Target crop identification method and device, electronic equipment and storage medium

Abstract

The application provides a target crop identification method and device, electronic equipment and storage medium, and relates to the technical field of target identification. The method comprises the following steps: acquiring multiple radar images of a region where a crop to be identified is located, and extracting at least three kinds of time series data in the radar images; fitting each time series data by using a nonlinear regression model to obtain model parameters corresponding to each time series data; determining a canopy structure difference parameter of the crop to be identified and a non-target crop according to the model parameters; and determining an identification result of the crop to be identified based on the canopy structure difference parameter. The application determines the identification result of the crop to be identified based on the canopy structure difference parameter, effectively avoids the problem that a large numerical fluctuation occurs due to factors such as regional characteristics, and leads to poor target crop identification effect, and thus can effectively improve the accuracy of the target crop identification result.

Description

Technical Field

[0001] This invention relates to the field of target recognition technology, and in particular to a method, apparatus, electronic device, and storage medium for identifying target crops. Background Technology

[0002] With the rapid development of intelligent and precision agriculture globally, target crop identification technology is playing an increasingly important role in agricultural production management, resource optimization, and sustainable development. For example, identifying cotton allows for timely and accurate acquisition of information on the spatial distribution and changes in cotton planting, which is crucial for yield estimation, management strategy formulation, and promoting the sustainable development of the cotton industry.

[0003] Existing target crop identification methods typically identify target crops by comparing the numerical differences between radar polarization data of target crops and non-target crops. However, relying solely on numerical differences is susceptible to large numerical fluctuations due to factors such as regional characteristics, resulting in poor target crop identification performance. Summary of the Invention

[0004] This invention provides a target crop identification method, device, electronic device, and storage medium to solve the technical problem that existing technologies rely solely on numerical differences, which are easily affected by factors such as regional characteristics, resulting in large numerical fluctuations and thus poor target crop identification performance.

[0005] This invention provides a method for identifying target crops, comprising: Acquire multiple radar images of the area where the crop to be identified is located, and extract at least three types of time-series data from the radar images; A nonlinear regression model is used to fit each of the time series data to obtain the model parameters corresponding to each time series data. Based on the model parameters, determine the canopy structure difference parameters between the crop to be identified and the non-target crop; The identification result of the crop to be identified is determined based on the canopy structure difference parameters.

[0006] According to a target crop identification method provided by the present invention, the extraction of at least three types of time-series data from the radar image includes: Candidate radar images are selected from multiple radar images based on the daily clear sky index. Using the location of the object to be identified as the center point, a buffer zone is determined in the candidate radar image; Extract the timing data from the buffer.

[0007] According to the target crop identification method provided by the present invention, the time series data includes dual-polarization time series data and polarization ratio time series data, and the canopy structure difference parameters include the total energy of backscattered signal and canopy time series characteristics; The step of determining the canopy structure difference parameters between the target crop and non-target crops based on the model parameters includes: Based on the model parameters obtained by fitting the dual-polarization time series data, the total energy of the backscattered signal of the crop to be identified is determined. Based on the model parameters obtained by fitting the polarization ratio time series data, the canopy time series characteristics of the crop to be identified are determined.

[0008] According to a target crop identification method provided by the present invention, the dual-polarization time series data includes VV time series data and VH time series data; The determination of the total energy of the backscattered signal of the crop to be identified, based on the model parameters obtained by fitting the dual-polarization time-series data, includes: Based on the model parameters obtained by fitting the VV time series data, the first phenological parameters of the crop to be identified are determined, and the area under the first curve of the first fitted curve is determined based on the first phenological parameters; wherein, the first fitted curve is obtained by substituting the model parameters obtained by fitting the VV time series data into the nonlinear regression model; Based on the model parameters obtained by fitting the VH time series data, the second phenological parameters of the crop to be identified are determined, and the area under the second curve of the second sound field curve is determined based on the second phenological parameters; wherein, the second fitted curve is obtained by substituting the model parameters obtained by fitting the VH time series data into the nonlinear regression model; The sum of the areas under the first curve and the areas under the second curve is taken as the total energy of the backscattered signal of the crop to be identified.

[0009] According to a target crop identification method provided by the present invention, the canopy temporal characteristics are determined based on model parameters fitted from the polarization ratio time-series data, including: The peak date characteristics of the curve are determined based on the peak date of the third fitted curve; wherein, the third fitted curve is obtained by substituting the model parameters obtained by fitting the polarization ratio time series data into the nonlinear regression model; The growth period of the crop to be identified in the third fitted curve is determined, wherein the growth period includes the non-growing season, the early growing season, and the late growing season; Determine the slope characteristics of the curve based on the slope characteristics of two adjacent growth stages; The peak date feature and the slope feature of the curve are used as canopy time series features.

[0010] According to a target crop identification method provided by the present invention, the slope feature includes a first slope feature and a second slope feature, and the step of determining the curve slope feature based on the slope features of two adjacent growth stages includes: Select a corresponding data point in each growth stage; Based on data points from two adjacent growth periods, determine the first slope feature and the second slope feature; The curve slope characteristics are determined based on the first slope characteristic and the second slope characteristic.

[0011] According to a target crop identification method provided by the present invention, determining the identification result of the crop to be identified based on the canopy structure difference parameters includes: The product of the total energy of the backscattered signal, the peak date feature of the curve, and the slope feature of the curve is used as the target crop identification index. If the target crop identification index is greater than or equal to a preset threshold, the crop to be identified is determined to be the target crop.

[0012] According to the present invention, a method for identifying a target crop is provided, wherein the target crop is cotton.

[0013] The present invention also provides a target crop identification device, comprising: The time-series data extraction module is used to acquire multiple radar images of the area where the crop to be identified is located, and extract at least three types of time-series data from the radar images; The model parameter determination module is used to fit each of the time series data using a nonlinear regression model to obtain the model parameters corresponding to each time series data. The canopy structure difference parameter determination module is used to determine the canopy structure difference parameters between the crop to be identified and the non-target crop based on the model parameters. The target crop identification module is used to determine the identification result of the crop to be identified based on the canopy structure difference parameters.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the target crop identification method as described above.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the target crop identification method as described above.

[0016] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the target crop identification method as described above.

[0017] This invention uses a nonlinear regression model to fit time-series data, obtains corresponding model parameters, and determines the canopy structure difference parameters between the target crop and non-target crops based on the model parameters. The identification result of the target crop is determined by these canopy structure difference parameters. This eliminates the need to rely on a large number of samples and avoids relying solely on numerical differences to identify the target crop. As a result, it can effectively avoid the problem of large numerical fluctuations caused by regional characteristics and other factors, which leads to poor identification results of the target crop. This can effectively improve the accuracy of the target crop identification results.

[0018] Furthermore, based on the dense canopy characteristics of cotton and the enhanced surface scattering caused by positive diurnal motion, but with limited impact on volume scattering, this invention utilizes the cumulative values ​​of the VV and VH backscattering coefficients throughout the entire growth period, as well as the later peak period and suppressed rate of rise shown by their ratio curve, thus exhibiting a unique curve characteristic of gradual rise and steep drop. By using the product of the total energy of the backscattered signal, the peak date characteristic of the curve, and the slope characteristic of the curve as the target crop identification index, and multiplying multiple indicators together as the identification index, this invention can accurately distinguish between cotton and non-cotton. Moreover, by combining the aforementioned unique intrinsic physiological characteristics of cotton, this invention requires a smaller sample size and has strong generalization ability, achieving good transferability between different regions. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a schematic flowchart of the target crop identification method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram showing the corresponding locations of phenological parameters provided in an embodiment of the present invention; Figure 3 This is the distribution of key parameters for various crops provided in the embodiments of the present invention; Figure 4 The distribution of phenological parameters for fitting the cotton VV curve and VH curve provided in the embodiments of the present invention; Figure 5 The slope characteristics of cotton, wheat, corn and other crops provided in the embodiments of the present invention; Figure 6A schematic diagram of the DCI histogram distribution results for cotton, wheat, corn and other crops provided in embodiments of the present invention; Figure 7 This is a schematic diagram of a target crop identification device provided in an embodiment of the present invention; Figure 8 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0022] Figure 1 This is a flowchart illustrating the target crop identification method provided by the present invention, as shown below. Figure 1 As shown, the method includes the following: S1. Acquire multiple radar images of the area where the crop to be identified is located, and extract at least three types of time-series data from the radar images; The target crops in this embodiment of the invention may include sunflowers and cotton, which belong to the Malvaceae family. Cotton is used as an example of the target crop in this embodiment of the invention.

[0023] In this embodiment of the invention, the acquired radar images are radar satellite image data of the crop to be identified during its growth period, using both VV (vertical transmission / vertical reception) and VH (vertical transmission / horizontal reception) polarization methods.

[0024] S2. A nonlinear regression model is used to fit each of the time series data to obtain the model parameters corresponding to each time series data. In this embodiment of the invention, a 7-parameter logistic model can be used to fit each time series data to obtain 7 model parameters corresponding to each time series data.

[0025] S3. Determine the canopy structure difference parameters between the crop to be identified and the non-target crop based on the model parameters; In this embodiment of the invention, the canopy structure difference parameters include the total energy of the backscattered signal and the temporal characteristics of the canopy. This embodiment can generate a corresponding fitted curve using model parameters, and use the area under the fitted curve and the curve characteristics as the canopy structure difference coefficient to characterize the canopy structure differences between the target crop and non-target crops, thereby accurately identifying the target crop.

[0026] S4. Determine the identification result of the crop to be identified based on the canopy structure difference parameters.

[0027] In this embodiment of the invention, a target crop identification index can be calculated based on the canopy structure difference parameters, and then the crop to be identified can be determined as the target crop based on the comparison result between the target crop identification index and the preset identification index threshold.

[0028] This invention uses a nonlinear regression model to fit time-series data, obtaining corresponding model parameters. Based on these parameters, it determines the canopy structure difference parameters between the target crop and non-target crops. The identification result of the target crop is then determined using these canopy structure difference parameters. This eliminates the need to rely on a large number of samples and avoids relying solely on numerical differences to identify the target crop. Consequently, it effectively avoids the problem of large numerical fluctuations caused by regional characteristics and other factors, which can lead to poor target crop identification results. This significantly improves the accuracy of the target crop identification results.

[0029] In one embodiment, step S1, extracting at least three types of time-series data from the radar image, includes: S11. Based on the daily Clearness Index (CI), candidate radar images are selected from multiple radar images. In this embodiment of the invention, a preset threshold for the Clear Sky Index (CSI) can be set, for example, a CSI preset threshold of 0.6. The CSI is used as a quantitative indicator to judge weather conditions. If the CSI > 0.6, the data is retained; if the CSI ≤ 0.6, the data is discarded, thereby filtering out radar images that meet the criteria. The CSI is used to quantify the degree of clear weather and is used to filter and retain radar image data under clear weather conditions.

[0030] This invention, through calculating the daily clear sky index, filters candidate radar images from multiple radar images, ensuring that the selected radar images were acquired under clear sky conditions. This effectively ensures the impact of the cotton's positive diurnal movement on the canopy structure during image acquisition, thereby improving the accuracy of target crop identification.

[0031] S12. Using the position of the object to be identified as the center point, determine a buffer zone in the candidate radar image; In this embodiment of the invention, a circular region with a preset radius can be generated as a buffer zone, using the location of the object to be identified as the center point. For example, a circular region with a radius of 30m, 35m, or 40m can be generated as a buffer zone to improve the stability of the data fitting results and thus improve the accuracy of target crop identification.

[0032] S13. Extract the timing data from the buffer.

[0033] In this embodiment of the invention, the timing data in the buffer may include dual-polarization timing data and polarization ratio timing data. The dual-polarization timing data includes VV (Vertical-Vertical polarization) timing data and VH (Vertical-Horizontal polarization) timing data, and the polarization ratio timing data is VH / VV timing data.

[0034] In one embodiment, S2, a nonlinear regression model is used to fit each of the time series data to obtain the model parameters corresponding to each time series data, including: A 7-parameter logistic model was used to fit each type of time series data to obtain the model parameters corresponding to each type of time series data. For example, model parameters were generated for the three types of time series data: VV time series data, VH time series data, and VH / VV time series data.

[0035] In one embodiment, the time-series data includes dual-polarization time-series data and polarization ratio time-series data, and the canopy structure difference parameters include the total energy of the backscattered signal and canopy time-series characteristics; Step S3: Based on the model parameters, determine the canopy structure difference parameters between the target crop and non-target crops, including: S31. Based on the model parameters obtained by fitting the dual-polarization time series data, determine the total energy of the backscattered signal of the crop to be identified; In this embodiment of the invention, the total energy of the backscattered signal is the sum of the energies of the VV-polarized backscattering intensity and the VH-polarized backscattering intensity. For example, if the target crop is cotton, its VH-polarized backscattering coefficient is higher than that of non-cotton crops due to its higher planting density and denser canopy. Simultaneously, cotton's positive diurnal movement and dynamic leaf adjustment result in a denser canopy structure in the direction of sunlight, further increasing the difference in its VV-polarized backscattering coefficient compared to non-cotton crops.

[0036] In this embodiment of the invention, model parameters obtained by fitting dual-polarization time series data can be substituted into a nonlinear model to obtain a corresponding fitting curve. The total energy of the backscattered signal of the crop to be identified can be determined based on the area under the curve in the fitting curve. The higher the polarization backscattering intensity, the larger the area under the curve.

[0037] This invention uses the total energy of the backscattered signal of the crop to be identified as one of the indicators for identifying the target crop, and can accurately identify the target crop based on the difference in canopy structure between the target crop and non-target crops.

[0038] S32. Based on the model parameters obtained by fitting the polarization ratio time series data, determine the canopy time series characteristics of the crop to be identified.

[0039] In this embodiment of the invention, taking into account the enhanced surface scattering of the canopy caused by the positive diurnal movement of cotton leaves, but with limited impact on volume scattering, the polarization ratio time series curve initially shows a slow upward trend due to the inhibition of the positive diurnal movement, and subsequently decays rapidly due to the influence of defoliants, forming a unique slowly rising and steeply falling curve characteristic. The canopy time series characteristics include the peak date characteristic and the slope characteristic. The peak date characteristic reflects the time lag of the peak value of the ratio time series curve caused by changes in the canopy structure due to the positive diurnal movement of cotton. The slope characteristic quantitatively characterizes the slowly rising and steeply falling shape of the cotton ratio time series curve.

[0040] In this embodiment of the invention, the model parameters obtained by fitting the time series data based on the polarization ratio can be substituted into the nonlinear regression model to obtain the corresponding fitting curve, and the peak date feature and curve slope feature can be determined according to the peak value and slope of the fitting curve.

[0041] By using the canopy temporal characteristics of the crop to be identified as one of the indicators for identifying the target crop, this invention can further accurately identify the target crop based on the differences in canopy structure between the target crop and non-target crops.

[0042] In one embodiment, the dual-polarization timing data includes VV timing data and VH timing data; Step S31, determining the total energy of the backscattered signal of the crop to be identified based on the model parameters obtained by fitting the dual-polarization time-series data, includes: S311. Based on the model parameters obtained by fitting the VV time series data, determine the first phenological parameters of the crop to be identified, and determine the area under the first curve of the first fitted curve based on the first phenological parameters; wherein, the first fitted curve is obtained by substituting the model parameters obtained by fitting the VV time series data into the nonlinear regression model. In this embodiment of the invention, phenological parameters include the start of growing season (SOS), the start of peak growth (SOP), the end of peak growth (EOP), and the end of growing season (EOS). Please refer to Table 1, which is an example table of phenological parameters provided in this embodiment of the invention.

[0043] Table 1: Example Table of Phenological Parameters phenological parameters formula Growing season start time [m3 - k - m4] Start time of peak period [m3+k-m4] End of peak period [m5 - k - m6] End of growing season <![CDATA[m5+k·m6]]> In this embodiment of the invention, a 7-parameter logistic model is used as the fitting model. The model parameters obtained by fitting each time series data are 7, including m1, m2, m3, m4, m5, m6, and m7. The expression for the 7-parameter logistic model is as follows: .

[0044] Where t represents the input time-series data. m1 is the curve baseline value, used to characterize the average level of the radar backscattering coefficient during the crop's non-growing season; m2 is the curve amplitude intercept, used to characterize the potential maximum increase in crop growth season relative to the background value (the actual summer peak is determined by both m2 and m7); m3 is the horizontal shift of the first set of logistic functions, controlling the time position of the inflection point in the rapid growth stage (SOS→SOP) (the smaller the value, the earlier the inflection point); m4 is the width parameter of the first set of exponential terms, used to control the steepness of the curve in the rapid growth stage (the slope at the inflection point is proportional to 1 / m4, the smaller the absolute value, the faster the growth rate); m5 is the horizontal shift of the second set of logistic functions, used to control the time position of the inflection point in the senescence stage (EOP→EOS) (the smaller the value, the earlier the inflection point); m6 is the width parameter of the second set of exponential terms, used to control the steepness of the curve in the senescence stage (the slope at the inflection point is proportional to 1 / m4, the smaller the absolute value, the faster the growth rate); (m6 is directly proportional; the smaller the absolute value, the faster the aging rate). m7 is the slope of the curve amplitude, which changes the shape of the "plateau period" between the two sets of logistic functions, making the peak value of the curve change linearly with time t. The model uses two sets of logistic functions to characterize the dynamic growth process of crops during their growth period, i.e., m3 and m4 control the first half, and m5 and m6 control the second half.

[0045] In this embodiment of the invention, m1, m2, and m7 only control the vertical amplitude of the curve, i.e., the baseline, amplitude intercept, and amplitude slope, and are independent of the time node, so their impact on the division of phenological stages is negligible. Therefore, this embodiment of the invention determines the phenological parameters through four parameters: m3, m4, m5, and m6. Specifically: By combining m3 and m4, the critical time point when crops transition from the sowing stage to significant growth can be accurately located. Specifically, a dynamically adjustable threshold k=2.5 is introduced, with m3–2.5·m4 taken as the start time of the growing season and m3+2.5·m4 taken as the start time of the vigorous growth period. The dynamically adjustable threshold can be set and adjusted according to actual needs or historical data to filter noise and adapt to the growth characteristics of the crop.

[0046] It should be noted that, in this embodiment of the invention, the value of k affects the percentage of the function value. When k is 2.5, for the growing season, the starting time is defined as the point where the function value rises from almost 0 to 7.6%, and the ending time is determined as the point where the function value reaches 92.4%. For the vigorous growth period, the value of k also affects the percentage of the function value.

[0047] This invention, through the combination of m5 and m6, can accurately pinpoint the critical time point when a crop transitions from vigorous growth to decline, i.e., the end of the vigorous growth period. Specifically, a dynamically adjustable threshold k=2.5 is introduced, with m5 – 2.5·m6 taken as the end time of the vigorous growth period; and m5 + 2.5·m6 taken as the end time of the growing season.

[0048] In this embodiment of the invention, the area under the first curve is the area of ​​the curve formed by the start time of the growing season to the end time of the growing season and the baseline.

[0049] Please see Figure 2 This is a schematic diagram of the corresponding positions of phenological parameters provided in an embodiment of the present invention.

[0050] S312. Based on the model parameters obtained by fitting the VH time series data, determine the second phenological parameters of the crop to be identified, and determine the area under the second curve of the second fitted curve based on the second phenological parameters; wherein, the second fitted curve is obtained by substituting the model parameters obtained by fitting the VH time series data into the nonlinear regression model; In this embodiment of the invention, the approach to determining the area under the second curve is the same as that for determining the area under the first curve: first, the second phenological parameter is determined based on the model parameters obtained by fitting the VH time series data, and then the area under the second curve is determined in the second fitted curve by referring to the phenological parameter example table shown in Table 1.

[0051] S313. The sum of the area under the first curve and the area under the second curve is taken as the total energy of the backscattered signal of the crop to be identified.

[0052] This invention, based on the dense canopy characteristics of cotton and the enhanced surface scattering caused by positive diurnal motion but with limited impact on volume scattering, utilizes the cumulative values ​​of the VV and VH backscattering coefficients throughout the entire growth period, as well as the later peak period and suppressed rate of increase shown in their ratio curve, thus exhibiting a unique curve characteristic of gradual rise and steep fall. By using the product of the total energy of the backscattered signal, the peak date characteristic of the curve, and the slope characteristic of the curve as the target crop identification index, and multiplying multiple indicators together as the identification index, this invention can accurately distinguish between cotton and non-cotton crops. Furthermore, by combining the aforementioned unique intrinsic physiological characteristics of cotton, this invention requires a smaller sample size and has strong generalization ability, achieving good transferability across different regions.

[0053] In one embodiment, step S32, determining the canopy time-series characteristics based on the model parameters fitted from the polarization ratio time-series data, includes: S321. Determine the peak date characteristics of the curve based on the peak date of the third fitted curve; wherein, the third fitted curve is obtained by substituting the model parameters obtained by fitting the polarization ratio time series data into the nonlinear regression model; In this embodiment of the invention, the DOY (Day of Year) corresponding to the maximum y-value of the third fitted curve can be used as the peak date feature of the curve. The expression for the peak date feature of the curve is as follows: .

[0054] S322. Determine the growth period of the crop to be identified in the third fitted curve, wherein the growth period includes the non-growing season, the early growing season, and the late growing season; In this embodiment of the invention, the third phenological parameter can be determined based on the model parameters obtained by fitting the time-series data of the polarization ratio. The determination method can be the same as the phenological parameter determination method shown in Table 1. After determining the third phenological parameter, the growth period can be further determined based on the third phenological parameter. For example, the time period from sowing to the start of the growing season can be determined as the non-growing season, the time period from the start of the growing season to the midpoint of the vigorous growth period can be determined as the early growing season, and the time period from the midpoint of the vigorous growth period to the end of the growing season can be determined as the late growing season.

[0055] S323. Determine the curve slope characteristics based on the slope characteristics of two adjacent growth stages, including: S3231. Select a corresponding data point in each growth stage; S3232. Based on the data points of two adjacent growth periods, determine the first slope feature and the second slope feature; In this embodiment of the invention, the non-growing season is adjacent to the early growing season, and the early growing season is adjacent to the late growing season. Three data points can be set correspondingly in the non-growing season, the early growing season, and the late growing season, labeled A, B, and C respectively. The first slope feature is the slope of the line connecting AB, and the second slope feature is the slope of the line connecting BC. In this embodiment of the invention, suitable locations can be selected as data points from each growth stage.

[0056] S3233. Determine the curve slope feature based on the first slope feature and the second slope feature.

[0057] In this embodiment of the invention, the slope feature of the curve can be formed by combining the slope of line AB and the slope of line BC. The expression is as follows: .

[0058] It should be noted that, in this embodiment of the invention, for the main crops in Xinjiang, the positive or negative sign of the slopes of the AB and BC lines is one of the key features for identifying cotton. Since wheat is sown in April and harvested in July, its peak growth period is relatively early. When some wheat enters its senescence stage, cotton and corn are still in a vigorous growth stage. Therefore, the slope of the AB line can be used to distinguish some wheat. Furthermore, due to the influence of positive diurnal motion and harvesting methods, the peak date of the cotton curve is later than that of non-cotton crop curves. When non-cotton crop curves enter their senescence stage, cotton reaches its peak growth. Therefore, the slope of the BC line of the cotton curve is positive, while that of the non-cotton crop curve is negative. This embodiment of the invention uses the joint constraint of the AB-BC slope as a discriminant feature, which can effectively improve the accuracy of cotton identification.

[0059] S324. The peak date feature of the curve and the slope feature of the curve are used as canopy time series features.

[0060] In this embodiment of the invention, since cotton adjusts its orientation according to the sun's position throughout the day, the third fitting curve exhibits a peak date lag and positive first and second slopes. This allows cotton to be distinguished from non-cotton crops such as wheat and corn from a physical mechanism perspective, avoiding the empirical defects of traditional methods that rely on numerical differences.

[0061] In one embodiment, the slope feature includes a first slope feature and a second slope feature. S323, Determining the curve slope feature based on the slope features of two adjacent growth stages includes: In one embodiment, step S4, determining the identification result of the crop to be identified based on the canopy structure difference parameters, includes: The product of the total energy of the backscattered signal, the peak date feature of the curve, and the slope feature of the curve is used as the target crop identification index. If the target crop identification index is greater than or equal to a preset threshold, the crop to be identified is determined to be the target crop.

[0062] In this embodiment of the invention, taking cotton as the target crop as an example, the expression of the target crop identification index DCI (Diaheliotropic Cotton Index) is as follows: in, This represents the total energy of the backscattered signal. The area under the first fitted curve is denoted as . This represents the area under the second fitted curve. The peak date characteristic of the curve, This represents the slope characteristic of the curve.

[0063] In one embodiment, to Figures 3-5 This invention provides a method for identifying a target crop, with cotton as an example of the target crop.

[0064] Please see Figure 3 Figure 3 shows the distribution of key parameters for cotton, wheat, corn, and other crops extracted using existing sample data, as provided in this embodiment of the invention. As can be seen from Figure 3, at the EOP (Expiration Point), both the VV (Volume Value) and VH (Volume Value) time-series data for cotton are higher than those for non-cotton crops. Further comparison of the areas under the time-series curves for different crops reveals that, due to the higher polarization values ​​during the vigorous growth period of cotton, the area under the cotton time-series curve is larger than that for non-cotton crops. Observing the fitting results of the VH / VV curves, it can be found that the peak date for non-cotton crops is earlier than that for cotton and is more prone to a bimodal distribution, while the gradual increase and steep decrease in the cotton curve makes the peak date close to the EOP date of the VH curve (i.e., the second fitted curve) and more concentrated.

[0065] Figure 4 The figure shows the distribution of phenological parameters fitted to the VV curve (first fitted curve) and VH curve (second fitted curve) of cotton crop provided in this embodiment of the invention. Figure 4 shows that the stage changes of the VH curve occur later than those of the VV curve. The SOP of the VH curve is about 15 days later than the SOP of the VV curve, and about 10 days later than the EOP. The rapid growth stage (SOS to SOP) of the VV curve is shorter than that of the VH curve, while the senescence stage (EOP to EOS) has a longer time span. This is because the VV time series data is sensitive to changes in the morphology and moisture content of cotton during the boll opening stage, causing its decline to occur earlier.

[0066] Figure 5 The slope characteristics of cotton, wheat, corn, and other crops are provided for embodiments of the present invention. From... Figure 5 As can be seen, wheat's peak growth period occurs earlier. While cotton and corn are still in their vigorous growth phase, some wheat has already entered the senescence stage. Therefore, the slope of the AB line can be used to distinguish some wheat crops. For the BC line, due to the positive diurnal movement of cotton leaves and the application of defoliants at harvest, the cotton curve reaches its peak later than non-cotton crops. Therefore, the slope of the BC line is mostly positive on the cotton curve, while the slope of non-cotton crops is mostly negative at this time, as they have already entered the senescence stage. Thus, using the combined AB-BC slope constraint can serve as a discriminant feature between cotton and non-cotton crops. Considering the phenological differences between crops in northern and southern Xinjiang, the specific locations of three constraint points for the slope feature were searched and set for both northern and southern Xinjiang. The locations of points A, B, and C in northern Xinjiang are DOY120, DOY219, and DOY269, respectively, while the locations of points A, B, and C in southern Xinjiang are DOY120, DOY165, and DOY286, respectively.

[0067] Figure 6 The DCI histogram distribution results for cotton, wheat, corn, and other crops provided for embodiments of the present invention. From... Figure 6 As can be seen, the DCI value of cotton samples is higher than that of non-cotton samples, and the mean DCI value of cotton samples is significantly higher than that of non-cotton samples. The mean DCI values ​​for cotton, wheat, corn, and other crops are 24.72, 2.88, 5.49, and 9.68, respectively. Therefore, cotton and non-cotton crops can be distinguished by setting a DCI threshold.

[0068] Implementing the embodiments of the present invention has the following beneficial effects: This invention uses a nonlinear regression model to fit time-series data, obtaining corresponding model parameters. Based on these parameters, it determines the canopy structure difference parameters between the target crop and non-target crops. The identification result of the target crop is then determined using these canopy structure difference parameters. This eliminates the need to rely on a large number of samples and avoids relying solely on numerical differences to identify the target crop. Consequently, it effectively avoids the problem of large numerical fluctuations caused by regional characteristics and other factors, which can lead to poor target crop identification results. This significantly improves the accuracy of the target crop identification results.

[0069] Furthermore, based on the dense canopy characteristics of cotton and the enhanced surface scattering caused by positive diurnal motion, but with limited impact on volume scattering, this invention utilizes the cumulative values ​​of the VV and VH backscattering coefficients throughout the entire growth period, as well as the later peak period and suppressed rate of increase shown in their ratio curve, thus exhibiting a unique curve characteristic of gradual rise and steep fall. By using the product of the total energy of the backscattered signal, the peak date characteristic of the curve, and the slope characteristic of the curve as the target crop identification index, and multiplying multiple indicators together as the identification index, this invention can accurately distinguish between cotton and non-cotton crops. Moreover, by combining the aforementioned unique intrinsic physiological characteristics of cotton, this invention requires a smaller sample size and has strong generalization ability, achieving good transferability across different regions.

[0070] Furthermore, in this embodiment of the invention, the product of the total energy of the backscattered signal, the peak date feature of the curve, and the slope feature of the curve is used as the target crop identification index. The multiplication of multiple indicators is used as the identification index, which can further distinguish between target crops and non-target crops, thereby further improving the accuracy of target crop identification.

[0071] The target crop identification device provided by the present invention is described below. The target crop identification device described below and the target crop identification method described above can be referred to in correspondence.

[0072] Please see Figure 7 This invention provides a target crop identification device, comprising: The time-series data extraction module 710 is used to acquire multiple radar images of the area where the crop to be identified is located, and extract at least three types of time-series data from the radar images. The model parameter determination module 720 is used to fit each of the time series data using a nonlinear regression model to obtain the model parameters corresponding to each time series data. The canopy structure difference parameter determination module 730 is used to determine the canopy structure difference parameters between the crop to be identified and the non-target crop based on the model parameters. The target crop identification module 740 is used to determine the identification result of the crop to be identified based on the canopy structure difference parameters.

[0073] In one embodiment, the extraction of at least three types of time-series data from the radar image includes: Candidate radar images are selected from multiple radar images based on the daily clear sky index. Using the location of the object to be identified as the center point, a buffer zone is determined in the candidate radar image; Extract the timing data from the buffer.

[0074] In one embodiment, the time-series data includes dual-polarization time-series data and polarization ratio time-series data, and the canopy structure difference parameters include the total energy of the backscattered signal and canopy time-series characteristics; The step of determining the canopy structure difference parameters between the target crop and non-target crops based on the model parameters includes: Based on the model parameters obtained by fitting the dual-polarization time series data, the total energy of the backscattered signal of the crop to be identified is determined. Based on the model parameters obtained by fitting the polarization ratio time series data, the canopy time series characteristics of the crop to be identified are determined.

[0075] In one embodiment, the dual-polarization timing data includes VV timing data and VH timing data; The determination of the total energy of the backscattered signal of the crop to be identified, based on the model parameters obtained by fitting the dual-polarization time-series data, includes: Based on the model parameters obtained by fitting the VV time series data, the first phenological parameters of the crop to be identified are determined, and the area under the first curve of the first fitted curve is determined based on the first phenological parameters; wherein, the first fitted curve is obtained by substituting the model parameters obtained by fitting the VV time series data into the nonlinear regression model; Based on the model parameters obtained by fitting the VH time series data, the second phenological parameters of the crop to be identified are determined, and the area under the second curve of the second fitted curve is determined based on the second phenological parameters; wherein, the second fitted curve is obtained by substituting the model parameters obtained by fitting the VH time series data into the nonlinear regression model; The sum of the areas under the first curve and the areas under the second curve is taken as the total energy of the backscattered signal of the crop to be identified.

[0076] In one embodiment, determining the canopy time-series characteristics based on the model parameters fitted from the polarization ratio time-series data includes: The peak date characteristics of the curve are determined based on the peak date of the third fitted curve; wherein, the third fitted curve is obtained by substituting the model parameters obtained by fitting the polarization ratio time series data into the nonlinear regression model; The growth period of the crop to be identified in the third fitted curve is determined, wherein the growth period includes the non-growing season, the early growing season, and the late growing season; Determine the slope characteristics of the curve based on the slope characteristics of two adjacent growth stages; The peak date feature and the slope feature of the curve are used as canopy time series features.

[0077] In one embodiment, the slope feature includes a first slope feature and a second slope feature, and determining the curve slope feature based on the slope features of two adjacent growth stages includes: Select a corresponding data point in each growth stage; Based on data points from two adjacent growth periods, determine the first slope feature and the second slope feature; The curve slope characteristics are determined based on the first slope characteristic and the second slope characteristic.

[0078] In one embodiment, determining the identification result of the crop to be identified based on the canopy structure difference parameter includes: The product of the total energy of the backscattered signal, the peak date feature of the curve, and the slope feature of the curve is used as the target crop identification index. If the target crop identification index is greater than or equal to a preset threshold, the crop to be identified is determined to be the target crop.

[0079] Figure 8 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 8 As shown, the electronic device may include: a processor 810, a communications interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communications interface 820, and the memory 830 communicate with each other via the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute a target crop identification method, which includes: Acquire multiple radar images of the area where the crop to be identified is located, and extract at least three types of time-series data from the radar images; A nonlinear regression model is used to fit each of the time series data to obtain the model parameters corresponding to each time series data. Based on the model parameters, determine the canopy structure difference parameters between the crop to be identified and the non-target crop; The identification result of the crop to be identified is determined based on the canopy structure difference parameters.

[0080] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0081] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer is able to execute a target crop identification method provided by the above methods, the method comprising: Acquire multiple radar images of the area where the crop to be identified is located, and extract at least three types of time-series data from the radar images; A nonlinear regression model is used to fit each of the time series data to obtain the model parameters corresponding to each time series data. Based on the model parameters, determine the canopy structure difference parameters between the crop to be identified and the non-target crop; The identification result of the crop to be identified is determined based on the canopy structure difference parameters.

[0082] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform a target crop identification method provided by the methods described above, the method comprising: Acquire multiple radar images of the area where the crop to be identified is located, and extract at least three types of time-series data from the radar images; A nonlinear regression model is used to fit each of the time series data to obtain the model parameters corresponding to each time series data. Based on the model parameters, determine the canopy structure difference parameters between the crop to be identified and the non-target crop; The identification result of the crop to be identified is determined based on the canopy structure difference parameters.

[0083] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0084] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying a target crop, characterized in that, include: Acquire multiple radar images of the area where the crop to be identified is located, and extract at least three types of time-series data from the radar images; A nonlinear regression model is used to fit each of the time series data to obtain the model parameters corresponding to each time series data. Based on the model parameters, determine the canopy structure difference parameters between the crop to be identified and the non-target crop; The identification result of the crop to be identified is determined based on the canopy structure difference parameters.

2. The target crop identification method as described in claim 1, characterized in that, The extraction of at least three types of time-series data from the radar image includes: Candidate radar images are selected from multiple radar images based on the daily clear sky index. Using the location of the object to be identified as the center point, a buffer zone is determined in the candidate radar image; Extract the timing data from the buffer.

3. The target crop identification method as described in claim 1, characterized in that, The time series data includes dual-polarization time series data and polarization ratio time series data, and the canopy structure difference parameters include the total energy of backscattered signals and canopy time series characteristics. The step of determining the canopy structure difference parameters between the target crop and non-target crops based on the model parameters includes: Based on the model parameters obtained by fitting the dual-polarization time series data, the total energy of the backscattered signal of the crop to be identified is determined. Based on the model parameters obtained by fitting the polarization ratio time series data, the canopy time series characteristics of the crop to be identified are determined.

4. The target crop identification method as described in claim 3, characterized in that, The dual-polarization timing data includes VV timing data and VH timing data; The determination of the total energy of the backscattered signal of the crop to be identified, based on the model parameters obtained by fitting the dual-polarization time-series data, includes: Based on the model parameters obtained by fitting the VV time series data, the first phenological parameters of the crop to be identified are determined, and the area under the first curve of the first fitted curve is determined based on the first phenological parameters; wherein, the first fitted curve is obtained by substituting the model parameters obtained by fitting the VV time series data into the nonlinear regression model; Based on the model parameters obtained by fitting the VH time series data, the second phenological parameters of the crop to be identified are determined, and the area under the second curve of the second fitted curve is determined based on the second phenological parameters; wherein, the second fitted curve is obtained by substituting the model parameters obtained by fitting the VH time series data into the nonlinear regression model; The sum of the areas under the first curve and the areas under the second curve is taken as the total energy of the backscattered signal of the crop to be identified.

5. The target crop identification method as described in claim 4, characterized in that, The step of determining the canopy time-series characteristics based on the model parameters fitted from the polarization ratio time-series data includes: The peak date characteristics of the curve are determined based on the peak date of the third fitted curve; wherein, the third fitted curve is obtained by substituting the model parameters obtained by fitting the polarization ratio time series data into the nonlinear regression model; The growth period of the crop to be identified in the third fitted curve is determined, wherein the growth period includes the non-growing season, the early growing season, and the late growing season; Determine the slope characteristics of the curve based on the slope characteristics of two adjacent growth stages; The peak date feature and the slope feature of the curve are used as canopy time series features.

6. The target crop identification method as described in claim 5, characterized in that, The slope feature includes a first slope feature and a second slope feature. Determining the curve slope feature based on the slope features of two adjacent growth stages includes: Select a corresponding data point in each growth stage; Based on data points from two adjacent growth periods, determine the first slope feature and the second slope feature; The curve slope characteristics are determined based on the first slope characteristic and the second slope characteristic.

7. The target crop identification method as described in claim 5, characterized in that, The determination of the identification result of the crop to be identified based on the canopy structure difference parameters includes: The product of the total energy of the backscattered signal, the peak date feature of the curve, and the slope feature of the curve is used as the target crop identification index. If the target crop identification index is greater than or equal to a preset threshold, the crop to be identified is determined to be the target crop.

8. The target crop identification method according to any one of claims 1-7, characterized in that, The target crop is cotton.

9. A target crop identification device, characterized in that, include: The time-series data extraction module is used to acquire multiple radar images of the area where the crop to be identified is located, and extract at least three types of time-series data from the radar images; The model parameter determination module is used to fit each of the time series data using a nonlinear regression model to obtain the model parameters corresponding to each time series data. The canopy structure difference parameter determination module is used to determine the canopy structure difference parameters between the crop to be identified and the non-target crop based on the model parameters. The target crop identification module is used to determine the identification result of the crop to be identified based on the canopy structure difference parameters.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the target crop identification method as described in any one of claims 1 to 8.

11. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the target crop identification method as described in any one of claims 1 to 8.

12. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the target crop identification method as described in any one of claims 1 to 8.

Images