processor; as well as A memory, connected to the processor, configured to provide the processor with instructions for processing the following processing steps: Generate a correlation matrix of each pixel according to the correlation of reflection information between each pixel of the satellite remote sensing reflection image and the corresponding neighborhood; Using a principal cause analysis algorithm to perform dimensionality reduction processing on the correlation matrix to generate a principal component matrix; as well as The principal component matrix is ​​processed using a common spatial pattern algorithm to determine the abundance value of the end member corresponding to the ground feature type in each pixel, where The operation of performing dimensionality reduction processing on the correlation matrix using the principal cause analysis algorithm to generate a principal component matrix includes: normalizing the correlation matrix to generate a standard matrix; generating a correlation matrix according to the standard matrix; and generating a principal component matrix according to a preset number of principal components, and wherein The operation of generating a principal component matrix according to a preset number of principal components includes: processing the correlation matrix according to the number of principal components to determine a unit eigenvector; and generating the principal component matrix according to the unit eigenvector and the correlation matrix.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to claim 1 are implemented.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to claim 1 are implemented.

Images