A remote sensing time series estimation method, system and storage medium for crop physiological parameters based on phenological segmentation DTW

Abstract

The present invention discloses a remote sensing time series estimation method, system and storage medium for crop physiological parameters based on phenological segmented DTW, including the following steps: preprocessing time series remote sensing images to obtain time series remote sensing data; calculating vegetation index time series data set based on time series remote sensing data; building a remote sensing estimation model for estimating crop physiological parameters in combination with crop sample measured data; building a crop standard vegetation index time series curve and a vegetation index time series curve data set based on plots as basic geographical units; determining the time range of crop growth in target plots based on vegetation index time series curves; obtaining crop time series matching paths based on dynamic time warping algorithms using vegetation index time series curves as inputs; and adjusting the time parameters of crops in target plots according to the obtained matching paths to estimate crop physiological parameters. The present invention overcomes the problem of low accuracy of remote sensing time series estimation of crop physiological parameters caused by phenological mismatch.

Description

Technical Field

[0001] The present invention belongs to the technical field of quantitative remote sensing and crop monitoring, and in particular relates to a remote sensing time series estimation method, system and storage medium for crop physiological parameters based on phenological segmented DTW. Background Art

[0002] Satellite remote sensing reveals the spatiotemporal dimensions of crop growth and yield through time series remote sensing vegetation indices, and has become the main method for obtaining crop information at local, regional and global scales. Analyzing the crop spectral response obtained by remote sensing through mathematical methods and computer technology to invert crop physiological parameters is a specific means to achieve crop growth monitoring. At present, most estimation models link vegetation indices with crop physiological parameters to construct expressions from observed data to target parameters. The advantage of this type of method is that there is a clear causal relationship between observed data and target parameters, which can reveal the role of variables in rice plant height models. However, the spatiotemporal generalization ability of this type of model is poor, and the estimation results often have deviations when applied in different regions or years. In addition, it is particularly important to resolve temporal changes when using inversion models to adapt to new years and regions. Due to different agricultural climate conditions and agricultural practices (including different sowing times, growth cycles, etc.), two plots of the same type may have different phenological evolutions. When the source training model is applied to the target year and region, time offset may cause errors in LAI estimation.

[0003] Some studies have used Savitzky-Golay filtering, Fourier transform, wavelet transform and other techniques to smooth the time series of vegetation indices, which can extract key nodes such as maximum values, minimum values ​​and inflection points, thereby accurately locating key time points in the crop growth cycle. However, this method is limited to monitoring specific crop growth stages and lacks coordinated analysis of the entire crop growth cycle.

[0004] Therefore, it is urgent to solve the above problems. Summary of the invention

[0005] Purpose of the invention: The first purpose of the present invention is to provide a method for remote sensing time series estimation of crop physiological parameters based on phenological segmented DTW, which aims to overcome the problem of low accuracy of remote sensing time series estimation of crop physiological parameters caused by phenological mismatch, and to achieve crop time series matching in different agricultural scenarios and high-precision time series estimation of crop physiological parameters.

[0006] The second object of the present invention is to provide a rice leaf area index estimation system based on dual-polarization SAR images with dynamic constraints on canopy structure.

[0007] A third object of the present invention is to provide a computer storage medium.

[0008] Technical solution: To achieve the above purpose, the present invention discloses a remote sensing time series estimation method for crop physiological parameters based on phenological segmented DTW, comprising the following steps:

[0009] (1) Preprocess the time series remote sensing images to obtain processed time series remote sensing data;

[0010] (2) Based on the time series remote sensing data, the vegetation index is calculated to obtain a vegetation index time series data set;

[0011] (3) Combined with the measured data of crop samples, a remote sensing estimation model for estimating crop physiological parameters is constructed;

[0012] (4) Taking plots as the basic geographical units, we constructed the crop standard vegetation index time series curve and vegetation index time series curve datasets respectively;

[0013] (5) Determine the time range of crop growth in the target plot based on the vegetation index time series curve;

[0014] (6) Based on the dynamic time warping algorithm, a phenological segmented time-weighted dynamic algorithm was constructed, and the crop time series matching path was obtained using the vegetation index time series curve as input;

[0015] (7) Adjust the time parameters of the target plot crops according to the obtained matching path, and estimate the crop physiological parameters based on the remote sensing estimation model and the time series vegetation index.

[0016] Wherein, in step (1), if the time series remote sensing image is an optical remote sensing image, the preprocessing includes radiation calibration, atmospheric correction, geometric correction and image filtering;

[0017] If the time series remote sensing image is a microwave remote sensing image, the preprocessing includes multi-view processing, speckle filtering, geocoding and radiometric calibration.

[0018] Preferably, step (3) specifically includes the following steps:

[0019] The least squares method is used to describe the relationship between crop physiological parameters and vegetation index in each growth period, and polynomials are used to describe the relationship between model parameters and time, so as to construct a remote sensing estimation model, which is expressed as follows:

[0020] (1)

[0021] (2)

[0022] (3)

[0023] In the formula Represents physiological parameters, is the vegetation index, kand b are the coefficients and intercepts of the least squares regression model for different growth periods, is the number of days in the year, n is the order of the polynomial, and are polynomial parameters, usually determined using the least squares method.

[0024] Furthermore, step (4) specifically includes the following steps:

[0025] Taking the plot as the basic geographical unit, the average value of the time series remote sensing data is used as the characteristic value, and the characteristic curve of each year is linearly interpolated at a time interval of 1 day. Then, the Savitaky-Golay filter is used to reduce the temporal noise of the curve. The polynomial order and time window size are set to 4 and 25 respectively, and the vegetation index time series curve data set of the plot unit is obtained. Among them, the vegetation index time series curve of the plot data used to construct the remote sensing estimation model is used as the standard vegetation index time series curve of crops.

[0026] Further, step (5) specifically includes the following steps:

[0027] According to the crop calendar data, the early growth range of crops is limited, and then the beginning of the crop growth range is identified; the minimum points of the vegetation index time series are extracted in the corresponding time series, and the number of minimum points is counted; if the number of minimum points is 0, the midpoint of the time series is defined as the beginning of the crop growth range by default; if the number of minimum points is 1, the minimum point is the beginning of the crop growth range; if the number of minimum points is greater than 1, the number of local minimum values ​​in the sequence whose left maximum value is greater than the right maximum value is counted, and the maximum value is taken if there is no maximum value; if the number of minimum values ​​that meet the conditions is 0, the earliest minimum point is selected as the beginning of the crop growth range; if the number of minimum values ​​that meet the conditions is 1, the minimum point is the beginning of the crop growth range; if the number of minimum values ​​that meet the conditions is greater than 1, the earliest minimum point that meets the conditions is selected as the beginning of the crop growth range;

[0028] According to the crop calendar data, the late growth range of crops is limited, and then the end of the crop growth range is identified; the minimum points of the vegetation index time series are extracted in the corresponding time series, and the number of minimum points is counted; if the number of minimum points is 0, the midpoint of the time series is defined as the end of the crop growth range by default; if the number of minimum points is 1, the minimum point is the end of the crop growth range; if the number of minimum points is greater than 1, the number of local minimum values ​​in the sequence whose left maximum value is greater than the right maximum value is counted, and the maximum value is taken if there is no maximum value; if the number of minimum values ​​that meet the conditions is 0, the earliest minimum point is selected as the end of the crop growth range; if the number of minimum values ​​that meet the conditions is 1, the minimum point is the end of the crop growth range; if the number of minimum values ​​that meet the conditions is greater than 1, the earliest minimum point that meets the conditions is selected as the end of the crop growth range.

[0029] Preferably, step (6) specifically includes the following steps:

[0030] The crop standard vegetation index time series curve and the vegetation index time series curve of the target plot are extracted, and the beginning and end of the crop growth range are used as the two segmentation points of the time series curve; the midpoint of the crop growth range of the target plot is selected as the third segmentation point of the vegetation index time series curve of the plot, and the third segmentation point of the crop standard vegetation index time series curve is set to the growing days corresponding to the midpoint of the crop growth range of the target plot; the segmented time series is used as the input of the dynamic time warping algorithm to obtain the matching path, and the matching paths of all segments are integrated.

[0031] Furthermore, the dynamic time warping algorithm in step (6) is:

[0032] Assume that the crop standard vegetation index time series curve and the target plot vegetation index time series curve are and Time Series and , , , the distance matrix constructed from these two time series :

[0033] (4)

[0034] In the formula It is the basis distance between two sequences. Generally, the absolute value of the difference between the node values ​​of the two sequences is taken as the basis distance of DTW, that is, ;

[0035] The cumulative distance matrix can be derived from the distance matrix , For any element in , whose value is Zhongcong arrive The minimum cumulative distance is usually calculated by recursively accumulating the minimum distance;

[0036] (5)

[0037] (6)

[0038] The dynamic time warping algorithm minimizes the cumulative distance through dynamic programming. During the dynamic programming process, the best matching path can be obtained by backtracking. , the matching path is used to define and The mapping between is the length of the matching path, No. The elements are defined as .

[0039] Further, step (7) specifically includes the following steps:

[0040] According to the matching path, matching the vegetation index time series curve of the target plot to the standard vegetation index time series curve of the crop;

[0041] If a single point of the vegetation index time series curve of the target plot corresponds to a single point of the standard vegetation index time series curve of the crop, then the point on the standard vegetation index time series curve of the crop is the matching point of the point on the vegetation index time series curve of the target plot;

[0042] If a single point of the target plot vegetation index time series curve corresponds to multiple points of the crop standard vegetation index time series curve, the average point of the points on the crop standard vegetation index time series curve is the matching point of the point on the target plot vegetation index time series curve;

[0043] If multiple points of the target plot vegetation index time series curve correspond to a single point of the crop standard vegetation index time series curve, then all points on the crop standard vegetation index time series curve are matching points of multiple points on the target plot vegetation index time series curve;

[0044] The physiological parameters of crops in the target plots are estimated based on curve matching results, vegetation index and remote sensing estimation model.

[0045] Based on the same inventive concept, the present invention discloses a crop physiological parameter remote sensing time series estimation system based on phenological segmented DTW, comprising:

[0046] The preprocessing module is used to preprocess the time series remote sensing images to obtain the processed time series remote sensing data;

[0047] The index calculation module is used to calculate the vegetation index based on the time series remote sensing data to obtain the vegetation index time series data set;

[0048] An estimation model building module is used to build a remote sensing estimation model for estimating crop physiological parameters by combining the measured data of crop samples;

[0049] The time series curve construction module is used to construct the crop standard vegetation index time series curve and the vegetation index time series curve dataset respectively with the plot as the basic geographical unit;

[0050] A growth time determination module is used to determine the time range of crop growth in the target plot based on the vegetation index time series curve;

[0051] The matching path acquisition module is used to construct a phenological segmented time weighted dynamic algorithm based on the dynamic time warping algorithm, and obtain the crop time series matching path using the vegetation index time series curve as input;

[0052] The parameter estimation module is used to adjust the time parameters of the target plot crops according to the obtained matching path, and estimate the crop physiological parameters based on the remote sensing estimation model and the time series vegetation index.

[0053] The present invention provides a computer storage medium, which stores a computer program. The computer program is run by a processor to execute the above-mentioned crop physiological parameter remote sensing time series estimation method based on phenological segmentation DTW.

[0054] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: the present invention takes into account the crop growth range and local matching, applies the DTW algorithm to the research on crop physiological parameter estimation of time series remote sensing images, and realizes high-precision time series estimation of crop physiological parameters in different agricultural scenarios; the present invention aims to overcome the problem of low accuracy of remote sensing time series estimation of crop physiological parameters caused by phenological mismatch, and realizes crop time series matching in different agricultural scenarios and high-precision time series estimation of crop physiological parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic diagram of the process of the present invention;

[0056] Figure 2 The phenological differences in different regions and years of the embodiments of the present invention;

[0057] Figure 3 A schematic diagram of crop growth range extraction according to an embodiment of the present invention;

[0058] Figure 4 A schematic diagram of model performance evaluation according to an embodiment of the present invention;

[0059] Figure 5Schematic diagram of estimated LAI timing changes according to an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0061] Example 1: This example uses microwave remote sensing data as an example to estimate the rice leaf area index. The observation data comes from the dual-polarization SAR data and supporting ground observation data collected in the Changshu Rice Experimental Area in 2019 and the Yingtan Rice Experimental Area in 2017, 2019, and 2020. Among them, the data from the Changshu Rice Experimental Area in 2019 is used to establish the model, and the data from the Yingtan Rice Experimental Area in 2017, 2019, and 2020 is used for model testing. Figure 2 As shown, the rice phenological calendars of the two locations in different years are displayed.

[0062] like Figure 1 As shown, the present invention discloses a remote sensing time series estimation method for crop physiological parameters based on phenological segmentation DTW, comprising the following steps:

[0063] (1) Preprocess the time series remote sensing images to obtain processed time series remote sensing data;

[0064] Since the time series remote sensing images are microwave remote sensing images, the preprocessing includes multi-view processing, speckle filtering, geocoding and radiometric calibration;

[0065] (2) Based on the time series remote sensing data, the vegetation index is calculated to obtain a vegetation index time series data set;

[0066] (3) Based on the 2019 data of Changshu area and combined with the measured data of crop samples, a remote sensing estimation model for estimating crop physiological parameters was constructed;

[0067] Step (3) specifically includes the following steps:

[0068] The conventional least squares method is used to describe the relationship between crop physiological parameters and vegetation index in each growth period, and polynomials are used to describe the relationship between model parameters and time, so as to construct a remote sensing estimation model, which is expressed as follows:

[0069] (1)

[0070] (2)

[0071] (3)

[0072] In the formula Represents physiological parameters, is the vegetation index, k and bare the coefficients and intercepts of the least squares regression model for different growth periods, is the number of days in the year, n is the order of the polynomial, and are polynomial parameters, usually determined using the least squares method;

[0073] (4) Taking plots as the basic geographical units, we constructed the crop standard vegetation index time series curve and the vegetation index time series curve dataset respectively. The vegetation index time series curve of Changshu area in 2019 is the crop standard vegetation index time series curve;

[0074] Step (4) specifically includes the following steps:

[0075] Taking the plot as the basic geographical unit, the average value of the time series remote sensing data is used as the characteristic value, and the characteristic curve of each year is linearly interpolated at a time interval of 1 day. Then, the Savitaky-Golay filter is used to reduce the temporal noise of the curve. The polynomial order and time window size are set to 4 and 25 respectively, and the vegetation index time series curve data set of the plot unit is obtained; among them, the vegetation index time series curve of Changshu area in 2019 is used as the standard vegetation index time series curve of crops;

[0076] (5) Determine the time range of crop growth in the target plot based on the vegetation index time series curve;

[0077] Step (5) specifically includes the following steps:

[0078] According to the crop calendar data, the early growth range of crops is limited, and then the beginning of the crop growth range is identified; the minimum points of the vegetation index time series are extracted in the corresponding time series, and the number of minimum points is counted; if the number of minimum points is 0, the midpoint of the time series is defined as the beginning of the crop growth range by default; if the number of minimum points is 1, the minimum point is the beginning of the crop growth range; if the number of minimum points is greater than 1, the number of local minimum values ​​in the sequence whose left maximum value is greater than the right maximum value is counted, and the maximum value is taken if there is no maximum value; if the number of minimum values ​​that meet the conditions is 0, the earliest minimum point is selected as the beginning of the crop growth range; if the number of minimum values ​​that meet the conditions is 1, the minimum point is the beginning of the crop growth range; if the number of minimum values ​​that meet the conditions is greater than 1, the earliest minimum point that meets the conditions is selected as the beginning of the crop growth range;

[0079] According to the crop calendar data, the late growth range of crops is limited, and then the end of the crop growth range is identified; the minimum points of the vegetation index time series are extracted in the corresponding time series, and the number of minimum points is counted; if the number of minimum points is 0, the midpoint of the time series is defined as the end of the crop growth range by default; if the number of minimum points is 1, the minimum point is the end of the crop growth range; if the number of minimum points is greater than 1, the number of local minimum values ​​in the sequence whose left maximum value is greater than the right maximum value is counted, and the maximum value is taken if there is no maximum value; if the number of minimum values ​​that meet the conditions is 0, the earliest minimum point is selected as the end of the crop growth range; if the number of minimum values ​​that meet the conditions is 1, the minimum point is the end of the crop growth range; if the number of minimum values ​​that meet the conditions is greater than 1, the earliest minimum point that meets the conditions is selected as the end of the crop growth range; Figure 3 As shown, the results of crop growth range extraction in Yingtan area in each year are shown;

[0080] (6) Based on the dynamic time warping algorithm, a phenological segmented time-weighted dynamic algorithm was constructed, and the crop time series matching path was obtained using the vegetation index time series curve as input;

[0081] Step (6) specifically includes the following steps:

[0082] Extract the crop standard vegetation index time series curve and the vegetation index time series curve of the target plot, and use the beginning and end of the crop growth range as the two segmentation points of the time series curve; select the midpoint of the crop growth range of the target plot as the third segmentation point of the vegetation index time series curve of the plot, and set the third segmentation point of the crop standard vegetation index time series curve to the growing days corresponding to the midpoint of the crop growth range of the target plot; use the segmented time series as the input of the dynamic time warping algorithm to obtain the matching path, and integrate all segmented matching paths;

[0083] The dynamic time warping algorithm, i.e., the DTW algorithm, is:

[0084] Assume that the crop standard vegetation index time series curve and the target plot vegetation index time series curve are and Time Series and , , , the distance matrix constructed from these two time series :

[0085] (4)

[0086] In the formula It is the basis distance between two sequences. Generally, the absolute value of the difference between the node values ​​of the two sequences is taken as the basis distance of DTW, that is, ;

[0087] The cumulative distance matrix can be derived from the distance matrix ,for Any element in , whose value is Zhongcong arrive The minimum cumulative distance is usually calculated by recursively accumulating the minimum distance;

[0088] (5)

[0089] (6)

[0090] The dynamic time warping algorithm minimizes the cumulative distance through dynamic programming. During the dynamic programming process, the best matching path can be obtained by backtracking. , the matching path is used to define and The mapping between is the length of the matching path, No. The elements are defined as ;

[0091] (7) Adjust the time parameters of the target plot crops according to the obtained matching path, and estimate the crop physiological parameters based on the remote sensing estimation model and time series vegetation index;

[0092] According to the matching path, matching the vegetation index time series curve of the target plot to the standard vegetation index time series curve of the crop;

[0093] If a single point of the vegetation index time series curve of the target plot corresponds to a single point of the standard vegetation index time series curve of the crop, then the point on the standard vegetation index time series curve of the crop is the matching point of the point on the vegetation index time series curve of the target plot;

[0094] If a single point of the target plot vegetation index time series curve corresponds to multiple points of the crop standard vegetation index time series curve, the average point of the points on the crop standard vegetation index time series curve is the matching point of the point on the target plot vegetation index time series curve;

[0095] If multiple points of the target plot vegetation index time series curve correspond to a single point of the crop standard vegetation index time series curve, then all points on the crop standard vegetation index time series curve are matching points of multiple points on the target plot vegetation index time series curve;

[0096] The physiological parameters of crops in the target plots are estimated based on curve matching results, vegetation index and remote sensing estimation model.

[0097] In this embodiment, if Figure 4As shown in the figure, the LAI inversion accuracy in Yingtan area in 2017, 2018 and 2020 is shown, and the conventional method is compared. The results show that the crop physiological parameter estimation method based on phenological segmented DTW proposed in the present invention can maintain a high accuracy (R 2 =0.8085, RMSE=0.9831). Figure 5 As shown, the LAI time series results of Yingtan area in 2017, 2018 and 2020 are shown. It can be seen that the method proposed in the present invention can achieve high-precision time series estimation of crop physiological parameters in different agricultural scenarios.

[0098] Embodiment 2: The present invention discloses a crop physiological parameter remote sensing time series estimation system based on phenological segmented DTW, comprising:

[0099] The preprocessing module is used to preprocess the time series remote sensing images to obtain the processed time series remote sensing data;

[0100] Since the time series remote sensing images are microwave remote sensing images, the preprocessing includes multi-view processing, speckle filtering, geocoding and radiometric calibration;

[0101] The index calculation module is used to calculate the vegetation index based on the time series remote sensing data to obtain the vegetation index time series data set;

[0102] The estimation model construction module is used to combine the measured data of crop samples to construct a remote sensing estimation model for estimating crop physiological parameters. The estimation model construction module uses the conventional least squares method to describe the relationship between crop physiological parameters and vegetation index in each growth period, and uses polynomials to describe the relationship between model parameters and time, so as to construct a remote sensing estimation model. The expression is:

[0103] (1)

[0104] (2)

[0105] (3)

[0106] In the formula Represents physiological parameters, is the vegetation index, k and b are the coefficients and intercepts of the least squares regression model for different growth periods, is the number of days in the year, n is the order of the polynomial, and are polynomial parameters, usually determined using the least squares method;

[0107] The time series curve construction module is used to construct the crop standard vegetation index time series curve and the vegetation index time series curve dataset respectively, taking the plot as the basic geographical unit; the vegetation index time series curve in Changshu area in 2019 is the crop standard vegetation index time series curve;

[0108] In the time series curve construction module, the plot is used as the basic geographical unit, the average value of the time series remote sensing data is used as the characteristic value, and the characteristic curve of each year is linearly interpolated at a time interval of 1 day. Then, the Savitaky-Golay filter is used to reduce the temporal noise of the curve. The polynomial order and time window size are set to 4 and 25 respectively, and the vegetation index time series curve data set of the plot unit is obtained; among them, the vegetation index time series curve of Changshu area in 2019 is used as the standard vegetation index time series curve of crops;

[0109] The growth time determination module is used to determine the time range of crop growth in the target plot based on the vegetation index time series curve; in the growth time determination module, the early growth range of crops is limited according to the crop calendar data, and then the beginning of the crop growth range is identified; the minimum value points of the vegetation index time series are extracted in the corresponding time series, and the number of minimum value points is counted; if the number of minimum value points is 0, the midpoint of the time series is defined as the beginning of the crop growth range by default; if the number of minimum value points is 1, the minimum value point is the beginning of the crop growth range; if the number of minimum value points is greater than 1, the number of local minimum values ​​in the sequence whose left maximum value is greater than the right maximum value is counted, and the maximum value is taken if there is no maximum value; if the number of minimum values ​​that meet the conditions is 0, the earliest minimum value point is selected as the beginning of the crop growth range; if the number of minimum values ​​that meet the conditions is 1, the minimum value point is the beginning of the crop growth range; if the number of minimum values ​​that meet the conditions is greater than 1, the earliest minimum value point that meets the conditions is selected as the beginning of the crop growth range;

[0110] According to the crop calendar data, the late growth range of crops is limited, and then the end of the crop growth range is identified; the minimum points of the vegetation index time series are extracted in the corresponding time series, and the number of minimum points is counted; if the number of minimum points is 0, the midpoint of the time series is defined as the end of the crop growth range by default; if the number of minimum points is 1, the minimum point is the end of the crop growth range; if the number of minimum points is greater than 1, the number of local minimum values ​​in the sequence whose left maximum value is greater than the right maximum value is counted, and the maximum value is taken if there is no maximum value; if the number of minimum values ​​that meet the conditions is 0, the earliest minimum point is selected as the end of the crop growth range; if the number of minimum values ​​that meet the conditions is 1, the minimum point is the end of the crop growth range; if the number of minimum values ​​that meet the conditions is greater than 1, the earliest minimum point that meets the conditions is selected as the end of the crop growth range; Figure 3As shown, the results of crop growth range extraction in Yingtan area in each year are shown;

[0111] The matching path acquisition module is used to construct a phenological segmented time weighted dynamic algorithm based on the dynamic time warping algorithm, and obtain the crop time series matching path using the vegetation index time series curve as input;

[0112] In the matching path acquisition module, the crop standard vegetation index time series curve and the vegetation index time series curve of the target plot are extracted, and the beginning and end of the crop growth range are used as the two segmentation points of the time series curve; the midpoint of the crop growth range of the target plot is selected as the third segmentation point of the vegetation index time series curve of the plot, and the third segmentation point of the crop standard vegetation index time series curve is set to the growth days corresponding to the midpoint of the crop growth range of the target plot; the segmented time series is used as the input of the dynamic time warping algorithm to obtain the matching path, and all segmented matching paths are integrated;

[0113] The dynamic time warping algorithm, i.e., the DTW algorithm, is:

[0114] Assume that the crop standard vegetation index time series curve and the target plot vegetation index time series curve are and Time Series and , , , the distance matrix constructed from these two time series :

[0115] (4)

[0116] In the formula It is the basis distance between two sequences. Generally, the absolute value of the difference between the node values ​​of the two sequences is taken as the basis distance of DTW, that is, ;

[0117] The cumulative distance matrix can be derived from the distance matrix ,for Any element in , whose value is Zhongcong arrive The minimum cumulative distance is usually calculated by recursively accumulating the minimum distance;

[0118] (5)

[0119] (6)

[0120] The dynamic time warping algorithm minimizes the cumulative distance through dynamic programming. During the dynamic programming process, the best matching path can be obtained by backtracking. , the matching path is used to define and The mapping between is the length of the matching path, No. The elements are defined as ;

[0121] A parameter estimation module is used to adjust the time parameters of the target plot crops according to the obtained matching path, and estimate the crop physiological parameters based on the remote sensing estimation model and the time series vegetation index;

[0122] In the parameter estimation module, according to the matching path, the vegetation index time series curve of the target plot is matched to the standard vegetation index time series curve of the crop;

[0123] If a single point of the vegetation index time series curve of the target plot corresponds to a single point of the standard vegetation index time series curve of the crop, then the point on the standard vegetation index time series curve of the crop is the matching point of the point on the vegetation index time series curve of the target plot;

[0124] If a single point of the target plot vegetation index time series curve corresponds to multiple points of the crop standard vegetation index time series curve, the average point of the points on the crop standard vegetation index time series curve is the matching point of the point on the target plot vegetation index time series curve;

[0125] If multiple points of the target plot vegetation index time series curve correspond to a single point of the crop standard vegetation index time series curve, then all points on the crop standard vegetation index time series curve are matching points of multiple points on the target plot vegetation index time series curve;

[0126] The physiological parameters of crops in the target plots are estimated based on curve matching results, vegetation index and remote sensing estimation model.

[0127] Example 3: Corresponding to the method of Example 1 of the present invention, Example 3 of the present invention also provides a computer storage medium for data collection and reception, wherein the computer storage medium stores a computer program, and the computer program is run by a processor to execute a remote sensing time series estimation method for crop physiological parameters based on phenological segmented DTW provided in Example 1 of the present invention.

[0128] Each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware, or in the form of software functional unit, or in the form of a combination of software and hardware.

[0129] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: a mobile hard disk, a USB flash drive, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk, and other media that can store program codes.

[0130] The present invention provides a method, system and storage medium for remote sensing time series estimation of crop physiological parameters based on phenological segmented DTW. The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the protection scope of the present invention. All components not specified in this embodiment can be implemented by existing technologies.

Claims

1. A remote sensing time series estimation method for crop physiological parameters based on phenological segmented DTW, characterized in that: The following steps are involved: (1) Preprocess the time series remote sensing images to obtain processed time series remote sensing data; (2) Based on the time series remote sensing data, the vegetation index is calculated to obtain a vegetation index time series data set; (3) Combined with the measured data of crop samples, a remote sensing estimation model for estimating crop physiological parameters is constructed; (4) Taking plots as the basic geographical units, we constructed the crop standard vegetation index time series curve and vegetation index time series curve datasets respectively; (5) Determine the time range of crop growth in the target plot based on the vegetation index time series curve; (6) Based on the dynamic time warping algorithm, a phenological segmented time-weighted dynamic algorithm was constructed, and the crop time series matching path was obtained using the vegetation index time series curve as input; The step (6) specifically includes the following steps: Extract the crop standard vegetation index time series curve and the vegetation index time series curve of the target plot, and use the beginning and end of the crop growth range as the two segmentation points of the time series curve; select the midpoint of the crop growth range of the target plot as the third segmentation point of the vegetation index time series curve of the plot, and set the third segmentation point of the crop standard vegetation index time series curve to the growing days corresponding to the midpoint of the crop growth range of the target plot; use the segmented time series as the input of the dynamic time warping algorithm to obtain the matching path, and integrate all segmented matching paths; (7) Adjust the time parameters of the target plot crops according to the obtained matching path, and estimate the crop physiological parameters based on the remote sensing estimation model and time series vegetation index; The step (7) specifically includes the following steps: According to the matching path, matching the vegetation index time series curve of the target plot to the standard vegetation index time series curve of the crop; If a single point of the vegetation index time series curve of the target plot corresponds to a single point of the standard vegetation index time series curve of the crop, then the point on the standard vegetation index time series curve of the crop is the matching point of the point on the vegetation index time series curve of the target plot; If a single point of the target plot vegetation index time series curve corresponds to multiple points of the crop standard vegetation index time series curve, the average point of the points on the crop standard vegetation index time series curve is the matching point of the point on the target plot vegetation index time series curve; If multiple points of the target plot vegetation index time series curve correspond to a single point of the crop standard vegetation index time series curve, then all points on the crop standard vegetation index time series curve are matching points of multiple points on the target plot vegetation index time series curve; The physiological parameters of crops in the target plots are estimated based on curve matching results, vegetation index and remote sensing estimation model.

2. The method for remote sensing time series estimation of crop physiological parameters based on phenological segmented DTW according to claim 1, characterized in that: In step (1), if the time series remote sensing image is an optical remote sensing image, the preprocessing includes radiation calibration, atmospheric correction, geometric correction and image filtering; If the time series remote sensing image is a microwave remote sensing image, the preprocessing includes multi-view processing, speckle filtering, geocoding and radiometric calibration.

3. The method for remote sensing time series estimation of crop physiological parameters based on phenological segmented DTW according to claim 1, characterized in that: The step (3) specifically includes the following steps: The least squares method is used to describe the relationship between crop physiological parameters and vegetation index in each growth period, and polynomials are used to describe the relationship between model parameters and time, so as to construct a remote sensing estimation model, which is expressed as follows: （1） （2） （3） In the formula Represents physiological parameters, is the vegetation index, k and b are the coefficients and intercepts of the least squares regression model for different growth periods, is the number of days in the year, n is the order of the polynomial, and are polynomial parameters determined using the least squares method.

4. The method for remote sensing time series estimation of crop physiological parameters based on phenological segmented DTW according to claim 1, characterized in that: The step (4) specifically includes the following steps: Taking the plot as the basic geographical unit, the average value of the time series remote sensing data is used as the characteristic value, and the characteristic curve of each year is linearly interpolated at a time interval of 1 day. Then, the Savitaky-Golay filter is used to reduce the temporal noise of the curve. The polynomial order and time window size are set to 4 and 25 respectively, and the vegetation index time series curve data set of the plot unit is obtained. Among them, the vegetation index time series curve of the plot data used to construct the remote sensing estimation model is used as the standard vegetation index time series curve of crops.

5. The method for remote sensing time series estimation of crop physiological parameters based on phenological segmented DTW according to claim 1, characterized in that: The step (5) specifically includes the following steps: According to the crop calendar data, the early growth range of crops is limited, and then the beginning of the crop growth range is identified; the minimum points of the vegetation index time series are extracted in the corresponding time series, and the number of minimum points is counted; if the number of minimum points is 0, the midpoint of the time series is defined as the beginning of the crop growth range by default; if the number of minimum points is 1, the minimum point is the beginning of the crop growth range; if the number of minimum points is greater than 1, the number of local minimum values ​​in the sequence whose left maximum value is greater than the right maximum value is counted, and the maximum value is taken if there is no maximum value; if the number of minimum values ​​that meet the conditions is 0, the earliest minimum point is selected as the beginning of the crop growth range; if the number of minimum values ​​that meet the conditions is 1, the minimum point is the beginning of the crop growth range; if the number of minimum values ​​that meet the conditions is greater than 1, the earliest minimum point that meets the conditions is selected as the beginning of the crop growth range; Based on crop calendar data, the late crop growth range is defined to identify the end of the crop growth range; The minimum points of the vegetation index time series are extracted from the corresponding time series, and the number of minimum points is counted; if the number of minimum points is 0, the midpoint of the time series is defined as the end of the crop growth range by default; if the number of minimum points is 1, the minimum point is the end of the crop growth range; if the number of minimum points is greater than 1, the number of local minimum values ​​in the sequence whose left maximum value is greater than the right maximum value is counted, and the maximum value is taken if there is no maximum value; if the number of minimum values ​​that meet the conditions is 0, the earliest minimum point is selected as the end of the crop growth range; if the number of minimum values ​​that meet the conditions is 1, the minimum point is the end of the crop growth range; if the number of minimum values ​​that meet the conditions is greater than 1, the earliest minimum point that meets the conditions is selected as the end of the crop growth range.

6. The method for remote sensing time series estimation of crop physiological parameters based on phenological segmented DTW according to claim 1, characterized in that: The dynamic time warping algorithm in step (6) is: Assume that the crop standard vegetation index time series curve and the target plot vegetation index time series curve are and Time Series and , , , the distance matrix constructed from these two time series : （4） In the formula is the base distance between the two sequences, and the absolute value of the difference between the node values ​​of the two sequences is taken as the base distance of DTW, that is, ; Deriving the cumulative distance matrix from the distance matrix ,for Any element in , whose value is Zhongcong arrive The minimum cumulative distance is calculated by recursively accumulating the minimum distance; （5） （6） The dynamic time warping algorithm minimizes the cumulative distance through dynamic programming. In the dynamic programming process, the best matching path is obtained by backtracking. Used to define and The mapping between is the length of the matching path, No. The elements are defined as .

7. A crop physiological parameter remote sensing time series estimation system based on phenological segmented DTW, characterized in that: include: The preprocessing module is used to preprocess the time series remote sensing images to obtain the processed time series remote sensing data; The index calculation module is used to calculate the vegetation index based on the time series remote sensing data to obtain the vegetation index time series data set; An estimation model building module is used to build a remote sensing estimation model for estimating crop physiological parameters by combining the measured data of crop samples; The time series curve construction module is used to construct the crop standard vegetation index time series curve and the vegetation index time series curve dataset respectively with the plot as the basic geographical unit; A growth time determination module is used to determine the time range of crop growth in the target plot based on the vegetation index time series curve; The matching path acquisition module is used to construct a phenological segmented time weighted dynamic algorithm based on the dynamic time warping algorithm, and obtain the crop time series matching path using the vegetation index time series curve as input; The matching path acquisition module extracts the crop standard vegetation index time series curve and the vegetation index time series curve of the target plot, and takes the beginning and end of the crop growth range as the two segmentation points of the time series curve; the midpoint of the crop growth range of the target plot is selected as the third segmentation point of the vegetation index time series curve of the plot, and the third segmentation point of the crop standard vegetation index time series curve is set to the growing days corresponding to the midpoint of the crop growth range of the target plot; the segmented time series is used as the input of the dynamic time warping algorithm to obtain the matching path, and all segmented matching paths are integrated; A parameter estimation module is used to adjust the time parameters of the target plot crops according to the obtained matching path, and estimate the crop physiological parameters based on the remote sensing estimation model and the time series vegetation index; The parameter estimation module matches the target plot vegetation index time series curve to the crop standard vegetation index time series curve according to the matching path; If a single point of the vegetation index time series curve of the target plot corresponds to a single point of the standard vegetation index time series curve of the crop, then the point on the standard vegetation index time series curve of the crop is the matching point of the point on the vegetation index time series curve of the target plot; If a single point of the target plot vegetation index time series curve corresponds to multiple points of the crop standard vegetation index time series curve, the average point of the points on the crop standard vegetation index time series curve is the matching point of the point on the target plot vegetation index time series curve; If multiple points of the target plot vegetation index time series curve correspond to a single point of the crop standard vegetation index time series curve, then all points on the crop standard vegetation index time series curve are matching points of multiple points on the target plot vegetation index time series curve; The physiological parameters of crops in the target plots are estimated based on curve matching results, vegetation index and remote sensing estimation model.

8. A computer storage medium, characterized in that: The computer storage medium stores a computer program, which is executed by a processor to perform the method according to any one of claims 1 to 6.

Images