A remote sensing identification method for desert plants based on multi-source information fusion

Abstract

The invention discloses a remote sensing identification method for desert plants based on multi-source information fusion, comprising the following steps: step 1: obtaining initial remote sensing data A of deserts in a designated area, and collecting source data B of desert plants; by gridding the desert and obtaining source data B, in the remote sensing identification process, real-time remote sensing data C is compared with the source data of B, and plant growth coordinates can be quickly obtained, the error of C is initially reduced, and desert plants N can be quickly identified according to the coordinates, and before fusion, control points are selected based on B to N through image registration or correction, so as to increase the accuracy of fusion, and thus make remote sensing identification more accurate, and different plant distribution areas are calculated according to the coordinate distribution range of the fused image, so as to know the survival status of desert plants, and thus facilitate timely remediation, bringing convenience to desert control.

Description

Technical Field

[0001] The invention belongs to the technical field of remote sensing identification, and in particular relates to a desert plant remote sensing identification method based on multi-source information fusion. Background Art

[0002] Desert plants refer to plants that can survive in desert conditions; most desert plants are drought-resistant or salt-resistant plants; some store water in their roots, stems, and leaves; some have a large rhizome system that can reach the groundwater layer, block the soil, and prevent soil erosion; some have larger stems and leaves that can reduce wind speed and preserve sand.

[0003] The natural environment is changing, and the desert is a typical variable environment; the characteristics of the desert are drought and little rain, large temporal variation rates of temperature, humidity and precipitation, and high degree of heterogeneity in spatial distribution; such harsh and changeable habitat conditions are extremely unfavorable for the growth and survival of plants; limited precipitation directly affects the growth, reproduction, primary productivity of annual plants, and the spatial distribution and abundance of species. However, the long-term biological evolution process has also selected some plants that are adapted to survive in this habitat, and thus formed a sparse but very precious desert vegetation; according to the theory of evolutionary ecology, the reason why certain plants can persist in a specific habitat is that these plants have formed life history strategies that adapt to extreme arid environments through natural selection in evolution. Water is the main limiting factor affecting plant growth and survival in the desert. The adaptive characteristics of desert plants are all related to the utilization of water resources (mainly natural precipitation). Taking desert plants as the research object, analyzing the life history of desert plants, and studying their life history strategies can help us clarify the internal mechanisms of the formation and maintenance of biodiversity in desert plant communities and even the entire desert ecosystem. Combined with the analysis of the main limiting factors, it is easier to reveal the ecological adaptation strategies and evolutionary mechanisms of plants.

[0004] In reality, some desert plants will die as time and climate change. In order to better monitor the desert and carry out timely remedial management, remote sensing identification methods are usually used for monitoring. During the use of traditional remote sensing identification methods, due to climate change, the distribution of desert plants is uneven, and there are many locations with sparse plants. Compared with the entire desert, the sparser plants are often difficult to identify accurately, and thus the growth and distribution of desert plants cannot be accurately obtained, which brings more inconvenience to desert management. Therefore, a desert plant remote sensing identification method based on multi-source information fusion is needed. Summary of the invention

[0005] The technical problem to be solved by the present invention is to overcome the existing defects and provide a desert plant remote sensing identification method based on multi-source information fusion to solve the problem mentioned in the above background technology that sparse plants are often unable to be accurately identified, and thus the growth distribution of desert plants cannot be accurately obtained, which brings more inconvenience to desert control.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for remote sensing identification of desert plants based on multi-source information fusion, comprising the following steps:

[0007] Step 1: Obtain the initial remote sensing data A of the desert in the designated area and collect the source data B of the desert plants;

[0008] Step 2: Preprocess the data based on A and B;

[0009] Step 3: Perform remote sensing identification on desert plants in the designated area and obtain real-time remote sensing data C;

[0010] Step 4: Fusion of multi-source information based on C;

[0011] Step 5: Obtain desert plant growth distribution images based on multi-source information fusion.

[0012] Preferably, the step 1 comprises the following steps:

[0013] Step 1.1: Obtain the desert’s spectrum, texture, shape, and landform information in A;

[0014] Step 1.2: Obtain the spectrum, texture, shape, and species information of desert plants in B.

[0015] Preferably, step 2 comprises the following steps:

[0016] Step 2.1: Grid the information based on A;

[0017] Step 2.2: Perform multi-level segmentation based on B to obtain modules such as B1, B2, and B3, and obtain information such as spectra, textures, and types at different scales.

[0018] Preferably, step 3 comprises the following steps:

[0019] Step 3.1: Identify the desert plant N in A based on B, and obtain the coordinate range of N based on the grid information of A;

[0020] Step 3.2: Based on B, obtain the image spectrum, texture and other factors of desert plant N in C.

[0021] Preferably, step 4 comprises the following steps:

[0022] Step 4.1: Based on the real-time remote sensing data C, N is segmented into multiple levels to obtain modules such as N1, N2, and N3;

[0023] Step 4.2: align N based on B;

[0024] Step 4.3: According to the principle that B1 corresponds to N1, B2 corresponds to N2, and B3 corresponds to N3, multi-source information fusion is performed on N to obtain identification information M;

[0025] Step 4.4: Calculate the desert plant distribution area S according to the coordinate range of the identification information M.

[0026] Preferably, the step 2.1 comprises the following steps:

[0027] Step 2.1.1: Assume that when inputting into the computer, the coordinates of each known discrete point are (xi, yi), and now find the value of every ten grid points to be generated. Take point O' as an example, take the unknown point O' as the origin of a new plane coordinate x'O'y', that is, x'o=O, y'o=O, and the coordinates of other known discrete points are x'i=xi-xo, y'i=yi-yo; take O' as the center of the circle, and the density of discrete points as the radius. If the discrete points are sparse, the value is large, and if the discrete points are dense, the value is small, so as to ensure that there are enough discrete points in the circle; take a quadratic polynomial for fitting, and the value of point O' Zo is Zo=ax'·x'+bx'·y'+cy'·y'+dx'+ey'+f. There are 6 unknown coefficients in the formula, and at least 6 known point data are required to form 6 equations to solve. Here, only f is required, Zo=fo. If there are more than 6 discrete known points, the least squares method is used to calculate the f value;

[0028] Step 2.1.2: After the value of this grid point is calculated, move one point to the right and calculate the value of the next point in the same way. Move one by one until the end of this row and then move to the next row until the entire image is scanned.

[0029] Preferably, the step 4.2 comprises the following steps:

[0030] Step 4.2.1: Select control points based on B and N;

[0031] Step 4.2.2: Image registration or correction;

[0032] Step 4.2.3: Check the registration accuracy;

[0033] Step 4.2.4: If the registration requirements are not met, continue to encrypt the control points until the registration requirements are met.

[0034] Preferably, the step 4.3 comprises the following steps:

[0035] Step 4.3.1: Let LRTM, LGTM, and LBTM be the brightness values ​​of TM4, 3, and 2 bands respectively, LSPoT be the brightness value of the SPOT panchromatic band, and G be the weight function. Then the brightness values ​​L of the three new images are generated as follows:

[0036] LR complex=G·LspoT·LRTM / (LRTM+LGTM+LBTM),

[0037] LG complex=G·LspoT·LGTM / (LRTM+LGTM+LBTM),

[0038] LB complex=G·LspoT·LBTM / (LRTM+LGTM+LBTM);

[0039] Step 4.3.2: Assign different colors to the newly generated images LR, LG, and LB respectively, and synthesize them to generate a composite image.

[0040] Compared with the prior art, the present invention provides a method for remote sensing identification of desert plants based on multi-source information fusion, which has the following beneficial effects:

[0041] 1. The present invention grids the desert and obtains source data B. In the remote sensing identification process, the real-time remote sensing data C is compared with the source data of B, and the plant growth coordinates can be quickly obtained, which initially reduces the error of C, and then the desert plant N can be quickly identified according to the coordinates;

[0042] 2. The present invention obtains information such as spectra, textures and species of different scales by performing multi-level segmentation of desert plants N based on real-time remote sensing data C to obtain modules such as N1, N2 and N3, and performs multi-level segmentation based on B to obtain modules such as B1, B2 and B3, and performs multi-source information fusion on N according to the principle that B1 corresponds to N1, B2 corresponds to N2 and B3 corresponds to N3. Before fusion, control points are selected based on B for N through image registration or correction, thereby increasing the accuracy of fusion and making remote sensing identification more accurate.

[0043] 3. When the present invention is integrated, different modules identified are color composited to make the identified image clearer, and the distribution areas of different plants are calculated according to the coordinate distribution range of the integrated image, so as to know the survival status of desert plants, thereby facilitating timely remediation and bringing convenience to desert control. DETAILED DESCRIPTION

[0044] The described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] Embodiment 1

[0046] The present invention provides a technical solution: a method for remote sensing identification of desert plants based on multi-source information fusion, comprising the following steps:

[0047] Step 1: Obtain the initial remote sensing data A of the desert in the designated area and collect the source data B of the desert plants;

[0048] Step 2: Preprocess the data based on A and B;

[0049] Step 3: Perform remote sensing identification on desert plants in the designated area and obtain real-time remote sensing data C;

[0050] Step 4: Fusion of multi-source information based on C;

[0051] Step 5: Obtain desert plant growth distribution images based on multi-source information fusion.

[0052] The working principle and use process of the present invention are as follows: when in use, the time-phase spectrum, shape and texture of a certain desert module are collected through remote sensing identification, and then the plants in the desert are preliminarily identified. The identified spectrum, texture, shape, type and other information are used for source data B, and multi-source information fusion of N and B is performed based on real-time remote sensing data C. The survival status of desert plants can be known based on the fused image, which facilitates timely remediation and brings convenience to desert control.

[0053] Embodiment 2

[0054] The present invention provides a technical solution: a method for remote sensing identification of desert plants based on multi-source information fusion, comprising the following steps:

[0055] Step 1: Obtain the initial remote sensing data A of the desert in the designated area and collect the source data B of the desert plants;

[0056] Step 2: Preprocess the data based on A and B;

[0057] Step 3: Perform remote sensing identification on desert plants in the designated area and obtain real-time remote sensing data C;

[0058] Step 4: Fusion of multi-source information based on C;

[0059] Step 5: Obtain desert plant growth distribution images based on multi-source information fusion.

[0060] In the present invention, preferably, step 1 comprises the following steps:

[0061] Step 1.1: Obtain the desert’s spectrum, texture, shape, and landform information in A;

[0062] Step 1.2: Obtain the spectrum, texture, shape, and species information of desert plants in B.

[0063] In the present invention, preferably, step 2 comprises the following steps:

[0064] Step 2.1: Grid the information based on A;

[0065] Step 2.2: Perform multi-level segmentation based on B to obtain modules such as B1, B2, and B3, and obtain information such as spectra, textures, and types at different scales.

[0066] In the present invention, preferably, step 3 comprises the following steps:

[0067] Step 3.1: Identify the desert plant N in A based on B, and obtain the coordinate range of N based on the grid information of A;

[0068] Step 3.2: Based on B, obtain the image spectrum, texture and other factors of desert plant N in C.

[0069] In the present invention, preferably, step 4 comprises the following steps:

[0070] Step 4.1: Based on the real-time remote sensing data C, N is segmented into multiple levels to obtain modules such as N1, N2, and N3;

[0071] Step 4.2: align N based on B;

[0072] Step 4.3: According to the principle that B1 corresponds to N1, B2 corresponds to N2, and B3 corresponds to N3, multi-source information fusion is performed on N to obtain identification information M;

[0073] Step 4.4: Calculate the desert plant distribution area S according to the coordinate range of the identification information M.

[0074] In the present invention, preferably, step 2.1 comprises the following steps:

[0075] Step 2.1.1: Assume that when inputting into the computer, the coordinates of each known discrete point are (xi, yi), and now find the value of every ten grid points to be generated. Take point O' as an example, take the unknown point O' as the origin of a new plane coordinate x'O'y', that is, x'o=O, y'o=O, and the coordinates of other known discrete points are x'i=xi-xo, y'i=yi-yo; take O' as the center of the circle, and the density of discrete points as the radius. If the discrete points are sparse, the value is large, and if the discrete points are dense, the value is small, so as to ensure that there are enough discrete points in the circle; take a quadratic polynomial for fitting, and the value of point O' Zo is Zo=ax'·x'+bx'·y'+cy'·y'+dx'+ey'+f. There are 6 unknown coefficients in the formula, and at least 6 known point data are required to form 6 equations to solve. Here, only f is required, Zo=fo. If there are more than 6 discrete known points, the least squares method is used to calculate the f value;

[0076] Step 2.1.2: After the value of this grid point is calculated, move one point to the right and calculate the value of the next point in the same way. Move one by one until the end of this row and then move to the next row until the entire image is scanned.

[0077] The working principle and use process of the present invention are as follows: when in use, the time spectrum, shape and texture of a certain desert module are collected through remote sensing identification, and the desert is gridded. When generating grid data, it is assumed that when input into the computer, the coordinates of each known discrete point are (xi, yi). Now the value of every ten grid points to be generated is calculated. Taking point O' as an example, the point to be determined O' is taken as the origin of a new plane coordinate x'O'y', that is, x'o=O, y'o=O, and the coordinates of other known discrete points are x'i=xi-xo, y'i=yi-yo; take O' as the center of the circle, the density of discrete points as the radius, and the value of discrete points is sparse. The circle is large, the discrete points are dense, and the value is small to ensure that there are enough discrete points in the circle; a quadratic polynomial is used for fitting, and the value of point O′, Zo, is Zo=ax′·x′+bx′·y′+cy′·y′+dx′+ey′+f. There are 6 unknown coefficients in the formula, and at least 6 known point data are required to form 6 equations to solve. Here, only f, Zo=fo, is required. If there are more than 6 discrete known points, the least squares method is used to calculate the f value; after the value of this grid point is solved, move one point to the right and calculate the value of the next point in the same way, move one by one until the end of this line and then proceed to the next line until the entire image is scanned; then The plants in the desert are preliminarily identified, and the identified spectrum, texture, shape, and type information are used for source data B, and the coordinates of the plant's growth range are obtained based on the grid data. With the change of time and climate, the plants in the desert will also die, which will lead to the diversity of monitoring information in the later monitoring process. In the remote sensing identification process, the real-time remote sensing data C can be compared with the source data of B to quickly obtain the plant growth coordinates, which initially reduces the error of C, and then the desert plant N can be quickly identified according to the coordinates. Because the growth of the plant has changed, it is necessary to The image spectrum, texture and other factors of N are obtained. In order to increase the accuracy of recognition, N is segmented into multiple levels based on the real-time remote sensing data C to obtain modules such as N1, N2, and N3, and is segmented into multiple levels based on B to obtain modules such as B1, B2, and B3. Information such as spectra, textures, and types of N are obtained at different scales. Multi-source information fusion is performed on N according to the principle that B1 corresponds to N1, B2 corresponds to N2, and B3 corresponds to N3. The distribution areas of different plants are calculated according to the coordinate distribution range of the fused image. The survival status of desert plants can be known, which is convenient for timely remediation and brings convenience to desertification control.

[0078] Embodiment 3

[0079] The present invention provides a technical solution: a method for remote sensing identification of desert plants based on multi-source information fusion, comprising the following steps:

[0080] Step 1: Obtain the initial remote sensing data A of the desert in the designated area and collect the source data B of the desert plants;

[0081] Step 2: Preprocess the data based on A and B;

[0082] Step 3: Perform remote sensing identification on desert plants in the designated area and obtain real-time remote sensing data C;

[0083] Step 4: Fusion of multi-source information based on C;

[0084] Step 5: Obtain desert plant growth distribution images based on multi-source information fusion.

[0085] In the present invention, preferably, step 1 comprises the following steps:

[0086] Step 1.1: Obtain the desert’s spectrum, texture, shape, and landform information in A;

[0087] Step 1.2: Obtain the spectrum, texture, shape, and species information of desert plants in B.

[0088] In the present invention, preferably, step 2 comprises the following steps:

[0089] Step 2.1: Grid the information based on A;

[0090] Step 2.2: Perform multi-level segmentation based on B to obtain modules such as B1, B2, and B3, and obtain information such as spectra, textures, and types at different scales.

[0091] In the present invention, preferably, step 3 comprises the following steps:

[0092] Step 3.1: Identify the desert plant N in A based on B, and obtain the coordinate range of N based on the grid information of A;

[0093] Step 3.2: Based on B, obtain the image spectrum, texture and other factors of desert plant N in C.

[0094] In the present invention, preferably, step 4 comprises the following steps:

[0095] Step 4.1: Based on the real-time remote sensing data C, N is segmented into multiple levels to obtain modules such as N1, N2, and N3;

[0096] Step 4.2: align N based on B;

[0097] Step 4.3: According to the principle that B1 corresponds to N1, B2 corresponds to N2, and B3 corresponds to N3, multi-source information fusion is performed on N to obtain identification information M;

[0098] Step 4.4: Calculate the desert plant distribution area S according to the coordinate range of the identification information M.

[0099] In the present invention, preferably, step 2.1 comprises the following steps:

[0100] Step 2.1.1: Assume that when inputting into the computer, the coordinates of each known discrete point are (xi, yi), and now find the value of every ten grid points to be generated. Take point O' as an example, take the unknown point O' as the origin of a new plane coordinate x'O'y', that is, x'o=O, y'o=O, and the coordinates of other known discrete points are x'i=xi-xo, y'i=yi-yo; take O' as the center of the circle, and the density of discrete points as the radius. If the discrete points are sparse, the value is large, and if the discrete points are dense, the value is small, so as to ensure that there are enough discrete points in the circle; take a quadratic polynomial for fitting, and the value of point O' Zo is Zo=ax'·x'+bx'·y'+cy'·y'+dx'+ey'+f. There are 6 unknown coefficients in the formula, and at least 6 known point data are required to form 6 equations to solve. Here, only f is required, Zo=fo. If there are more than 6 discrete known points, the least squares method is used to calculate the f value;

[0101] Step 2.1.2: After the value of this grid point is calculated, move one point to the right and calculate the value of the next point in the same way. Move one by one until the end of this row and then move to the next row until the entire image is scanned.

[0102] In the present invention, preferably, step 4.2 comprises the following steps:

[0103] Step 4.2.1: Select control points based on B and N;

[0104] Step 4.2.2: Image registration or correction;

[0105] Step 4.2.3: Check the registration accuracy;

[0106] Step 4.2.4: If the registration requirements are not met, continue to encrypt the control points until the registration requirements are met.

[0107] In the present invention, preferably, step 4.3 comprises the following steps:

[0108] Step 4.3.1: Let LRTM, LGTM, and LBTM be the brightness values ​​of TM4, 3, and 2 bands respectively, LSPoT be the brightness value of the SPOT panchromatic band, and G be the weight function. Then the brightness values ​​L of the three new images are generated as follows:

[0109] LR complex=G·LspoT·LRTM / (LRTM+LGTM+LBTM),

[0110] LG complex=G·LspoT·LGTM / (LRTM+LGTM+LBTM),

[0111] LB complex=G·LspoT·LBTM / (LRTM+LGTM+LBTM);

[0112] Step 4.3.2: Assign different colors to the newly generated images LR, LG, and LB respectively, and synthesize them to generate a composite image.

[0113] The working principle and use process of the present invention are as follows: when in use, the time phase spectrum, shape and texture of a certain desert module are collected through remote sensing identification, and the desert is gridded. When generating grid data, it is assumed that the coordinates of each known discrete point when input into the computer are (xi, yi), and the value of every ten grid points to be generated is now calculated. Taking point O' as an example, the point to be determined O' is taken as the origin of a new plane coordinate x'O'y', that is, x'o=O, y'o=O, and the coordinates of other known discrete points are x'i=xi-xo, y'i=yi-yo; O' is taken as the center of the circle, and the density of discrete points is taken as the radius. If the discrete points are sparse, the value is large, and if the discrete points are dense, the value is small, so as to ensure that there are enough discrete points in the circle; take two A polynomial of degree is fitted, and the value of point O′ is Zo=ax′·x′+bx′·y′+cy′·y′+dx′+ey′+f. There are 6 unknown coefficients in the formula, and at least 6 known points are required to form 6 equations to solve. Here, only f is required, Zo=fo. If there are more than 6 discrete known points, the least squares method is used to calculate the f value. After the value of this grid point is solved, move one point to the right and calculate the value of the next point in the same way. Move one by one until the end of this line and then proceed to the next line until the entire map is scanned. Then, the plants in the desert are preliminarily identified, and the identified spectrum, texture, shape, and type information are used for the source data B, and the gridded data are used for The coordinates of the plant's growth range are obtained. With the change of time and climate, the plants in the desert will also die, which will lead to the diversity of monitoring information in the later monitoring process. In the remote sensing identification process, the real-time remote sensing data C can be compared with the source data of B to quickly obtain the plant growth coordinates, which initially reduces the error of C. Then, the desert plant N can be quickly identified according to the coordinates. Due to the change in plant growth, it is necessary to obtain factors such as N's image spectrum and texture. In order to increase the accuracy of identification, N is multi-level segmented based on the real-time remote sensing data C to obtain modules such as N1, N2, and N3, and B is multi-level segmented based on B to obtain B1, B 2. Modules B3 and so on, obtain information such as spectra, textures and types of different scales, and fuse N with multi-source information according to the principle that B1 corresponds to N1, B2 corresponds to N2, and B3 corresponds to N3. Before fusion, control points are selected based on B to N through image registration or correction, and then the registration accuracy is checked to see if it meets the registration requirements. If it does not meet the registration requirements, the control points are encrypted until the registration requirements are met. During fusion, the identified different modules are color composited. During composite, LRTM, LGTM, and LBTM are respectively the brightness values ​​of TM4, 3, and 2 bands, LSPoT is the brightness value of the SPOT panchromatic band, and G is the weight function. The brightness values ​​L of the three new images are generated as follows:

[0114] LR complex=G·LspoT·LRTM / (LRTM+LGTM+LBTM),

[0115] LG complex=G·LspoT·LGTM / (LRTM+LGTM+LBTM),

[0116] LB complex = G·LspoT·LBTM / (LRTM+LGTM+LBTM); assign different colors to the newly generated images LR complex, LG complex, and LB complex, respectively, and generate a composite image after synthesis. According to the coordinate distribution range of the fused image, the distribution areas of different plants are calculated to know the survival status of desert plants, which is convenient for timely remediation and brings convenience to desert control.

[0117] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for remote sensing identification of desert plants based on multi-source information fusion, characterized in that: The following steps are involved: Step 1: Obtain the initial remote sensing data A of the desert in the designated area and collect the source data B of the desert plants; Step 2: Preprocess the data based on A and B; Step 3: Perform remote sensing identification on desert plants in the designated area and obtain real-time remote sensing data C; Step 4: Fusion of multi-source information based on C; Step 5: Obtain desert plant growth distribution images based on multi-source information fusion; The step 1 comprises the following steps: Step 1.1: Obtain the spectrum, texture, shape and geomorphic information of the desert in A; Step 1.2: Obtain the spectrum, texture, shape, and species information of desert plants in B; The step 2 comprises the following steps: Step 2.1: Grid the information based on A; Step 2.2: Perform multi-level segmentation based on B to obtain B1, B2, and B3 modules to obtain spectrum, texture, and type information at different scales; The step 3 comprises the following steps: Step 3.1: Identify the desert plant N in A based on B, and obtain the coordinate range of N based on the grid information of A; Step 3.2: Based on B, obtain the image spectrum and texture factor of desert plant N in C; The step 4 comprises the following steps: Step 4.1: Based on the real-time remote sensing data C, perform multi-level segmentation on N to obtain N1, N2, and N3 modules; Step 4.2: align N based on B; Step 4.3: According to the principle that B1 corresponds to N1, B2 corresponds to N2, and B3 corresponds to N3, multi-source information fusion is performed on N to obtain identification information M; Step 4.4: Calculate the desert plant distribution area S according to the coordinate range of the identification information M.

2. The method for remote sensing identification of desert plants based on multi-source information fusion according to claim 1 is characterized in that: The step 2.1 comprises the following steps: Step 2.1.1: Assume that when inputting into the computer, the coordinates of each known discrete point are (xi, yi), and now find the value of every ten grid points to be generated. Take point 0' as an example, take the unknown point 0' as a new plane coordinate x'0'y' origin, that is, x'o=0, y'o=0, and the coordinates of other known discrete points are x'i=xi-xo, y'i=yi-yo; take 0' as the center of the circle, and the density of discrete points as the radius. If the discrete points are sparse, the value is large, and if the discrete points are dense, the value is small, so as to ensure that there are enough discrete points in the circle; take a quadratic polynomial for fitting, and the value of point 0' Zo is Zo=ax'·x'+bx'·y'+cy'·y'+dx'+ey'+f. There are 6 unknown coefficients in the formula, and at least 6 known point data are required to form 6 equations to solve. Here, only f is required, Zo=fo. If there are more than 6 discrete known points, the least squares method is used to calculate the f value; Step 2.1.2: After the value of this grid point is calculated, move one point to the right and calculate the value of the next point in the same way. Move one by one until the end of this row and then move to the next row until the entire image is scanned.

3. The method for remote sensing identification of desert plants based on multi-source information fusion according to claim 2 is characterized in that: The step 4.2 comprises the following steps: Step 4.2.1: Select control points based on B and N; Step 4.2.2: Image registration or correction; Step 4.2.3: Check the registration accuracy; Step 4.2.4: If the registration requirements are not met, continue to encrypt the control points until the registration requirements are met.

4. The method for remote sensing identification of desert plants based on multi-source information fusion according to claim 3 is characterized in that: The step 4.3 comprises the following steps: Step 4.3.1: Let LRTM, LGTM, and LBTM be the brightness values ​​of TM4, 3, and 2 bands respectively, LSPoT be the brightness value of SPOT panchromatic band, and G be the weight function. Then the brightness values ​​of the three new images L complex are generated as follows: LR complex = G·LspoT·LRTM / (LRTM+LGTM+LBTM). LG complex=G·LspoT·LGTM / (LRTM+LGTM+LBTM), LB complex=G·LspoT·LBTM / (LRTM+LGTM+LBTM); Step 4.3.2: Assign different colors to the newly generated images LR, LG, and LB respectively, and synthesize them to generate a composite image.