A cultivated land planting purpose change monitoring processing method based on hyperspectral remote sensing

Abstract

The application discloses a cultivated land planting purpose change monitoring processing method based on hyperspectral remote sensing, wherein the method comprises detecting the membership of a certain crop in a certain piece of permanent basic farmland in sequence according to a crop planting detection index, and if the membership of a certain crop is greater than or equal to the planting condition membership, the current planted crop is output and reported, otherwise, the membership of different farmland uncultivated states is detected. The application can meet the conventional monitoring processing of the cultivated land planting purpose change of the permanent basic farmland, thereby relieving the long image processing time and slow response speed of the current technology after the fusion of the panchromatic image and the hyperspectral image, and effectively reducing the calculation cost.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of image data processing, and particularly relates to a cultivated land planting purpose change monitoring processing method based on hyperspectral remote sensing. BACKGROUND

[0002] The cultivated land planting purpose change monitoring can accurately and quickly extract the spatial and temporal distribution information of agricultural land and the detailed conditions of cultivated land planting, and provide strong technical support for extracting the use conditions of regional agricultural land, optimizing the configuration and utilization of agricultural land, adjusting the planting structure of crops and guaranteeing food safety, is conducive to realizing the rapid and accurate supervision of the utilization conditions of agricultural land, is conducive to implementing the most strict cultivated land protection system, and has important significance for firmly holding the red line of cultivated land.

[0003] Hyper-Spectral Imaging (HSI) technology can acquire hundreds of continuous characteristic wavelength images in the same time and the same scene, and has a wide range of applications in the field of computer vision and remote sensing, such as change detection, target recognition and segmentation, environmental monitoring and agricultural analysis. However, due to the physical constraints of the design of optical sensors, there is a trade-off between spatial resolution and spectral resolution in this imaging modality. Specifically, the output of the imaging system carries a large number of characteristic wavelengths but damages the spatial resolution or high spatial resolution but has few spectral bands, such as multi-spectral imaging (MSI) or panchromatic (PAN) images. In order to obtain high-resolution HSI, HSI fusion technology fuses HSI and MSI (or PAN) in the same area, thereby attracting extensive research attention. Hyper-spectral remote sensing image is a kind of three-dimensional image containing rich spatial, spectral and other information formed by using hyper-spectral imaging system to detect ground object information based on the characteristics of electromagnetic wave reflection. Different ground objects have different spectral absorption and reflection characteristics, so the hyper-spectral imaging system can obtain the reflection signal of the ground object by using this characteristic, and then convert the reflection signal into the spectral curve of the ground object, and the continuous spectral curve presents the target ground object information. The reflection signal is dispersed into different wavelength spectral signals by optical equipment, so that several peaks and troughs are formed in the spectral curve. Finally, the spectral curve is analyzed by some specific algorithms, and many useful information for classification is extracted. The key task of image fusion technology in the field of space spectrum is to integrate three types of data combinations: panchromatic image and multi-spectral image, panchromatic image and hyper-spectral image, and multi-spectral image and hyper-spectral image. The process is essentially to combine the image data with unique advantages, i.e. the high spatial resolution of panchromatic image, the multi-band information of multi-spectral image, and the high spectral resolution of hyper-spectral image, through precise fusion algorithm. Whether it is to integrate high-detail panchromatic features into multi-band multi-spectral or hyper-spectral images, or to enhance the spatial performance of multi-spectral and hyper-spectral images, the core is to create a new type of image that not only maintains high spatial resolution but also contains rich spectral details, thereby greatly improving the accuracy and application range of remote sensing image analysis. Before image fusion, the image usually needs to be processed for radiation calibration and atmospheric correction, etc. to ensure that the final fused image can be accurately registered and show the information characteristics of each band.

[0004] The uncertainty in the division of things can be studied by fuzzy mathematics. Fuzzy pattern recognition is a recognition method based on the abstraction and description of fuzzy phenomena by fuzzy mathematical methods, which can simplify the structure of the recognition system, more widely and deeply simulate the thinking process of the human brain, and thus more effectively classify and recognize the target. Compared with other methods, fuzzy pattern recognition is good at dealing with the identification type of given objects in complex systems, can solve many problems that traditional evaluation methods cannot solve, and can avoid the subjective influence of human factors on the evaluation results in the calculation process, making the results of classification decision more accurate.

[0005] Considering that hyperspectral remote sensing technology has the characteristics of rapidness, wide coverage and cost-effectiveness, the cultivated land is monitored by analyzing and classifying the surface features of the hyperspectral remote sensing data in the monitoring area, and the satellite image and the hyperspectral image of the unmanned aerial vehicle are used as the core data source, the artificial intelligence recognition technology is used to extract and analyze the change of the cultivated land use, and the related vector data is matched and analyzed, the monitoring results are output, the change reasons are analyzed, the supervision and management of the illegal change of the cultivated land use and other behaviors of illegally occupying the cultivated land are effectively assisted, and the rapid monitoring of a large range of cultivated land is realized. However, the existing land type classification recognition method has the problems of large amount of hyperspectral image data processing in the process of small plot conventional detection, and the problems of position difference and insufficient registration of the image after the fusion of the panchromatic image and the hyperspectral image. SUMMARY

[0006] In order to solve the problems in the prior art, the purpose of the present application is to provide a cultivated land planting purpose change monitoring processing method based on hyperspectral remote sensing, which can analyze and judge the current planting purpose by referring to the planting crop detection index, detecting the planting condition membership degree of different crops in turn with the aid of the panchromatic image and the hyperspectral image, and detecting the membership condition of different uncultivated states when monitoring and processing the planting purpose change of a piece of cultivated land, so as to meet the conventional monitoring and processing of the planting purpose change of the cultivated land contained in the cultivated land database of a certain area, and to alleviate the long processing time and slow response speed of the image after the fusion of the panchromatic image and the hyperspectral image, and to effectively reduce the calculation cost.

[0007] The present application adopts the following technical scheme: a cultivated land planting purpose change monitoring processing method based on hyperspectral remote sensing, comprising the following steps:

[0008] Step S1, establishing a planting crop detection index of the cultivated land in the cultivated land database of a certain area;

[0009] Further, the specific method of step S1 is:

[0010] Let Φ be a farmland database for a certain region containing several farmland plots, f be the identifier of the farmland plot, and f∈Φ. Let i be the identifier of the crop, and I be the set of crop types for the region. For farmland f, the crop r can be registered on the farmland use change monitoring date d. f,d Where d satisfies d∈Θ, and Θ is the set of dates for which changes in cultivated land use were previously monitored in the region, then a set of crops Υ for cultivated land f can be constructed. f For {r f,d |d∈Θ}, where | is the condition symbol; according to fuzzy mathematics, the universe of discourse U={i|i∈I} can be set as follows: and Let U be the fuzzy set of "crops" on the domain U with respect to cultivated land f and the fuzzy set of "crops" on the region, respectively. These can be represented using Zadeh notation as follows:

[0011]

[0012] When using Zadeh notation, the right-hand side of the formula is not a summation of fractions; it is merely a notation. The denominator represents the element of the universe of discourse U, and the numerator represents the membership degree of the corresponding element. `size` is a function that gives the number of elements in the input set, and `true` is a function that determines whether the condition is true or false. If true, the function outputs 1; otherwise, it outputs 0.

[0013] If there exists a pre-constructed set of already planted crops Υ for cultivated land f. f Then the f corresponding to According to element i Elements i are arranged in descending order of membership degree. If elements have the same membership degree, they are randomly arranged. The set of these elements is then used as the crop detection index L corresponding to cultivated land f. f If a set of already planted crops (Υ) has not been constructed for cultivated land (f); f Then the corresponding region can be According to the membership of element i The elements i are arranged in descending order of their degree of importance, and then the set of the above elements is used as the detection index L of the crop planted on the cultivated land f. f If elements have the same membership degree, the corresponding elements are randomly arranged.

[0014] Step S2: Determine the set of pixel gray values ​​of different crops at different growth stages in the crop type set of the region, or the set of reflectance values ​​of different crops at different characteristic wavelengths in the hyperspectral image, as well as the set of pixel gray values ​​of different uncultivated states in the panchromatic image and the set of reflectance values ​​of different uncultivated states at different characteristic wavelengths in the hyperspectral image, in the uncultivated state set of the region.

[0015] Furthermore, the specific method for step S2 is as follows:

[0016] The growth process of crop i can be divided into l i There are several growth stages, of which l i Let j be the number of growth stages that crop i can be divided into, and let j be the identifier of the growth stage, where 1 ≤ j ≤ l. i For crop i in growth stage j within the region, its panchromatic and hyperspectral images can be acquired simultaneously for the same scene, and for any crop i in growth stage j at time m and scene n, the panchromatic image P... i,j (m,n) and hyperspectral image Q i,j (m,n) can be fused to generate image R i,j (m,n), where n is the scene number for acquiring panchromatic and hyperspectral images in this region, and m is the time when acquiring panchromatic and hyperspectral images in this region; let X i,j For a panchromatic image dataset of crop i at growth stage j within this region, the X... i,j The panchromatic image P contained within i,j After grayscale conversion of (m,n), the P can be given. i,j The gray value α of any pixel in (m,n) i,j Let (m,n,k) be the set of pixel numbers in the panchromatic image, and K0 be the set of all pixel numbers in the panchromatic image. Considering that in mathematics, methods such as interval methods, set methods, and number lines are often used to represent the range of values ​​for ease of understanding and calculation, the set method can be used to determine the set A of pixel grayscale values ​​of crop i at growth stage j in the panchromatic image. i,j for:

[0017]

[0018] Where ∪ is the union symbol, N is the set of scenes in which panchromatic and hyperspectral images are acquired within the region, and M is the time range in which panchromatic and hyperspectral images are acquired within the region;

[0019] Let Y be the hyperspectral image dataset of crop i at growth stage j in this region. i,j The hyperspectral image dataset Y of crop i at growth stage j can be used. i,j The specific set of wavelengths exhibiting spectral characteristics can be determined as Ω. i,j Then through the Y i,j The hyperspectral image Q included i,j Select the region of interest r in (m,n) to extract the reflectance β at the characteristic wavelength h. i,j (m,n,h,r), where h is the identifier of the characteristic wavelength in the hyperspectral image, and for crop i, h∈H should be satisfied.i , H i is the set of all feature wavelengths for feature extraction in the hyperspectral image of crop i, r is the number of the region of interest in the hyperspectral image, R0is the set of all numbers of the regions of interest in the hyperspectral image, and r∈R0, thus the set of values B of the reflectivity of crop i at growth stage j on feature wavelength h in the hyperspectral image can be determined i,j,h is:

[0020]

[0021] Set W as the set of uncultivated states in the region, w is the uncultivated state in W, and the panchromatic image u of the uncultivated state w in the region at time m and scene n can be collected w (m, n) and the hyperspectral image v w (m, n), set the panchromatic image dataset and the hyperspectral image dataset of the uncultivated state w in the basic farmland protection zone as U w and V w , the panchromatic image u w (m, n) contained in U w can be given the gray value φ w (m, n, k) after being grayed, thus the set of values p of the uncultivated state w on the pixel gray value in the panchromatic image can be determined w is:

[0022]

[0023] Select the region of interest r in the hyperspectral image v w (m, n) contained in V w to extract the reflectivity on feature wavelength h wherein for the uncultivated state w, h∈H w , H w is the set of feature wavelengths for feature extraction in the hyperspectral image of the uncultivated state, r∈R0, thus the set of values s of the reflectivity of the uncultivated state w on feature wavelength h in the hyperspectral image can be determined w,h is:

[0024]

[0025] Step S3, for a piece of cultivated land, the planting condition membership degree corresponding to the crop can be detected in sequence according to the detection index of the planting crop corresponding to the cultivated land given in step S1, if the membership condition corresponding to a crop satisfies greater than or equal to the planting condition membership degree, the crop is output as the current planted crop of the cultivated land, otherwise the crop possibly planted in the cultivated land is input to step S4;

[0026] Further, the specific method of step S3 is:

[0027] According to the value set A of the pixel gray scale of crop i in growth stage j in the full-color image in step S2 i,j The pixel gray scale belongs to the fuzzy set of "crop i in growth stage j" The membership function of the fuzzy set The membership function of the fuzzy set Can be defined as:

[0028]

[0029] Wherein, x is the variable of the input function, Ψ is the set of the input function, Closest(Ψ, x) is a function of finding the element in the set Ψ closest to z, above(Ψ, x) is a function of finding the element in the set Ψ greater than z, below(Ψ, x) is a function of finding the element in the set Ψ smaller than z;

[0030] According to the value set B of the reflectance of crop i in growth stage j at feature wavelength h in the hyperspectral image in step S2 i,j,h The reflectance of the region of interest belongs to the fuzzy set of "crop i in growth stage j at feature wavelength h" The membership function of the fuzzy set The membership function of the fuzzy set Can be defined as:

[0031]

[0032] Wherein, biggest is a function of giving the element with the largest value in the input set, smallest is a function of giving the element with the smallest value in the input set;

[0033] For the farmland f, set the full-color image P f (m, n) and the hyperspectral image Q f (m, n) collected when the farmland f is monitored for change in use for planting at time m and scene n, wherein the scene n should be set within the geometric range of the f, and the pixel point number set K f (m, n) is set for random sampling in the P f (m, n), the number set R f of the region of interest within the geometric figure composed of the pixel points in the K f , then the membership status ZWLSZK(f, i) of the planting of the f for crop i can be defined as:

[0034]

[0035] Wherein, αf (m,n,k) is the panchromatic image P f (m,n) is the K f corresponding to the pixel k, β f (m,n,h,r) is the hyperspectral image Q f (m,n) is the R f reflectivity of the region of interest r at the characteristic wavelength h, and for the crop i should satisfy h∈H i max is the function of taking the maximum value;

[0036] Set γ as the crop membership threshold, thereby the planting crop detection index L f corresponding to the cultivated land f in step S1 is detected in turn f The membership status ZWLSZK(f,i) corresponding to the crop i contained in the L f is output and reported; if the membership status ZWLSZK(f,i) corresponding to the crop i is greater than or equal to the planting status membership degree γ, then the crop i is taken as the current planting crop κ of the cultivated land f f is output and reported; if the membership status ZWLSZK(f,i) corresponding to all crops i in the L is taken as the membership status τ of the current planting status of the cultivated land f f , the crop i corresponding to is taken as the current possible planting crop of the cultivated land f is input into step S4;

[0037] Step S4, the maximum value of the membership status of the different uncultivated states of the cultivated land in step S3 is compared with the crop membership threshold in step S3, if greater than or equal to, then the corresponding uncultivated state is output and reported, otherwise the maximum value of the membership status is compared with the membership status of the current planting status of the cultivated land in step S3, the uncultivated state or the current planting status corresponding to the larger value is output and reported;

[0038] Further, the specific method of step S4 is:

[0039] According to the value set ρ of the pixel gray scale of the uncultivated state w in step S2 w The fuzzy set of the pixel gray scale belonging to the uncultivated state w can be set The membership function of the The membership function of the

[0040]

[0041] ​According to step S2, the set of values ​​σ for the reflectance of the hyperspectral image at the characteristic wavelength h in the uncultivated state w is given by the following values. w,h You can set the fuzzy set of reflectance of the region of interest to belong to the "uncultivated state w". The membership function It can be defined as:

[0042]

[0043] Define the membership status WGZLSZK(f,w) of the uncultivated land f in step S3 as follows:

[0044]

[0045] Based on step S2, W provides WGZLSZK(f,w) corresponding to different values ​​of w, and the maximum value among them is determined. and the corresponding uncultivated state w; the aforementioned Compare with the crop membership threshold γ described in step S3, if Then The corresponding w represents the current uncultivated state of arable land f. f Perform output reporting; if Then The affiliation status τ of the current planting status of cultivated land f as described in step S3 f If a comparison is made, Then the crops that can currently be planted on the cultivated land f mentioned in step S3 will be... The affiliation status of cultivated land f under its current planting status τ f Output and report, otherwise... The corresponding w represents the current possible uncultivated state of arable land f. and will The affiliation status ζ of the currently uncultivated state of the cultivated land f f And will determine the current possible uncultivated state of arable land. The affiliation status of arable land f in its current uncultivated state ζ f Output and report. Attached Figure Description

[0046] The accompanying drawings are defined as part of this invention and are used to provide a further understanding of the invention.

[0047] Figure 1 This is a flowchart illustrating a method for monitoring and processing changes in cultivated land use based on hyperspectral remote sensing, as described in an embodiment of the present invention.

[0048] Figure 2 This is a partial remote sensing image of a certain area that has been divided into different cultivated lands in an embodiment of the present invention;

[0049] Figure 3 a membership function corresponding to a fuzzy set of "rice (1) in tillering stage (2)" to which the pixel gray scale in the embodiment of the present application belongs;

[0050] Figure 4 a membership function corresponding to a fuzzy set of "rice (1) in tillering stage (2) about characteristic wavelengths 670nm / 780nm" to which the reflectivity of the region of interest in the embodiment of the present application belongs;

[0051] Figure 5 a gray scale value corresponding to a pixel point which is randomly sampled after the panchromatic image is collected in the embodiment of the present application;

[0052] Figure 6 reflectivity of the region of interest in a geometric figure composed of pixel points on different characteristic wavelengths after the hyperspectral image is collected in the embodiment of the present application. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the present application are clearly and completely described below in combination with the drawings and specific embodiments. The described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other equivalent or obvious modified embodiments obtained by those skilled in the art without creative effort fall within the scope of protection of the present application.

[0054] As shown in Figure 1 The present application provides a cultivated land planting purpose change monitoring processing method based on hyperspectral remote sensing, which comprises the following steps:

[0055] Step S1, form a planting crop detection index of cultivated land in a cultivated land database of a certain region;

[0056] Step S2, determine a value set of pixel gray scale in a panchromatic image or a value set of reflectivity on different characteristic wavelengths of a hyperspectral image of different crops in different growth stages in a crop type set of the region, and a value set of pixel gray scale in the panchromatic image of different uncultivated states and a value set of reflectivity on different characteristic wavelengths of the hyperspectral image of different uncultivated states in the uncultivated state set of the region;

[0057] Step S3, for a certain piece of cultivated land, the planting condition membership degree of the crop corresponding to the cultivated land is detected in turn according to the planting crop detection index corresponding to the cultivated land given in step S1, if the membership condition corresponding to a certain crop satisfies the planting condition membership degree, the crop currently planted in the cultivated land is output and reported, otherwise the crop currently possibly planted in the cultivated land is input to step S4;

[0058] Step S4, comparing the maximum of the membership conditions of the different uncultivated states of the farmland in step S3 with the crop membership threshold in step S3, if greater than or equal to, the corresponding uncultivated state is output and reported, otherwise, comparing the maximum of the membership conditions with the membership conditions of the current planting state of the farmland in step S3, the uncultivated state or the current planting state corresponding to the larger value is output and reported;

[0059] More specifically, the S1 is provided for a certain region of farmland database, any piece of farmland crop type can be registered and the constructed for a piece of farmland has been planted crop set; using fuzzy mathematics, the farmland related crop type set for the region is set as the domain, and the specific operation of setting the fuzzy set of "crop" on a piece of farmland or the fuzzy set of "crop" on the region in the domain to formulate the corresponding planting crop detection index of the farmland is as follows:

[0060] Setting the farmland database Φ for a certain region contains several pieces of farmland, f is the identification of the farmland, and f∈Φ, setting i is the identification of the crop, I is the crop type set of the region, for the farmland f, the crop r f,d , wherein d can satisfy d∈Θ, Θ is the date set of the farmland planting use change monitoring of the basic farmland protection zone, then the constructed for the farmland f has been planted crop set Υ f is {r f,d |d∈Θ}, wherein | is the conditional symbol;

[0061] According to fuzzy mathematics, the domain U={i|i∈I} can be set, and is the fuzzy set of "crop" on the farmland f in the domain U, which can be expressed by Zadeh notation as:

[0062]

[0063] Wherein, when using Zadeh notation, the right end of the formula is not a fractional sum, it is only a symbol, the denominator position is the element of the domain U, and the numerator position is the membership degree of the corresponding element; size is a function of giving the number of elements in the input set, true is a function of judging whether it is true or not, if true, output 1, if not, output 0;

[0064] Setting is the fuzzy set of "crop" on the region in the domain U, which can be expressed by Zadeh notation as:

[0065]

[0066] If there is a constructed has been planted crop set Υ fThe f corresponding to the region can be determined by the following steps: The elements i are arranged in descending order according to the membership degree of the element i, and if there is the same membership degree, the corresponding elements are arranged randomly, and then the set of the above elements is taken as the planting crop detection index corresponding to the cultivated land f The elements i are arranged in descending order according to the membership degree of the element i, and if there is the same membership degree, the corresponding elements are arranged randomly, and then the set of the above elements is taken as the planting crop detection index corresponding to the cultivated land f

[0067] If the planted crop set Y of the cultivated land f is not constructed f The f corresponding to the region can be determined by the following steps: The elements i are arranged in descending order according to the membership degree of the element i, and if there is the same membership degree, the corresponding elements are arranged randomly, and then the set of the above elements is taken as the planting crop detection index corresponding to the cultivated land f The elements i are arranged in descending order according to the membership degree of the element i, and if there is the same membership degree, the corresponding elements are arranged randomly, and then the set of the above elements is taken as the planting crop detection index corresponding to the cultivated land f The elements i are arranged in descending order according to the membership degree of the element i, and if there is the same membership degree, the corresponding elements are arranged randomly, and then the set of the above elements is taken as the planting crop detection index corresponding to the cultivated land f

[0068] More specifically, the specific operation of determining the value set of the pixel gray scale of different crops in different growth stages in the panchromatic image and the value set of the reflectivity of different crops in different growth stages on different characteristic wavelengths in the hyperspectral image in S2 through the panchromatic image data set and the hyperspectral image data set of different crop types in different growth stages in the region is as follows:

[0069] The growth process of the crop i in the I in step S1 can be divided into l i growth stages, where l i is the number of growth stages that the crop i can be divided into, j is the identification of the growth stage, and 1≤j≤l i ; The panchromatic image and the hyperspectral image of the crop i in the region in the growth stage j can be collected at the same time for the same scene, and the panchromatic image P i,j (m, n) and the hyperspectral image Q i,j (m, n) of the crop i in the growth stage j at time m and scene n can be fused to generate an image R i,j (m, n), where n is the scene number of collecting the panchromatic image and the hyperspectral image in the region, and m is the time of collecting the panchromatic image and the hyperspectral image in the region, thereby the panchromatic image data set X i,j of the crop i in the growth stage j and the hyperspectral image data set Y i,j of the crop i in the growth stage j can be made, which can be respectively expressed as:

[0070]

[0071] Where N is the scene set of collecting the panchromatic image and the hyperspectral image in the region, and M is the time range of collecting the panchromatic image and the hyperspectral image in the region;

[0072] The X i,j The full-color image P i,j (m,n) is grayed, the gray value a i,j (m,n) of any pixel point in the P i,j (m,n,k), wherein k is the number of the pixel point in the full-color image, K0 is the total number set of the pixel points in the full-color image, and considering that in mathematics, interval method, set method and number axis method are often used to represent the value range, so as to facilitate understanding and calculation, the value set A i,j of the crop i in the growth stage j about the pixel gray value in the full-color image can be determined by the set method as follows:

[0073]

[0074] Wherein, ∪ is the union symbol;

[0075] Suppose that the hyperspectral image data set Y i,j of the crop i in the growth stage j is determined, and the specific wavelength set representing the spectral characteristics is Ω i,j , then the hyperspectral image Q i,j contained in the Y i,j is selected to extract the reflectivity β i,j (m,n,h,r) on the feature wavelength h, wherein h is the identification of the feature wavelength in the hyperspectral image, and for the crop i, h∈H i , H i is the total feature wavelength set of the hyperspectral image of the crop i for feature extraction, r is the number of the region of interest in the hyperspectral image, R0 is the total number set of the regions of interest in the hyperspectral image, and r∈R0, thus the value set B i,j,h of the crop i in the growth stage j about the reflectivity of the hyperspectral image on the feature wavelength h can be determined as follows:

[0076]

[0077] More specifically, the specific operation of determining the value set of the different uncultivated states about the pixel gray value in the full-color image and the value set of the different uncultivated states about the reflectivity of the hyperspectral image on the different feature wavelengths in S2 according to the full-color image data set and the hyperspectral image data set of different uncultivated states is as follows:

[0078] Suppose that W is the set of uncultivated states in the region, and w is the uncultivated state in the W, the full-color image u w (m,n) and the hyperspectral image v w(m,n), from which a panchromatic image dataset U of the uncultivated state w in the basic farmland protection area can be created. w and hyperspectral image dataset V w , can be represented as:

[0079] U w ={u w (m,n)|w∈W,n∈N,m∈M}

[0080] V w ={v w (m,n)|w∈W,n∈N,m∈M}

[0081] The U w The panchromatic image u contained w After converting (m,n) to grayscale, the grayscale value φ of any pixel can be obtained. w (m,n,k), thus determining the set of pixel gray values ​​ρ for the uncultivated state w in the panchromatic image. w for:

[0082]

[0083] V w The hyperspectral image v included w Select the region of interest r in (m,n) to extract the reflectance at the characteristic wavelength h. For the uncultivated state w, h∈H should be satisfied. w H w Let r∈R0 be the set of feature wavelengths for feature extraction from a hyperspectral image of an uncultivated state. From this, we can determine the set of values ​​σ for the reflectance of the uncultivated state w at the feature wavelength h in the hyperspectral image. w,h for:

[0084]

[0085] More specifically, in S3, for a certain piece of farmland, when monitoring changes in farmland planting use, its panchromatic image and hyperspectral image can be collected. Randomly sample pixels from the panchromatic image and determine the region of interest in the hyperspectral image based on the geometric shape formed by the pixels. The specific operation of defining the affiliation of the farmland to a certain crop is as follows:

[0086] According to step S2, the crop i at growth stage j takes the set A of pixel grayscale values ​​in the panchromatic image. i,j The pixel grayscale can be set to belong to the fuzzy set of "crop i at growth stage j". The membership function It can be defined as:

[0087]

[0088] Where x is the variable of the input function, Ψ is the set of input functions, Closest(Ψ,x) is the function that finds the element in set Ψ whose value is closest to z, above(Ψ,x) is the function that finds the element in set Ψ whose value is greater than z, and below(Ψ,x) is the function that finds the element in set Ψ whose value is less than z.

[0089] According to step S2, the set B of values ​​for the reflectance of crop i at the characteristic wavelength h of the hyperspectral image during its growth stage j is as described. i,j,h The reflectance of the region of interest can be set to belong to the fuzzy set of "crop i with respect to characteristic wavelength h at growth stage j". The membership function It can be defined as:

[0090]

[0091] Here, `biggest` is a function that returns the largest element in the input set, and `smallest` is a function that returns the smallest element in the input set.

[0092] For cultivated land f, let P be the panchromatic image collected when monitoring changes in cultivated land use at time m and scene n. f (m,n) and the hyperspectral image are Q f (m,n), where the scenario n should be set within the geometric range of f, and P is defined as follows. f The set of pixel numbers randomly sampled from (m,n) is K. f It can be assumed that K is the K f The set of numbers for the regions of interest within the geometric shape composed of medium-sized pixels is R. f Then, the membership status ZWLSZK(f,i) of cultivated land on f belonging to crop i can be defined as:

[0093]

[0094] Where, α f (m,n,k) represents the panchromatic image P. f K in (m,n) f The gray value corresponding to pixel k, β f (m,n,h,r) represents the hyperspectral image Q. f R in (m,n) f The reflectance of the region of interest r at the characteristic wavelength h, and for crop i, h∈Hi max is a function that takes the maximum value;

[0095] More specifically, in step S3, the crop status membership degree corresponding to the cultivated land is detected sequentially according to the crop detection index given in step S1. If the membership degree of a certain crop is greater than or equal to the planting status membership degree, then the crop is output and reported as the crop currently planted on the cultivated land. Otherwise, the maximum value of the membership degree of the crop planted on the cultivated land is taken as the membership degree of the current planting status of the cultivated land, and the corresponding crop is taken as the possible crop currently planted on the cultivated land. The process then proceeds to step S4.

[0096] Set γ as the crop membership threshold, and then detect the crop index L corresponding to cultivated land f according to step S1. f The L are detected sequentially f The membership status ZWLSZK(f,i) corresponding to crop i is included. If the membership status ZWLSZK(f,i) corresponding to a certain crop i satisfies greater than or equal to the membership degree γ of the planting status, then crop i is regarded as the crop κ currently planted on cultivated land f. f Perform output reporting; if the L f If the membership status ZWLSZK(f,i) corresponding to all crops i in the matrix is ​​less than the stated γ, then... The corresponding crop i is the crop that can currently be planted on the cultivated land f. and will The subordinate status τ of the current planting status of the cultivated land f f Then proceed to step S4;

[0097] More specifically, the specific operation in S4 to define the affiliation status of the cultivated land in step S3 as being in a certain uncultivated state is as follows:

[0098] According to step S2, the uncultivated state w has the same value set ρ as the pixel grayscale value set in the panchromatic image. w You can set the pixel grayscale to belong to the fuzzy set of "uncultivated state w". The membership function It can be defined as:

[0099]

[0100] According to step S2, the set of values ​​σ for the reflectance of the hyperspectral image at the characteristic wavelength h in the uncultivated state w is given by the following values. w,h You can set the fuzzy set of reflectance of the region of interest to belong to the "uncultivated state w". The membership function It can be defined as:

[0101]

[0102] Define the membership status WGZLSZK(f,w) of the uncultivated land f in step S3 as follows:

[0103]

[0104] More specifically, in step S4, based on the different uncultivated states in the set of uncultivated states in the basic farmland protection area described in step S2, the corresponding affiliation status is given, and the maximum value of the affiliation status and the uncultivated state corresponding to the maximum value of the affiliation status are determined. The maximum value of the affiliation status is compared with the crop affiliation threshold described in step S3. If it is greater than or equal to the threshold, the uncultivated state corresponding to the maximum value of the affiliation status is output and reported as the current uncultivated state of the farmland. If it is less than the threshold, the maximum value of the affiliation status is compared with the affiliation status of the current planting status of the farmland described in step S3. If the affiliation status of the current planting status of the farmland is greater, the affiliation status of the currently possible crops to be planted on the farmland and the affiliation status of the current planting status of the farmland described in step S3 are output and reported. Otherwise, the uncultivated state corresponding to the maximum value of the affiliation status is taken as the current possible uncultivated state of the farmland, and the maximum value of the affiliation status is taken as the affiliation status of the current uncultivated state of the farmland. The specific operation of outputting and reporting the above two is as follows:

[0105] Based on step S2, W provides WGZLSZK(f,w) corresponding to different values ​​of w, and the maximum value among them is determined. and the corresponding uncultivated state w; the aforementioned Compare with the crop membership threshold γ described in step S3, if Then The corresponding w represents the current uncultivated state of arable land f. f Perform output reporting; if Then The affiliation status τ of the current planting status of cultivated land f as described in step S3 f If a comparison is made, Then the crops that can currently be planted on the cultivated land f mentioned in step S3 will be... The affiliation status of cultivated land f under its current planting status τ f Output and report, otherwise... The corresponding w represents the current possible uncultivated state of arable land f. and will The affiliation status ζ of the currently uncultivated state of the cultivated land f f And will determine the current possible uncultivated state of arable land. The affiliation status of arable land f in its current uncultivated state ζ f Perform output reporting;

[0106] The effectiveness of this invention can be further illustrated by the following experimental results and analysis.

[0107] set up Figure 2 The area to the left of the dashed line represents a region where land use change monitoring was conducted. The rectangle containing this region is labeled 17, and it can be recorded as region 17. The arable land labeled 1783, within the ellipse, is outlined by the dashed line and can be recorded as arable land 1783. Table 1 shows the crops registered for arable land 1783 on different land use change monitoring dates. It also sets the... Figure 2 The crop type set I of the 17 regions mentioned above can be represented as {rice (1), corn (2), tobacco (3)} and the set W of uncultivated states in the 17 regions can be set as {water (4), straw residue (5), mud (6)}, where the numbers in the set are the identifiers of the corresponding crops or uncultivated states. It can be seen that for cultivated land 1783, the Zadeh notation can be used to represent the crop type I of the 17 regions. Represented as Therefore, step S1 can be used to obtain the crop detection index L corresponding to cultivated land 1783. 1783 ={1,2,3};

[0108] Table 1

[0109] Date of monitoring of change in use of arable land for cultivation Crop Date of monitoring of change in use of arable land for cultivation Crop March 2, 2023 Rice June 2, 2024 Rice August 14, 2023 Rice September 11, 2024 Rice October 22, 2023 Corn November 15, 2024 Tobacco December 3, 2023 Tobacco January 2, 2025 Corn February 11, 2024 Tobacco March 12, 2025 Rice April 1, 2024 Rice May 1, 2025 Rice

[0110] Let it be Figure 2 Panchromatic image datasets and hyperspectral image datasets of different crops at different growth stages were created in 17 regions of China. This was to determine the set of pixel gray values ​​of different crops at different growth stages in the panchromatic images and the set of reflectance values ​​of different crops at different growth stages in the hyperspectral images at different characteristic wavelengths. Considering that the R-channel histogram has a bimodal structure with a large distance, in this embodiment of the invention, only the R-channel gray value is processed in the gray value processing for the convenience of demonstration and subsequent processing. Taking rice (1) as an example, its growth process is divided into 8 growth stages, namely seedling stage (1), tillering stage (2), jointing stage (3), booting stage (4), heading stage (5), flowering stage (6), grain filling stage (7) and maturity stage (8). The above numbers are the identifiers of rice (1) at different growth stages. The pixel gray value defined in step S3 belongs to the fuzzy set of "crop i at growth stage j". membership function It can be determined that the pixel gray level belongs to the membership function corresponding to the fuzzy set of "rice (1) in the tillering stage (2)", as shown in the specific situation. Figure 3As shown in the figure. The set H1 of all feature wavelengths for feature extraction in the hyperspectral image of rice (1) is set to {670, 780}, where the wavelength unit is nm. The reflectance of the region of interest defined in step S3 belongs to the fuzzy set of "crop i with respect to feature wavelength h at growth stage j". membership function The membership function corresponding to the fuzzy set of "rice (1) with respect to characteristic wavelengths 670nm / 780nm during tillering stage (2)" can be determined separately for the reflectance of the region of interest. The specific situation is as follows: Figure 4 (a) and Figure 4 As shown in (b).

[0111] For cultivated land 1783, the pixel numbers and corresponding grayscale values ​​of randomly sampled pixels after acquiring a panchromatic image during the monitoring and processing of changes in cultivated land use are as follows: Figure 5 As shown in the figure, and the set of numbered regions of interest within the geometric shape formed by the aforementioned pixels after the hyperspectral image was acquired during the monitoring and processing of changes in cultivated land use, and the reflectance of the aforementioned regions of interest at different characteristic wavelengths are respectively as follows: Figure 6 As shown in the figure. The crop affiliation threshold is set to 0.9. The affiliation status of the crops included in the above-mentioned cultivated land is detected in sequence according to the crop detection index. If the affiliation status of rice (1) in the tillering stage (2) on cultivated land 1783 is 0.93, then rice can be output and reported as the crop currently planted on cultivated land 1783, thereby completing the monitoring and processing of changes in cultivated land planting use for a single cultivated land.

[0112] All parts not explicitly stated in the above embodiments of the present invention can be implemented using existing technologies. The above embodiments of the present invention are merely illustrative of the technical solutions disclosed in the present invention and are not intended to limit the present invention in any way. Any simple modifications, equivalent changes, and alterations made by those skilled in the art to the above embodiments based on the technical content of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for monitoring and processing changes in cultivated land use based on hyperspectral remote sensing, characterized in that, Includes the following steps: Step S1: Develop a crop detection index for cultivated land in a specific region's cultivated land database; Step S2: Determine the set of pixel gray values ​​in the panchromatic image or the set of reflectance values ​​at different characteristic wavelengths in the hyperspectral image for different crops at different growth stages in the set of crop types in this region, as well as the set of pixel gray values ​​in the panchromatic image and the set of reflectance values ​​at different characteristic wavelengths in the hyperspectral image for different uncultivated states in the set of uncultivated states in this region. The specific method is as follows: The growth process of crop i is divided into l i There are several growth stages, of which l i Let j be the number of growth stages for crop i, and let j be the identifier of the growth stage, where 1 ≤ j ≤ l. i For crop i in growth stage j within the region, its panchromatic and hyperspectral images are acquired simultaneously for the same scene, and for any crop i in growth stage j at time m and scene n, the panchromatic image P... i,j (m,n) and hyperspectral image Q i,j (m,n) are fused to generate image R i,j (m,n), where n is the scene number for acquiring panchromatic and hyperspectral images in this region, and m is the time when acquiring panchromatic and hyperspectral images in this region; let X i,j For a panchromatic image dataset of crop i at growth stage j within this region, the X... i,j The panchromatic image P contained within i,j After grayscale conversion of (m,n), the P is given. i,j The gray value α of any pixel in (m,n) i,j Let (m,n,k) be the set of pixel numbers in the panchromatic image, and K0 be the set of all pixel numbers in the panchromatic image. Considering that in mathematics, methods such as interval methods, set methods, and number lines are often used to represent the range of values ​​for ease of understanding and calculation, the set method can be used to determine the set A of pixel grayscale values ​​of crop i at growth stage j in the panchromatic image. i,j for: ； in, For the union symbol, This is a collection of scenes for which panchromatic and hyperspectral images were acquired within the region. The time range for acquiring panchromatic and hyperspectral images within this region; Let Y be the hyperspectral image dataset of crop i at growth stage j in this region. i,j Using the hyperspectral image dataset Y of crop i at growth stage j i,j The specific set of wavelengths exhibiting spectral characteristics is defined as Ω. i,j Then through the Y i,j The hyperspectral image Q included i,j Select the region of interest r in (m,n) to extract the reflectance β at the characteristic wavelength h. i,j (m,n,h,r), where h is the identifier of the characteristic wavelength in the hyperspectral image, and for crop i, h∈H should be satisfied. i H i Let Ri be the set of all feature wavelengths for feature extraction from the hyperspectral image of crop i, r be the region of interest number in the hyperspectral image, and R0 be the set of all region of interest numbers in the hyperspectral image, where r ∈ R0. From this, we can determine the set B of reflectance values ​​of crop i at growth stage j with respect to the feature wavelength h in the hyperspectral image. i,j,h for: ； Let W be the set of uncultivated states in the region, and w be an uncultivated state in W. At time m and scene n, a panchromatic image u of the uncultivated state w in the region is acquired. w (m,n) and hyperspectral image v w (m,n), let the panchromatic image dataset and hyperspectral image dataset of the uncultivated state w in the basic farmland protection area be U, respectively. w and V w , will the U w The panchromatic image u contained w After converting (m,n) to grayscale, the grayscale value φ of any pixel can be obtained. w (m,n,k), thus determining the set of pixel gray values ​​ρ for the uncultivated state w in the panchromatic image. w for: ； V w The hyperspectral image v included w Select the region of interest r in (m,n) to extract the reflectance at the characteristic wavelength h. For the uncultivated state w, h∈H should be satisfied. w H w Let r∈R0 be the set of feature wavelengths for feature extraction from a hyperspectral image of an uncultivated state. This sets σ represent the set of reflectance values ​​for the uncultivated state w at the feature wavelength h. w,h for: ； Step S3: For a certain piece of farmland, the planting status membership degree of the crops corresponding to the farmland is detected sequentially according to the crop detection index given in step S1. If the membership degree of a certain crop satisfies ≥ planting status membership degree, it is output and reported as the crop currently planted in the farmland. Otherwise, the possible crops currently planted in the farmland are input to step S4. Step S4: Compare the maximum value of the affiliation status of different uncultivated states on the cultivated land described in Step S3 with the crop affiliation threshold in Step S3. If it is greater than or equal to the threshold, output and report the corresponding uncultivated state. Otherwise, compare the maximum value of the affiliation status with the affiliation status of the current planting status of the cultivated land described in Step S3, and output and report the uncultivated state or the current planting status corresponding to the larger value.

2. The method for monitoring and processing changes in cultivated land planting use based on hyperspectral remote sensing according to claim 1, characterized in that, The specific method for step S1 is as follows: Suppose that a certain region's arable land database Φ contains several arable land plots, f is the identifier of the arable land, and f∈Φ. Let i be the identifier of the crop, and I be the set of crop types in the region. For arable land f, register the crop r on the arable land use change monitoring date d. f,d Where d satisfies d∈Θ, and Θ is the set of dates for which changes in cultivated land use were previously monitored in this region, then construct the set of crops Υ planted on cultivated land f. f For {r f,d |d∈Θ}, where | is the condition symbol; according to fuzzy mathematics, let the universe of discourse U={i|i∈I}, and set and Let U be the fuzzy set of "crops" on the domain U and the fuzzy set of "crops" on the region U, respectively; if there exists a pre-constructed set of already planted crops Y for the domain U. f Then the f corresponding to According to element i Elements i are arranged in descending order of membership degree. If elements have the same membership degree, they are randomly arranged. The set of these elements is then used as the crop detection index L corresponding to cultivated land f. f If a set of already planted crops (Υ) has not been constructed for cultivated land (f); f Then the corresponding region According to the membership of element i The elements i are arranged in descending order of their degree of importance, and then the set of the above elements is used as the detection index L of the crop planted on the cultivated land f. f If elements have the same membership degree, the corresponding elements are randomly arranged.

3. The method for monitoring and processing changes in cultivated land planting use based on hyperspectral remote sensing according to claim 1, characterized in that, The specific method for step S3 is as follows: Let A be the set of pixel grayscale values ​​of crop i at growth stage j in the panchromatic image, as described in step S2. i,j The pixel grayscale can be set to belong to the fuzzy set of "crop i at growth stage j". And according to step S2, the set B of values ​​for the reflectance of crop i at the characteristic wavelength h of the hyperspectral image during growth stage j. i,j,h Set the region of interest to a fuzzy set whose reflectance belongs to "crop i with respect to characteristic wavelength h at growth stage j". For cultivated land f, let P be the panchromatic image collected when monitoring changes in cultivated land use at time m and scene n. f (m,n) and the hyperspectral image are Q f (m,n), where the scenario n should be set within the geometric range of f, and P is defined as follows. f The set of pixel numbers randomly sampled from (m,n) is K. f Let K be the K f The set of numbers for the regions of interest within the geometric shape composed of medium-sized pixels is R. f Then, the membership status ZWLSZK(f,i) of cultivated land on f belonging to crop i is defined as: ； in, and The respective and membership function, α f (m,n,k) represents the panchromatic image P. f K in (m,n) f The gray value corresponding to pixel k, β f (m,n,h,r) represents the hyperspectral image Q. f R in (m,n) f The reflectance of the region of interest r at the characteristic wavelength h, and for crop i, h∈H i max is a function that takes the maximum value; Set γ as the crop membership threshold, and then detect the crop index L corresponding to cultivated land f according to step S1. f The L are detected sequentially f The membership status ZWLSZK(f,i) corresponding to crop i is included. If the membership status ZWLSZK(f,i) corresponding to a certain crop i satisfies greater than or equal to the membership degree γ of the planting status, then crop i is regarded as the crop κ currently planted on cultivated land f. f Perform output reporting; if the L f If the membership status ZWLSZK(f,i) corresponding to all crops i in the matrix is ​​less than the stated γ, then... The subordinate status τ of the current planting status of the cultivated land f f ,Will The corresponding crop i is the crop that can currently be planted on the cultivated land f. Enter the information in step S4.

4. The method for monitoring and processing changes in cultivated land planting use based on hyperspectral remote sensing according to claim 1, characterized in that, The specific method for step S4 is as follows: According to step S2, the uncultivated state w has the same value set ρ as the pixel grayscale value set in the panchromatic image. w Set the pixel grayscale to belong to the fuzzy set of "uncultivated state w". And the set of values ​​σ for the reflectance of the uncultivated state w in step S2 with respect to the characteristic wavelength h of the hyperspectral image. w,h Set the region of interest to a fuzzy set whose reflectance belongs to the "uncultivated state w". Let the above and The membership functions are respectively and Therefore, the membership status WGZLSZK(f,w) of the uncultivated land w on cultivated land f in step S3 is defined as: ； Based on step S2, W provides WGZLSZK(f,w) corresponding to different values ​​of w, and the maximum value among them is determined. and the corresponding uncultivated state w; the aforementioned Compare with the crop membership threshold γ described in step S3, if Then The corresponding w represents the current uncultivated state of arable land f. f Perform output reporting; if Then The affiliation status τ of the current planting status of cultivated land f as described in step S3 f If a comparison is made, Then the crops that can currently be planted on the cultivated land f mentioned in step S3 will be... The affiliation status of cultivated land f under its current planting status τ f Output and report, otherwise... The corresponding w represents the current possible uncultivated state of arable land f. and will The affiliation status ζ of the currently uncultivated state of the cultivated land f f And will determine the current possible uncultivated state of arable land. The affiliation status of arable land f in its current uncultivated state ζ f Output and report.

Images