Remote sensing growth vigor monitoring method and device for dynamic matching of growth period

Abstract

The invention relates to the technical field of agricultural remote sensing crop growth monitoring, in particular to a growth period dynamic matching remote sensing growth monitoring method and equipment, which perform growth period time dynamic matching when a historical standard curve is constructed, solve the problem of curve error caused by inter-annual crop growth period difference when the historical standard curve is constructed, and improve the crop growth monitoring accuracy. And the aim of improving the precision of inter-annual data iteration is fulfilled. In the growth vigor monitoring process, the growth period is matched dynamically and circularly, the optimal historical matching time is searched for the latest data of each period, the optimal grade judgment threshold value of the period is obtained, and the accuracy of growth vigor monitoring is improved.

Description

Technical Field

[0001] The invention relates to the technical field of agricultural remote sensing crop growth monitoring, and in particular to a remote sensing growth monitoring method and equipment with dynamic matching of growth periods. Background Art

[0002] Crop growth refers to the growth status and trend of crops. Crop growth can be described by individual and group characteristics. The traditional method of obtaining crop growth is ground survey, while remote sensing technology is mainly used to monitor crop growth status and trend in modern agricultural production. Remote sensing monitoring of crop growth fully reflects the macroscopic, objective, timely and economical characteristics of remote sensing technology, which can provide timely decision-making support information for field management and provide a basis for early estimation of yield. The basis of remote sensing monitoring of growth is that there must be biological growth factors that can be monitored by remote sensing. Vegetation remote sensing parameters represented by vegetation index and leaf area index are recognized as remote sensing monitoring indicators that can reflect crop growth. Among them, the normalized vegetation index (NDVI) is the most widely used in crop planting information extraction, growth monitoring and yield forecasting. In particular, with the rapid development of "3S integrated application technology, high-resolution satellite data and big data computing technology, remote sensing monitoring information of crop growth has become an indispensable and important information for guiding agricultural production.

[0003] Crop growth monitoring mainly includes process monitoring and real-time monitoring. Process monitoring is also called monitoring over time. It mainly uses time-series NDVI images to calculate the required parameter values ​​to construct the crop growth process, and compares the growth conditions of each period between years to reflect the growth conditions of crops. Real-time monitoring refers to the use of real-time NDVI image values, compared with the NDVI values ​​of last year, multi-year average or a specified year, and the NDVI difference values ​​are graded, counted and regionally displayed to achieve real-time reflection of crop growth differences.

[0004] However, whether it is process monitoring or real-time monitoring, when making inter-annual comparisons, most of the comparisons are based on absolute time. However, there are differences in the growth period of crops between years, and direct comparisons will result in certain errors. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a remote sensing growth monitoring method and equipment with dynamic matching of growth periods, so as to solve the problem that there are certain errors in the comparison in the prior art.

[0006] According to a first aspect of an embodiment of the present invention, a remote sensing growth monitoring method with dynamic matching of growth periods is provided, characterized in that it comprises:

[0007] Obtain historical crop growth data, and derive filter signals, slope signals and fluctuation signals based on the original time series of each year in the historical crop growth data;

[0008] According to the filtering signal, the slope signal and the fluctuation signal of each year, an optimal bending path is calculated, and the optimal bending path is applied to the original time series to obtain a similarity measurement result;

[0009] Iterate the standard curve according to the similarity measurement result to construct a historical standard curve;

[0010] Acquire the latest crop growth data, and calculate the best value, good value and median value based on the latest crop growth data;

[0011] Adding the good value to the growth curve of the planting season, matching the good value with the historical standard curve to obtain the best bending path of the original time series of the two, and setting the current growth level threshold according to the best bending path of the original time series of the two;

[0012] The growth grade of the latest crop growth data is determined according to the current growth grade threshold.

[0013] Preferably, obtaining a filtered signal according to the original time series includes:

[0014] Use a sliding window to perform moving average on the original time series to obtain a filtered signal;

[0015] The filtered signal F={f1,…,f n}, the length of the sliding window is 2r+1;

[0016] f i Calculated by the following formula:

[0017]

[0018] Among them, the original time series is X={x1,…,x n}; r is the filter signal parameter; k is the kth period data of the filter window; n is the total number of data periods.

[0019] Preferably, obtaining a slope signal according to the original time series includes:

[0020] Segmenting the filtered signal according to the positions of the local extreme value points of the filtered signal to obtain a plurality of monotonically changing subsequence segments;

[0021] Perform linear fitting on each subsequence segment, and use the slope of the fitted line segment as the slope of the corresponding subsequence segment of the original time series;

[0022] Merge the slopes of the subsequence segments to obtain the slope signal corresponding to the original time series;

[0023] The slope signal is K = {k1,…,k n}, k i The calculation formula is as follows:

[0024]

[0025] Among them, f v is the value of the local extreme point of the filtered signal F, f p is the location of the local extreme point of the filtered signal F.

[0026] Preferably, the fluctuation signal is obtained according to the original time series, including:

[0027] The time series is decomposed using the additive model to obtain the fluctuation signal C = {c1,…,c n};

[0028] c i The calculation formula is: i =x i -f i ,i=1,…n.

[0030] Preferably, the calculating of the best curved path according to the annual filtering signal, the slope signal and the fluctuation signal comprises:

[0031] The DTW algorithm is used to measure the similarity distances of the filter signal, the slope signal, and the fluctuation signal to obtain a cumulative distance matrix. The optimal bending path of the original time series is obtained according to the cumulative distance matrix. The formula is as follows:

[0032] DTW(X,Y)=D(n,m)

[0033]

[0034] F(i,j)=(F x (i)-F y (j)) 2

[0035] K(i,j)=(K x (i)-K y (j)) 2

[0036] C(i,j)=(C x (i)-C y (j)) 2

[0037] Where D is the cumulative cost matrix, F(i,j) is the distance between two points of the filtered signal, K(i,j) is the distance between two points of the slope signal, and C(i,j) is the distance between two points of the fluctuation signal; η is the slope signal weight; μ is the fluctuation signal weight.

[0038] Preferably, performing standard curve iteration according to the similarity measurement result to construct a historical standard curve includes:

[0039] The standard curve was iterated using the following formula:

[0040]

[0041] Among them, NDVI i is the value at the i-th time after iteration, is the value of the historical standard curve at time i, is the value of the iterated time series j corresponding to i in the optimal bending path D(i,j) of the curve, and a is the cumulative year of the historical curve.

[0042] Preferably, the best value, good value and median value are calculated based on the latest crop growth data, and the formula is as follows:

[0043]

[0044] Among them, NDVI l This is the latest good value, NDVI y It is the latest figure of merit, NDVI z is the median value of the latest period, NDVI t is the mean NDVI of the tth plot, and k is the total number of plots in the region.

[0045] Preferably, the current growth level threshold is set by the following formula:

[0046]

[0047]

[0048] NDVI l i is a good value at time i, NDVI y i is the figure of merit at time i, NDVI z i is the median value at time i, is the historical standard curve data corresponding to j when i=n in the optimal curved path D(i,j), is the good level threshold at time i, is the priority threshold at time i, is the intermediate threshold at time i.

[0049] According to a second aspect of an embodiment of the present invention, there is provided a remote sensing growth monitoring device with dynamic matching of growth periods, comprising:

[0050] A main controller, and a memory connected to the main controller;

[0051] a memory in which program instructions are stored;

[0052] The main controller is used to execute program instructions stored in the memory to perform any of the above methods.

[0053] The technical solution provided by the embodiments of the present invention may have the following beneficial effects:

[0054] It is understandable that the technical solution shown in the present invention performs dynamic matching of growth period time when constructing the historical standard curve, which solves the problem of curve error caused by the difference in crop growth period between years when constructing the historical standard curve, and meets the purpose of improving the accuracy of inter-annual data iteration. During the growth monitoring process, the growth period is matched in a dynamic cycle, the best historical matching time is found for the latest data of each period, and the best grade judgment threshold for the period is obtained, thereby improving the accuracy of growth monitoring. The data of different plots in different years are normalized to the relative time based on the growth period, which is more in line with the law of crop growth and meets the accuracy requirements of crop growth monitoring in different periods.

[0055] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0057] Figure 1 It is a schematic diagram of the steps of a remote sensing growth monitoring method with dynamic matching of growth periods according to an exemplary embodiment. DETAILED DESCRIPTION

[0058] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0059] In one embodiment, Figure 1 is a schematic diagram of the steps of a remote sensing growth monitoring method for dynamic matching of growth periods according to an exemplary embodiment, see Figure 1 , provides a remote sensing growth monitoring method with dynamic matching of growth period, characterized by comprising:

[0060] Step S11, obtaining historical crop growth data, and obtaining a filter signal, a slope signal and a fluctuation signal according to the original time series of each year in the historical crop growth data.

[0061] Step S12: Calculate the best curved path based on the filtered signal, slope signal and fluctuation signal of each year, and apply the best curved path to the original time series to obtain a similarity measurement result.

[0062] Step S13: Iterate the standard curve according to the similarity measurement result to construct a historical standard curve.

[0063] Step S14, obtaining the latest crop growth data, and calculating the best value, good value and median value based on the latest crop growth data.

[0064] Step S15, adding the good value to the growth curve of the planting season, matching the good value with the historical standard curve, obtaining the best curved path of the original time series of the two, and setting the current growth level threshold according to the best curved path of the original time series of the two.

[0065] Step S16: determining the growth grade of the latest crop growth data according to the current growth grade threshold.

[0066] It is understandable that the technical solution shown in this embodiment performs dynamic matching of growth period time when constructing the historical standard curve, which solves the problem of curve error caused by the difference in growth period of crops between years when constructing the historical standard curve, and meets the purpose of improving the accuracy of inter-annual data iteration. During the growth monitoring process, the growth period is matched dynamically and cyclically, the best historical matching time is found for the latest data of each period, and the best grade judgment threshold for the period is obtained, thereby improving the accuracy of growth monitoring. The data of different years and different plots are normalized to the relative time based on the growth period, which is more in line with the law of crop growth and meets the accuracy requirements of crop growth monitoring in different periods.

[0067] In practice, in order to ensure the objectivity and accuracy of the setting of growth thresholds in each period during the growth monitoring process, it is necessary to analyze and process the historical growth data of crops by region and construct a standard growth curve. In the past, the historical standard curve construction was mainly based on absolute time as a reference for the iteration of multi-year data, without considering the impact of inter-annual growth period differences. The constructed curve had low accuracy for plot-level growth evaluation.

[0068] In order to improve the reliability of the standard curve, the technical solution of this embodiment firstly performs an NDVI analysis on the historical standard curve before iterating the historical data. h And the NDVI curve for this planting season d The data is dynamically matched to the growth period.

[0069] In step S11, it should be noted that the filtered signal is obtained according to the original time series, including:

[0070] Use a sliding window to perform moving average on the original time series to obtain a filtered signal;

[0071] The filtered signal F={f1,…,f n}, the length of the sliding window is 2r+1;

[0072] f i Calculated by the following formula:

[0073]

[0074] Among them, the original time series is X={x1,…,x n}; r is the filter signal parameter; k is the kth period data of the filter window; n is the total number of data periods.

[0075] It should be noted that the fluctuation signal obtained based on the original time series includes:

[0076] While filtering eliminates the original time series band, a lot of detail information is lost. In order to retain both the filtered signal and the fluctuation signal, the additive model is used to decompose the time series, and the fluctuation signal C = {c1,…,c n};c i The calculation formula is: i =x i -f i ,i=1,…n.

[0078] It should be noted that the slope signal K is added to represent the slope change of the time series. The slope signal is obtained based on the original time series, including:

[0079] The filtered signal is segmented according to the position of the local extreme point of the filtered signal to obtain a plurality of monotonically changing subsequence segments; a linear fit is performed on each subsequence segment, and the slope of the fitted line segment is used as the slope of the subsequence segment corresponding to the original time series; the slopes of the subsequence segments are combined to obtain a slope signal corresponding to the original time series; the slope signal is K={k1,…,k n}, k i The calculation formula is as follows:

[0080]

[0081] Among them, f v is the value of the local extreme point of the filtered signal F, f p is the location of the local extreme point of the filtered signal F.

[0082] It can be understood that the original time series X is decomposed into a filter signal F, a slope signal K and a fluctuation signal C through the above calculation. Preferably, according to the characteristics of the time series data and actual needs, the weight of the decomposed signal is adjusted to obtain a better similarity measurement result.

[0083] It should be noted that the best curved path is calculated based on the annual filtering signal, slope signal and fluctuation signal, including:

[0084] The DTW algorithm is used to measure the similar distances of the filter signal, the slope signal and the fluctuation signal to obtain a cumulative distance matrix, and the optimal bending path of the original time series is obtained according to the cumulative distance matrix.

[0085] Two vegetation index NDVI time series X and Y with different lengths of n and m in different years, the filtered signal F is obtained by time series decomposition x and F y , slope signal K x and K y , and the fluctuation signal C x and C y , the calculation formula is as follows:

[0086] DTW(X,Y)=D(n,m)

[0087]

[0088] F(i,j)=(F x (i)-F y (j)) 2

[0089] K(i,j)=(K x (i)-K y (j)) 2

[0090] C(i,j)=(C x (i)-C y (j)) 2

[0091] Where D is the cumulative cost matrix, F(i,j) is the distance between two points of the filtered signal, K(i,j) is the distance between two points of the slope signal, and C(i,j) is the distance between two points of the fluctuation signal; η is the slope signal weight; μ is the fluctuation signal weight.

[0092] It can be understood that the best curved path is determined according to the cumulative distance matrix, and the alignment path is applied to the original time series to obtain the similarity measurement result. The inter-annual time series curve data is adjusted from absolute time alignment to relative time alignment to meet the needs of matching the growth period.

[0093] It should be noted that the standard curve is iterated according to the similarity measurement result to construct a historical standard curve, including:

[0094] The standard curve was iterated using the following formula:

[0095]

[0096] Among them, NDVI i is the value at the i-th time after iteration, is the value of the historical standard curve at time i, is the value of the iterated time series j corresponding to i in the optimal bending path D(i,j) of the curve, and a is the cumulative year of the historical curve.

[0097] After the historical standard curve is constructed, the growth monitoring of the entire growth period can be carried out after the next planting season begins. Similarly, there are differences in growth periods during the monitoring process. To improve the accuracy of monitoring, the data is curve matched period by period, the growth period matching time is dynamically adjusted, and the growth level judgment threshold is determined. The specific steps include S14 to S16.

[0098] It should be noted that, in step S14, the best value, good value and median value are calculated based on the latest crop growth data, and the formula is as follows:

[0099]

[0100]

[0101] Among them, NDVI l This is the latest good value, NDVI y It is the latest figure of merit, NDVI z is the median value of the latest period, NDVI t is the mean NDVI of the tth plot, and k is the total number of plots in the region.

[0102] In step S15, the good value NDVI l Add it to the growth curve of the planting season, and use the dynamic time matching method of the growth period to convert the good value NDVI l ={X1,…,X n} and historical standard curve NDVI h ={Y1,…,Y m} are matched to obtain the best curved path of the original time series of the two, and the current growth level threshold is set according to the historical standard curve data corresponding to j when i=n in the best curved path D(i,j) of the original time series of the two.

[0103] It should be noted that the current growth level threshold is set by the following formula:

[0104]

[0105] NDVI l i is a good value at time i, NDVI y i is the figure of merit at time i, NDVI z i is the median value at time i, is the historical standard curve data corresponding to j when i=n in the optimal curved path D(i,j), is the good level threshold at time i, is the priority threshold at time i, is the intermediate threshold at time i.

[0106] In step S16, the growth level of the latest crop growth data is determined according to the current growth level threshold, and the formula is as follows:

[0107]

[0108] Among them, f(t) is the growth grade, f(t) equals 0 for excellent, f(t) equals 1 for good, f(t) equals 2 for medium, and f(t) equals 3 for poor. NDVI t is the mean NDVI of the tth plot.

[0109] In addition, it is also possible to search in real time whether there is the latest data. If there is a new period of data, the steps S14 to S16 are looped to calculate the threshold values ​​of each level and make a growth level determination.

[0110] According to a second aspect of an embodiment of the present invention, there is provided a remote sensing growth monitoring device with dynamic matching of growth periods, comprising:

[0111] A main controller, and a memory connected to the main controller;

[0112] a memory in which program instructions are stored;

[0113] The main controller is used to execute program instructions stored in the memory to perform any of the above methods.

[0114] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0115] It should be noted that, in the description of the present invention, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, the meaning of "plurality" refers to at least two.

[0116] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0117] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0118] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0119] In addition, each functional unit in each embodiment of the present invention may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0120] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0121] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0122] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A remote sensing growth monitoring method with dynamic matching of growth period, characterized in that: include: Obtain historical crop growth data, and derive filter signals, slope signals and fluctuation signals based on the original time series of each year in the historical crop growth data; According to the filtering signal, the slope signal and the fluctuation signal of each year, the best bending path is calculated, and the best bending path is applied to the original time series to obtain a similarity measurement result; Iterate the standard curve according to the similarity measurement result to construct a historical standard curve; Acquire the latest crop growth data, and calculate the best value, good value and median value based on the latest crop growth data; Adding the good value to the growth curve of the planting season, matching the good value with the historical standard curve to obtain the best bending path of the original time series of the two, and setting the current growth level threshold according to the best bending path of the original time series of the two; The growth grade of the latest crop growth data is determined according to the current growth grade threshold.

2. The remote sensing growth monitoring method with dynamic matching of growth period according to claim 1 is characterized in that: The filtered signal is obtained from the original time series, including: Use a sliding window to perform moving average on the original time series to obtain a filtered signal; The filtered signal F={f1,…,f n }, the length of the sliding window is 2r+1; f i Calculated by the following formula: The original time series is X = {x1,…,x n }; r is the filter signal parameter; k is the kth period data of the filter window; n is the total number of data periods.

3. The remote sensing growth monitoring method with dynamic matching of growth period according to claim 2 is characterized in that: The slope signal is obtained based on the original time series, including: Segmenting the filtered signal according to the positions of the local extreme value points of the filtered signal to obtain a plurality of monotonically changing subsequence segments; Perform linear fitting on each subsequence segment, and use the slope of the fitted line segment as the slope of the corresponding subsequence segment of the original time series; Merge the slopes of the subsequence segments to obtain the slope signal corresponding to the original time series; The slope signal is K = {k1,…,k n }, k i The calculation formula is as follows: Among them, f v is the value of the local extreme point of the filtered signal F, f p is the location of the local extreme point of the filtered signal F.

4. The remote sensing growth monitoring method with dynamic matching of growth period according to claim 3 is characterized in that: The volatility signals are derived from the original time series, including: The time series is decomposed using the additive model to obtain the fluctuation signal C = {c1,…,c n }; c i The calculation formula is: i =x i -f i ,i=1,…n.。 5. The remote sensing growth monitoring method with dynamic matching of growth period according to claim 4 is characterized in that: The optimal bending path is calculated based on the annual filtering signal, slope signal and fluctuation signal, including: The DTW algorithm is used to measure the similarity distances of the filter signal, the slope signal, and the fluctuation signal to obtain a cumulative distance matrix. The optimal bending path of the original time series is obtained according to the cumulative distance matrix. The formula is as follows: DTW(X,Y)=D(n,m) F(i,j)=(F x (i)-F y (j)) 2 K(i,j)=(K x (i)-K y (j)) 2 C(i,j)=(C x (i)-C y (j)) 2 Where D is the cumulative cost matrix, F(i,j) is the distance between two points of the filtered signal, K(i,j) is the distance between two points of the slope signal, and C(i,j) is the distance between two points of the fluctuation signal; η is the slope signal weight; μ is the fluctuation signal weight.

6. The remote sensing growth monitoring method with dynamic matching of growth period according to claim 5 is characterized in that: The standard curve is iterated according to the similarity measurement result to construct a historical standard curve, including: The standard curve was iterated using the following formula: Among them, NDVI i is the value at the i-th time after iteration, is the value of the historical standard curve at time i, is the value of the iterated time series j corresponding to i in the optimal bending path D(i,j) of the curve, and a is the cumulative year of the historical curve.

7. The remote sensing growth monitoring method with dynamic matching of growth period according to claim 6 is characterized in that: The excellent value, good value and median value are calculated based on the latest crop growth data, and the formula is as follows: When NDVIt≥NDVIl When NDVIt <NDVIl Among them, NDVI l This is the latest good value, NDVI y It is the latest figure of merit, NDVI z is the median value of the latest period, NDVI t is the mean NDVI of the tth plot, and k is the total number of plots in the region.

8. The remote sensing growth monitoring method with dynamic matching of growth period according to claim 7 is characterized in that: The current growth level threshold is set by the following formula: in is a good value for time i, is the figure of merit at time i, is the median value at time i, is the historical standard curve data corresponding to j when i=n in the optimal curved path D(i,j), is the good level threshold at time i, is the priority threshold at time i, is the intermediate threshold at time i.

9. A remote sensing growth monitoring device with dynamic matching of growth period, characterized in that: include: A main controller, and a memory connected to the main controller; a memory in which program instructions are stored; The main controller is used to execute program instructions stored in the memory and perform the method according to any one of claims 1 to 8.

Images