Automatic sugarcane identification method cooperating with soil-green-light time sequence linkage

Abstract

The invention provides an automatic sugarcane identification method cooperating with soil-green-light time sequence linkage, and the method comprises the steps: carrying out the detection of a cultivation disturbance event: quantifying the dynamic change of a soil structure in a spring cultivation period based on the radar texture heterogeneity induced by sugarcane ridging cultivation; canopy steady-state event detection: based on the scattering mechanism stability of the sugarcane closed canopy, quantifying the biomass steady state in the elongation mature period through normalized variance; group light competition event detection: based on chlorophyll absorption competitive change caused by sugarcane close planting, time sequence extreme value positioning and cumulative intensity analysis are carried out; multi-source collaborative decision making: coupling radar structure stability and optical physiological response characteristics, and fusing event detection results through linkage discriminant indexes; outputting the spatial distribution of the sugarcanes; wherein the detection of the farming disturbance event, the canopy steady state event and the group light competition event is based on the time sequence data linkage analysis of the whole growth period of the sugarcane.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of agricultural remote sensing, and particularly relates to a sugarcane automatic identification method based on soil-green-light time sequence linkage. BACKGROUND

[0002] As the core crop of the global sugar industry supply system, the planting distribution dynamics of sugarcane has a significant impact on the stable operation of the international sugar market. In the regional agricultural economic pattern, although the statistical area proportion of sugarcane is limited, it shows significant economic value and ecological service function per unit area in core production areas such as South China, and becomes an important characteristic crop supporting local agricultural economy. The current macro monitoring of sugarcane planting areas mainly relies on remote sensing technology, but still faces some technical bottlenecks in practical application.

[0003] As the main data source of traditional crop identification, optical remote sensing has some basic limitations in sugarcane monitoring. Sugarcane and other high-stalk crops with similar photosynthetic pathways show obvious reflection characteristic overlap in a certain spectral range, especially in the canopy closure stage of vigorous crop growth, and key vegetation indices are difficult to provide effective separability. This spectral confusion effect restricts the improvement of classification accuracy based on a single optical data source. At the same time, in the main production areas of sugarcane, frequent cloud cover in the rainy season leads to uncontrollable acquisition period of high-quality optical images, and continuous observation sequence in the key phenology period is easily interrupted, which directly affects the integrity of the dynamic analysis of the growth process.

[0004] Although the time sequence analysis technology of crop growth period characteristics has been widely used in large-scale crop identification, it faces special challenges in sugarcane monitoring. Due to the influence of ratoon regeneration characteristics and regional differentiated harvesting strategies, the length of sugarcane phenology cycle in different production areas is obviously different, and the time span of key growth stages varies significantly. Traditional phenology models are difficult to adapt to such spatio-temporal heterogeneity, resulting in a decrease in the comparability of time sequence curves and restricting the effectiveness of the construction of unified identification models across regions. Especially in high-proportion ratoon fields, the growth period variation caused by plant regeneration process further increases the uncertainty of time sequence characteristics.

[0005] In order to overcome the limitations of optical data, multi-source remote sensing collaborative monitoring has gradually become the direction of technological development. Radar satellites can theoretically make up for the lack of optical data under cloudy and rainy conditions due to their all-weather observation advantage. However, in actual application, there are natural differences between radar and optical data in terms of spatial and temporal resolution, imaging mechanism, etc., which makes effective collaboration face technical challenges. Current fusion methods are mostly limited to data superposition or simple decision integration, and have not fully released the inherent complementary value of radar scattering information and optical physiological parameters. Especially in expressing the collaborative law of the unique crown structure change and physiological response of sugarcane, the existing technical framework has not established a quantitative correlation model, which restricts the substantial breakthrough of identification accuracy.

[0006] Overall, the precise identification of sugarcane is urgent to break through three technical difficulties: crop physiological characteristics convergence leading to separability constraints, phenology dynamic response caused by agronomic practice differences, and algorithm adaptation bottleneck in multi-source data collaborative application. The existence of these basic problems makes the existing monitoring means still insufficient in the actual demand of sugar industry supply chain management. SUMMARY

[0007] In view of the defects and deficiencies of the prior art, the present application provides a kind of sugarcane automatic identification method and system of coordinated soil-green-light timing linkage, by analyzing the three types of growth events of sugarcane specific tillage disturbance, canopy steady state and group light competition, break through the dual technical bottleneck of high stalk crop spectrum confusion and phenology difference. First, based on the radar echo intensity mutation induced by sugarcane ridging tillage, the tillage disturbance event (soil) is quantified, and the soil stress effect is determined by combining the peak value and cumulative threshold of bare soil index. Second, the closed canopy scattering stability of sugarcane is captured by using the de-meaning variance model, and the biomass saturation stage is locked by using the exclusive steady state threshold. Further, according to the time sequence extreme value of the competitive change of chlorophyll absorption, the light competition window (light) is located, and the photosynthetic resource competition dynamics caused by dense canopy is analyzed. The core innovation lies in the construction of multi-source collaborative decision mechanism: through the antagonistic relationship model of radar structure stability and photosynthetic response, the three types of events are fused, the negative correlation between canopy steady state enhancement and photosynthetic inhibition is quantitatively expressed, and the detection window is dynamically calibrated by using the phenology adaptive algorithm, solving the time sequence misplacement problem caused by regional growth period difference. This method significantly improves the recognition robustness in cloudy areas, and realizes the precise mapping of sugarcane at plot to provincial scale.

[0008] The technical scheme specifically adopted by the application to solve its technical problems is:

[0009] A kind of sugarcane automatic identification method of coordinated soil-green-light timing linkage, comprising:

[0010] Tillage disturbance event detection: based on the radar texture heterogeneity induced by sugarcane ridging tillage, the dynamic change of soil structure in spring tillage period is quantified;

[0011] Canopy steady state event detection: based on the stability of closed canopy scattering mechanism of sugarcane, the biomass steady state in elongation mature period is quantified by normalized variance;

[0012] Group light competition event detection: based on the competitive change of chlorophyll absorption caused by dense planting of sugarcane, the time sequence extreme value is located and the cumulative intensity is analyzed;

[0013] Multi-source collaborative decision: coupling radar structure stability and photosynthetic response characteristics, the event detection results are fused by linkage discriminant index;

[0014] Output the spatial distribution of sugarcane;

[0015] The detection of the tillage disturbance event, the canopy steady-state event, and the population light competition event is based on linkage analysis of time series data of the sugarcane whole growth period.

[0016] Further, in the tillage disturbance event detection process:

[0017] The texture heterogeneity is quantified by a radar echo intensity mutation induced by tillage disturbance;

[0018] The mutation intensity satisfies a disturbance discrimination threshold specific to sugarcane planting;

[0019] The discrimination threshold is dynamically determined by a texture feature entropy change rate within a tillage period window.

[0020] Further, in the canopy steady-state event detection process:

[0021] The normalized variance is achieved by closed canopy scattering steady-state analysis of sugarcane;

[0022] An unmeant variance calculation model is adopted;

[0023] The steady-state determination threshold is an interval value of 0.3-0.55, which is related to the radar signal stability caused by the saturation of canopy biomass in the elongation and maturation period of sugarcane.

[0024] Further, in the multi-source collaborative decision-making process:

[0025] The linkage discrimination index is quantified by the antagonistic relationship between radar structural stability and photosynthetic physiological response;

[0026] The antagonistic relationship adopts a ratio operation model of radar and optical indexes;

[0027] The antagonistic relationship is manifested as a negative correlation between the strengthening of sugarcane canopy structure stability and the weakening of competitive chlorophyll absorption.

[0028] Further, before the tillage disturbance event detection, the canopy steady-state event detection, and the population light competition event detection, a data preprocessing step is further included:

[0029] Geocoding and terrain correction are performed on the radar time series data to ensure spatial alignment with optical data, and spatial tessellation and speckle suppression processing are performed;

[0030] Cloud mask and smoothing reconstruction are performed on the optical time series data.

[0031] Further, in the tillage disturbance event detection and the canopy steady-state event detection:

[0032] Based on the JM distance separability measure, the vertical-horizontal polarization ratio and the cross-polarization scattering entropy are selected to form the optimal polarization feature combination.

[0033] Further, the tillage disturbance event detection is associated with soil deficit event:

[0034] When the bare soil index peak value is greater than zero, and the cumulative bare soil index in the spring tillage time window exceeds a set threshold, it is determined that sugarcane planting induces soil stress effect.

[0035] Further, in the tillage disturbance event detection and the canopy steady state event detection:

[0036] The radar shape fingerprint index is used to analyze the phenological response shape in the whole growth period, including:

[0037] The tillage period signal drop shape is used to support the tillage disturbance event detection;

[0038] The elongation period signal stable shape is used to support the canopy steady state event detection;

[0039] The radar shape fingerprint index is generated by logical operation of the first-order difference of the time series signal.

[0040] Further, the time windows of the tillage disturbance event detection and the population light competition event detection are dynamically determined by the following methods respectively:

[0041] The tillage disturbance window: ±50 days range based on the day of the bare soil index peak value;

[0042] The light competition window: locking the double-peak extreme value point interval of the chlorophyll absorption rate, and partially overlapping with the canopy steady state event detection window.

[0043] And a system for implementing the method as described above, comprising:

[0044] The radar event perception module: used to perform the tillage disturbance event detection and the canopy steady state event detection;

[0045] The photosynthetic physiological inversion module: used to perform the population light competition event detection;

[0046] The multi-dimensional decision engine: used to perform multi-source collaborative decision;

[0047] The spatial distribution generation module: used to output the spatial distribution map of the sugarcane planting area.

[0048] And a computer device, comprising a memory, a processor and a computer program stored on the memory, the processor executes the computer program to implement the method as described above.

[0049] A non-transitory computer readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the method as described above.

[0050] Compared with the prior art, the present application and its preferred schemes at least include the following beneficial effects:

[0051] 1. Breakthrough of high-stalk crop spectrum confusion bottleneck

[0052] By coupling radar texture structure detection and optical chlorophyll response analysis, a "cultivation disturbance (soil)-canopy steady state (green)-light competition (light)" triple event synergistic decision mechanism is constructed, effectively separating the overlapping interference of the reflection spectrum of sugarcane and C4 crops such as corn and sorghum, and solving the misclassification problem caused by the convergence of physiological characteristics of traditional optical remote sensing.

[0053] 2. Overcome the constraints of phenological spatiotemporal heterogeneity

[0054] Based on the dynamic time window mechanism designed according to the sugarcane ratoon regeneration characteristics, the peak value of bare soil index and the extreme point of chlorophyll are adaptively locked to the cultivation period and light competition period, so that the key event detection is not affected by the regional growth period difference, and the universality of cross-regional identification is significantly improved.

[0055] 3. Enhance the monitoring robustness in cloudy areas

[0056] Fusion of radar all-weather observation advantage and optical physiological parameter sensitivity, under the condition of optical data missing rate in the core sugarcane area in the rainy season, still guarantee the continuous analysis ability of canopy steady state scattering characteristics (VH polarization) and light competition event (CARI index).

[0057] 4. Realize mechanism-driven precision decision

[0058] Innovatively construct the canopy steady state-light competition antagonism model (CSLCI), quantitatively characterize the photosynthetic inhibition effect caused by biomass saturation, deepen the understanding of the "sustained high biomass-phase low light efficiency" contradiction under sugarcane dense planting cultivation, and support the optimization of planting and harvesting strategy.

[0059] 5. Provide multi-level mapping adaptation capability

[0060] This method significantly improves the recognition robustness in cloudy areas, realizes the precision mapping of sugarcane from plot to provincial scale, and through large-scale empirical verification, its classification accuracy is significantly better than the benchmark level of existing technology, meeting the precision requirements of sugar industry supply chain management. BRIEF DESCRIPTION OF DRAWINGS

[0061] The present application will be further described in detail below in combination with the drawings and specific embodiments:

[0062] Figure 1 The design and implementation process of the embodiment of the present application is shown in the figure;

[0063] Figure 2 The VH time series signal diagram of sugarcane in the embodiment of the present application is shown in the figure;

[0064] Figure 3 DBSI timing signal data chart for an embodiment of the application;

[0065] Figure 4 CARI timing signal data chart for an embodiment of the application;

[0066] Figure 5 Sugarcane spatial distribution map for an embodiment of the application in a certain province A city and B city. DETAILED DESCRIPTION

[0067] In order to make the features and advantages of the present application more apparent, the following specific examples are provided for further illustration, and are described in detail as follows:

[0068] It should be noted that the following detailed description is exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0069] It is to be understood that the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. It is further understood that the terms "comprising," "including," and / or "containing" when used herein, do not exclude the presence of other elements. It is also understood that the terms "including" and / or "comprising" when used herein, specify the presence of stated features, integers, steps, operations, devices, components, and / or groups thereof, but do not preclude the presence or addition of one or more other features, integers, steps, operations, devices, components, and / or groups thereof.

[0070] In view of the deficiencies and challenges of the prior art, the present application breaks out the idea of evaluating the similarity of spectral index time series curve, comprehensively considers the site conditions and farming measures (soil) of sugarcane planting areas, the growth characteristics (green) of the persistent green leaf canopy of stem crops, and the light competition (light) of densely planted crop groups, and proposes a sugarcane automatic recognition scheme based on the synergistic soil-green-light time series linkage. In terms of sugarcane farming measures (soil), the formation of ridge and furrow on the ground surface after ridge cultivation measures leads to a significant decrease in radar backscattering coefficient; at the same time, the bare soil index value is high due to the increased reflectivity in the short-wave infrared light band. In terms of the growth characteristics (green) of the persistent green leaf canopy of sugarcane, the persistent green leaf canopy of sugarcane is covered by dense green leaves for a long time, resulting in a continuously high value of radar backscattering coefficient. In terms of the light competition (light) of densely planted crop groups, sugarcane is a typical densely planted crop, and the dense sugarcane leaves lead to poor light conditions for the lower leaves of the group, and reduce the chlorophyll absorption rate and photosynthetic rate of the group. The present application focuses on the multi-dimensional attributes of soil, green, and light of sugarcane, uses optical and radar time series data sets, and sequentially designs ridge cultivation, persistent green leaf canopy, and group light competition event detection methods, and then constructs a large-scale sugarcane automatic recognition method and system based on the time series linkage characteristics of soil-green-light in the growth cycle of sugarcane.

[0071] Considering the spectral feature similarity of high stem crops such as sugarcane and corn, it is difficult to effectively identify only by relying on optical data, especially in cloudy and rainy areas. The present application ingeniously integrates optical and radar data, comprehensively considers the site conditions and tillage measures (soil) of sugarcane planting areas, the growth characteristics (green) of the persistent green leaf canopy of stem crops, and the light competition of densely planted crops (light), and sequentially designs the sugarcane spring tillage measures, the elongation mature period green leaf canopy steady state, and the elongation mature period group light competition event detection method, and proposes a kind of sugarcane automatic identification method and system of soil-green-light time sequence linkage. The present application has the ability of explainability and cross-domain and cross-scale migration, and provides key support for the identification of characteristic economic crops-sugarcane in complex planting areas in tropical and subtropical regions. Based on multi-source features and mechanism-driven decision model, the present application constructs a multi-dimensional analysis framework of "tillage disturbance-green leaf canopy steady state-group light competition". The method does not need a large number of training samples, and can realize automatic identification of sugarcane and high-precision mapping in different regions.

[0072] The present application provides a construction process of a kind of sugarcane automatic identification method and system of soil-green-light time sequence linkage, and based on this, provides a sugarcane mapping method based on tillage disturbance-green leaf canopy steady state-group light competition multi-dimensional framework, including the following steps, as shown in Figure 1

[0073] Step S01, construct radar and optical image time sequence data set;

[0074] Step S02, establish preferred radar time sequence data based on crop radar time sequence separation degree;

[0075] Step S03, design spring tillage measures event detection module;

[0076] Step S04, design sugarcane elongation mature period green leaf canopy steady state detection module;

[0077] Step S05, design radar shape fingerprint index of multi-dimensional event time sequence linkage of sugarcane whole growth period;

[0078] Step S06, design spring soil deficit event detection module;

[0079] Step S07, design elongation mature period group light competition event detection module;

[0080] Step S08, design canopy steady state-light competition linkage discriminant index in cooperation with optical radar data;

[0081] Step S09, create a kind of sugarcane automatic identification method and system of soil-green-light time sequence linkage;

[0082] Step S10, construct sugarcane spatial distribution mapping product.​

[0083] As a preferred scheme of the embodiment, step S01 is specifically;

[0084] On the remote sensing cloud platform, all Sentinel-1 SAR and Sentinel-2 MSI image data meeting the conditions are screened out according to the research time period and the research area range; the Sentinel-1 VV and VH data are preprocessed, such as spatial inlaying, speckle filtering suppression and radiometric terrain correction, the original Sentinel-2 MSI image data are inlaid, cloud removed, cropped, linearly interpolated and Whittaker Smoother time series smoothed and reconstructed to obtain a 10-day maximized synthesized Sentinel-2 MSI image data set of the research area; the Sentinel-2 MSI image data set is used to establish a multi-dimensional spectral index time series data set of vegetation, soil and chlorophyll, including vegetation index EVI2, bare soil index DBSI and chlorophyll absorption rate index CARI (the CARI time series signal is as shown in Figure 4 The calculation formulas are as follows:

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091] Wherein, NIR represents the near-infrared band reflectivity, Red represents the red light band reflectivity, Green represents the green light band reflectivity, SWIR1 represents the short-wave infrared band reflectivity, and VRE1 represents the red edge band reflectivity; in step S01, the sugarcane growth period is dynamically acquired pixel by pixel, and the growth condition of the crop can be judged according to the fluctuation degree of the curve by analyzing the time series curve of the vegetation index EVI2. The value of EVI2 is positively correlated with the growth state of the crop, and the higher the value, the more vigorous the growth of the crop.

[0092] As a preferred scheme of the embodiment, in step S02, a multi-polarization time series feature set of the growth period is constructed based on the Sentinel-1 VH and VV data. The differences of sugarcane and other crops in polarization ratio, cross-polarization scattering entropy and other parameters are quantified by Jeffries-Matusita distance to establish a polarization feature separability evaluation model and construct an optimized radar time series data, specifically;

[0093] Most of the sugarcane production areas are located in the tropical and subtropical monsoon climate zone, which are affected by persistent cloud cover and atmospheric aerosol interference. Radar signals have the advantages of all-weather observation, vegetation structure sensitivity, and multi-polarization information, and have become an important tool for crop identification, especially in cloudy and rainy areas, complex crop classification and dynamic monitoring scenarios. Based on Sentinel-1 synthetic aperture radar data, radar signal time series data of different polarization modes and polarization combination modes are extracted. The polarization mode is determined by the vibration direction of the electric field vector during the electromagnetic wave propagation process. Among them, vertical transmission-vertical reception (VV) and vertical transmission-horizontal reception (VH) are the most typical dual-polarization modes of the synthetic aperture radar system: the former maintains the consistency of the electromagnetic wave polarization plane (both transmission and reception are vertical), and the latter generates cross-polarization components by orthogonal polarization reception. Based on the mathematical transformation of the polarization scattering matrix, linear combination (VV+VH) and ratio combination (VV / VH) and other derived polarization features can be further constructed. In addition, the Jeffries-Matusita distance (JM distance), which is a distance based on feature calculation of different class samples, is used to measure the separation degree of the class, and a separability analysis model of sugarcane and other crops is constructed to select the best radar polarization mode with the best separation degree. The calculation process can be represented as:

[0094]

[0095]

[0096] In the formula, represents the mean value of the sample feature, represents the variance of the sample feature, and B is an intermediate variable; wherein i=1,2.

[0097] As a preferred scheme of the present embodiment, step S03 constructs a ridge tillage event detection module based on radar texture parameters according to the characteristics of the dynamic change of soil structure after sugarcane ridge tillage in spring, and quantifies the radar texture feature heterogeneity characteristics induced by tillage disturbance, specifically:

[0098] According to the characteristics of the dynamic change of soil structure after sugarcane ridge tillage in spring, a radar gray level co-occurrence matrix is introduced to describe the matrix of image pixel gray level distribution, and the spatial relationship analysis between different pixels is revealed. Common texture features include contrast, entropy, correlation and energy, etc. The calculation formula is as follows:

[0099]

[0100]

[0101]

[0102]

[0103] Texture features such as contrast, entropy, correlation, and energy can be used to fully utilize the heterogeneity of radar texture features induced by ridged farmland from a texture perspective, starting from the degree of image grayscale difference, the complexity of image texture, the linear dependence between grayscale values, and the smoothness of texture.

[0104] As a preferred solution of this embodiment, step S04 is specifically as follows:

[0105] When sugarcane enters its elongation and maturity phase, its stems rapidly lengthen, and its leaves fully expand and shade each other, forming a dense canopy of green leaves. The overall biomass tends to a steady state. This stable, closed canopy structure ensures that the scattering mechanism between radar waves and the canopy remains constant.

[0106] In this example, during the elongation and maturity stage of sugarcane, the canopy morphology is characterized by a closed canopy structure composed of green sugarcane leaves. During this stage, the sugarcane canopy structure and biomass distribution tend to be stable, and the radar backscatter coefficient exhibits significant time series stability. Based on this, the Green Canopy Steady-state Scattering Index (GCSSI) was designed. The specific expression is:

[0107]

[0108]

[0109] in, 、 Respectively represent the minimum and maximum values ​​of VH in the study area, is the result after VH normalization, n is the number of periods in the time series, and the GCSSI steady-state threshold is recommended to be 0.3-0.55.

[0110] As a preferred solution of this embodiment, step S05 integrates the multi-phenological period radar time series response trajectories to design the Radar Shape Fingerprint Index (RSFI) of the radar backscatter coefficient time series curve for the entire growth period. By using logical operations, the heterogeneity of microwave scattering mechanisms in different growth stages is quantified and linked, and a unique identifier is assigned to the time series signal. Specifically,

[0111] Based on the optimal radar polarization mode screened out by the above data analysis and the radar characteristics of each phenological period, the radar shape fingerprint index of the polarization mode time sequence curve is designed. In the monitoring of complex systems with a single variable, the evolution of the shape of the time sequence signal implies the physical mechanism of the system state. Although traditional methods such as spectral analysis and trend decomposition can extract overall periodicity and long-term trend, they lack detailed mathematical description of non-stationary local jumps, segmented oscillations, and heteroscedastic fluctuations. The present application proposes a radar shape fingerprint index (RSFI) that fully utilizes the logical operation relationship, defines the shape category through the geometric characteristics of the time sequence curve, captures the phased shape changes of the time sequence curve over time, and gives the time sequence signal a unique identifier to help understand the internal structure of the time sequence. Sugarcane is an annual crop, and its life cycle can be divided into spring planting, germination, seedling, tillering, elongation and maturity. Its growth time lasts from the beginning of the year to the middle of February of the next year. According to the growth characteristics and time sequence signal change characteristics of different stages, the radar shape fingerprint index is constructed.

[0112]

[0113] wherein, represents the first-order difference of the corresponding time sequence signal at the moment, respectively represent different moments of time sequence data, and satisfy The radar shape fingerprint index designed according to the time sequence signal converts complex time sequence curves into quantifiable, interpretable and traceable shape fingerprints in two stages of shape coding and fingerprint generation, and has the characteristics of light weight and interpretability.

[0114] As a preferred scheme of the present embodiment, in step S06, the soil water loss, nutrient consumption and microbial activity of sugarcane in the spring planting period are decreased, a soil deficit event detection module is constructed, and the soil stress effect of sugarcane as a biomass intensive crop is quantified. The specific design process includes:

[0115] Considering that the above steps mainly rely on radar remote sensing data, the separability of sugarcane and other crops is measured by interpreting the characteristics of the backscattering signal of the target point, and the classification performance is strictly limited by the sample balance assumption. In the key phenological period of sugarcane elongation, time sequence analysis is easily disturbed by the intra-class variation caused by the change of phenological period, and the phenomenon of same object and different spectrum to some extent weakens the classification robustness. Therefore, the crop biophysical characteristics are fully utilized to construct a multi-dimensional identification system, and the classification is driven by the intrinsic properties of the crop, effectively breaking through the interpretation limit of a single data source.

[0116] Sugarcane is the highest biomass crop cultivated by human beings so far due to its efficient C4 photosynthetic pathway and strong carbon assimilation capacity. This high biomass output characteristic makes sugarcane a typical high nutrient demand crop, which has a predatory effect on soil nutrients. Specifically, it consumes 0.8-1.2 kg of nitrogen, 0.15-0.2 kg of phosphorus, and 1.5-2.0 kg of potassium per ton of production, which is 3-5 times the amount of rice. The high nutrient demand of sugarcane during the growth period makes the soil stress effect reach a peak during the spring cultivation period in the rotation cycle. At this time, the ground cover changes from crops to bare soil, and the lack of crop cover exacerbates water loss and reduces microbial activity, ultimately leading to a bare and degraded soil system. To accurately simulate this process, the present embodiment uses the dry bare soil index, a commonly used indicator that reflects the drought and barren conditions of the soil. Based on remote sensing data calculations, DBSI can quantify the soil stress effect caused by sugarcane as an intensive biomass crop and further reflect the dynamic characteristics of soil quality. , defined as the time point corresponding to the maximum value of the entire DBSI time series signal. The adaptive acquisition of the sugarcane spring cultivation period time window is determined by the following rule:

[0117]

[0118] wherein, is a set threshold value, recommended to be 1.95-2.7. According to the set threshold value , the change in soil moisture is analyzed to determine whether the soil has entered a state of drought or barrenness, achieving automatic locking of the sugarcane spring cultivation period window. The DBSI time series signal is shown in Figure 3 .

[0119] As a preferred scheme of the present embodiment, step S07 is the inversion of sugarcane elongation maturity period canopy chlorophyll content and photosynthetic capacity. During the elongation period of sugarcane, the sugarcane stem elongates rapidly, and the sugarcane leaf canopy gradually reaches a closed state, resulting in a sharp decrease in the interception of photosynthetically active radiation. This high-density canopy causes competition for light resources among plants. To quantify the canopy chlorophyll and light competition effect at this stage, a group light competition event detection module is designed to describe the intensity of photosynthesis and the absorption of chlorophyll, specifically:

[0120] A group chlorophyll light competition dynamic feedback model is established to quantify the intensity of photosynthesis and the absorption rate of chlorophyll. The photosynthetic rate of leaves during the elongation period is the highest among all growth periods of sugarcane, with a photosynthetic intensity of 10.9-17.0 mg CO2 / (h·100 cm 2 ) during the initial and peak elongation periods, and only 7.8 mg CO2 / (h·100 cm 2 ) and 9.2 mg CO2 / (h·100 cm 2). During the elongation period, sugarcane enters the rapid growth stage, which is characterized by the peak of sugarcane stem growth rate and the continuous enhancement of leaf area. In terms of optical index, the Enhanced Vegetation Index (EVI2) reaches the highest level during the whole growth period. After the peak of the elongation period, the larger the leaf area per plant, the more serious the mutual shading among individuals in the group, and the sugarcane canopy gradually reaches a closed state. At this time, the group structure presents the typical characteristics of canopy closure. In terms of light energy utilization, the mutual shading of leaves leads to a decrease in canopy transmittance and a decrease in the interception of photosynthetically active radiation per plant, triggering the competition effect of light resources among groups. In terms of material distribution, the base leaves start the aging program due to insufficient light, which is characterized by an accelerated disintegration rate of chloroplasts, a reduced leaf opening angle, and a rapid decline in chlorophyll content in a short period. In terms of group microenvironment, the air flow rate of the canopy decreases, leading to a continuous rise in relative humidity, forming a high-temperature and high-humidity environment. To accurately quantify the functional state of chloroplasts during this period, a dynamic feedback model of chlorophyll light competition is established, and the core of the model is the design of the Temporal Light Competition Index (TLCI), which can effectively describe the intensity of group photosynthesis and the absorption status of chlorophyll.

[0121] The specific discrimination rule is:

[0122]

[0123] Where start, heading, and end represent the initial, peak, and final stages of sugarcane growth, corresponding to 60, 210, and 360 days of accumulated days, respectively. Two periods are found to have local maximum values, and further judgment is made on the main period of sugarcane light competition.

[0124] As a preferred scheme of the present embodiment, step S08 links the radar steady-state characteristics of the elongation and maturation period with the optical light competition index to design the Canopy Steady-state Light Competition Index (CSLCI), which analyzes the specific morphological-metabolic activity state of the elongation and maturation period of sugarcane. Specifically:

[0125] During the elongation and maturation period of sugarcane, the canopy state is stable, and the radar data scattering remains stable. However, the optical index is very sensitive to the competition for light resources in the group during the elongation and maturation period. By synergizing multi-source remote sensing data, the radar steady-state characteristics of the elongation and maturation period are coupled with the optical light competition index to design the Canopy Steady-state Light Competition Index (CSLCI), which has both the ability to capture the geometric stability of radar and the sensitivity to physiological dynamics of optics. The specific index formula is:

[0126]

[0127] wherein start represents 30 days before the maximum value of VH in the whole time series signal, end represents 30 days after the maximum value of VH in the whole time series signal, and the VH time series signal graph is as shown in Figure 2 .

[0128] As a preferred scheme of the present embodiment, step S09 designs a sugarcane mapping scheme based on a plowing disturbance-green leaf canopy steady state-population light competition multi-dimensional attribute framework:

[0129] First, the Jeffries-Matusita distance is used to select a radar polarization feature combination, second, the phase-specific scattering mechanism is analyzed in stages, the ridging plowing event in the spring plowing period corresponds to high-resolution texture features, and the canopy evergreen event in the elongation mature period is manifested as the stability of the time series sequence of the radar polarization scattering coefficient; on this basis, the radar shape fingerprint index representing the morphological dynamics of sugarcane in the whole growth period is constructed. In the optical data dimension, a soil deficit detection model is established for the plowing disturbance event in the spring plowing period, and a light competition intensity quantification algorithm is developed for the light resource competition event caused by the closed canopy in the mature period. Further, by coupling the radar steady-state feature in the elongation mature period with the optical light competition index, a canopy steady state-light competition linkage discrimination index is designed, and finally a multi-level event linkage recognition model for the whole growth period of sugarcane is constructed, which integrates the radar morphological fingerprint, the physiological response of light, and the multi-source feature cooperation. Specifically:

[0130] Based on the radar shape function-crop physiological characteristic multi-dimensional attribute framework, the sugarcane remote sensing mapping method is designed by combining crop management measures, and the hierarchical structure of the method and the classification process are described as follows:

[0131] First, the separation efficiency of different radar polarization modes in sugarcane plant recognition is systematically evaluated. Based on the specific management measures and canopy scattering characteristics of sugarcane plants, a polarization feature optimization method based on statistical distance measurement is established: by calculating the JM distance values of different polarization features and polarization combination features between sugarcane and other ground object categories, the class separability is quantitatively characterized.

[0132] Second, the gray level co-occurrence matrix is introduced to extract the spatial texture features of the sugarcane canopy. According to the characteristics of the dynamic changes of soil structure after the ridging plowing in the spring plowing period of sugarcane, the second-order statistics are calculated using a multi-scale sliding window, including a texture feature set such as contrast, entropy, and correlation. The variance inflation factor is used to eliminate multicollinearity, the random forest algorithm is used to evaluate the feature importance, and the feature subset sensitive to the texture response of sugarcane is selected to construct a ridging plowing event detection module based on radar texture parameters.

[0133] Again, at the elongation and maturation stage of sugarcane, the canopy structure and biomass distribution of sugarcane tend to be stable, and the radar backscatter coefficient shows significant time series stability at this time. Based on this, the green leaf canopy steady-state mechanism of sugarcane at the elongation and maturation stage is constructed.

[0134] Then, the dynamic shape fingerprint index of the radar time series curve is designed by linking the radar time series response trajectory of multiple phenological stages. Based on the field investigation points, the VH time series signal diagram of sugarcane is obtained. According to the logical operation relationship, the morphological changes of the time series curve with time are captured, and the complex time series curve is converted into a quantifiable and interpretable morphological fingerprint index. This method can fully cope with the influence of spectral heterogeneity and environmental disturbance and form an adaptive adjustment mechanism.

[0135] Subsequently, from the perspective of crop physiology, the differences in growth habits between sugarcane and other crops are found out, and the performance on optical data is mapped. In terms of underlying surface, after the stalks of sugarcane are harvested, the vegetation coverage of the farmland changes from green leaf crops to bare soil, and enters the spring plowing period. During this period, the soil state changes, such as water loss, nutrient consumption, and microbial activity decline. The soil moisture changes rapidly, and its decay rate is significantly higher than that of the soil humidity after the harvest of conventional crops. This significant difference is caused by the thermal inertia and evaporation characteristics of the bare surface, and the dry bare soil index change characteristics provide a reliable spectral fingerprint for the identification of sugarcane ratoon field. When the numerical domain is greater than 0, it indicates that the surface soil enters a drought-poor state.

[0136] Then, the canopy dimension analysis is carried out. As an annual crop, sugarcane has a long green leaf coverage period, and the leaves are relatively wide. When the sugarcane field is covered with green leaves again as the crop grows, the plant enters the rapid growth period, and the growth rate of the plant height reaches the peak. At this time, the sugarcane field enters the ridge sealing state. At this stage, the sugarcane population structure presents typical canopy shading characteristics: the mutual shading of leaves leads to a decrease in canopy transmittance, and the interception of photosynthetically active radiation of individual plants decreases, triggering the light resource competition effect between plants. The base leaves start the aging program due to insufficient light, which leads to an accelerated rate of chloroplast disintegration, a reduced leaf opening angle, and a rapid decline in chlorophyll content in a short period. The air flow rate of the canopy decreases, leading to a continuous rise in relative humidity, and thus forming a high-temperature and high-humidity microenvironment. In addition, the fruits of sugarcane are located on the stalks below the leaves, so its high demand for light and flexible adjustment between leaves are not possessed by other crops. Based on this, a population light competition event detection method is designed.

[0137] Finally, the radar steady-state characteristics in the elongation and maturation period are linked to optical light competition indicators to construct a canopy steady-state-light competition discriminant index to analyze the specific morphological and metabolic activity state of sugarcane in the elongation and maturation period, highlight the physiological differences between sugarcane and other crops, and construct a multi-event time-series linkage detection method for sugarcane in the whole growth period from radar data to optical data, from soil moisture to canopy state.

[0138] As a preferred scheme of the embodiment, step S10 is specifically:

[0139] Taking A city and B city in a province as the research area, using the national standard administrative division vector map as the base map, and according to the above steps, the spatial distribution map of sugarcane in the research area is created. Through the multi-source information coupling strategy, the scattering mechanism characteristics and spectral biological markers are fused in multiple dimensions to construct a unique biophysical marker system for sugarcane, effectively interpret the phenotypic differences and habitat response heterogeneity between crops, and significantly improve the classification accuracy and spatial mapping accuracy of sugarcane in complex planting environments. The spatial distribution map of sugarcane in A city and B city of a province is obtained, as shown in Figure 5

[0140] The scheme provided in the above embodiment of the application system evaluates the separation efficiency of different radar polarization modes in sugarcane recognition. A polarization feature optimization method based on statistical distance measurement is established according to the specific management measures and canopy scattering characteristics of sugarcane plants. Secondly, the gray level co-occurrence matrix is introduced to extract the spatial texture features of the sugarcane canopy. A ridge tillage event detection module based on radar texture parameters is constructed according to the dynamic changes of soil structure after ridge tillage in the spring planting period of sugarcane. Then, in the elongation and maturation period of sugarcane, the canopy structure and biomass distribution of sugarcane tend to be stable, and the radar backscattering coefficient shows significant time series stability. Based on this, a green leaf canopy steady-state mechanism in the elongation and maturation period of sugarcane is constructed. Next, the dynamic shape fingerprint index of the radar time series curve in the whole growth period is designed by linking the multi-harvest period radar time series response trajectory, which converts the complex time series curve into a quantifiable, interpretable, and traceable morphological fingerprint. Then, from the perspective of crop physiology, the differences in growth habits between sugarcane and other crops are analyzed and mapped to optical data. In the spring planting period, soil water loss and nutrient consumption are constructed to build a soil deficit event detection module to quantify the soil stress effect caused by sugarcane as an intensive biomass crop. In the elongation and maturation period, the leaf canopy of sugarcane gradually reaches a closed state, and the interception of photosynthetically active radiation decreases sharply, triggering light resource competition between plants. A group light competition event detection module is designed to describe the intensity of group photosynthesis and the absorption status of chlorophyll. Finally, the radar steady-state characteristics in the elongation and maturation period are linked to optical light competition indicators to construct a canopy steady-state-light competition discriminant index to analyze the specific morphological and metabolic state of sugarcane in the elongation and maturation period, highlight the physiological differences between sugarcane and other crops, and provide support for accurate crop recognition.

[0141] ​The application breaks out the idea of evaluating the similarity of spectral index time series, comprehensively considers the site conditions and tillage measures (soil) of sugarcane planting areas, the growth characteristics (green) of the persistent green leaf canopy of stem crops, and the light competition (light) of densely planted crop groups, and proposes a method and system for automatically identifying sugarcane in the time series linkage of soil-green-light. From the aspect of sugarcane tillage measures (soil), the formation of ridge and furrow on the ground surface after ridge cultivation measures leads to a significant decrease in radar backscattering coefficient; at the same time, the bare soil index value is high due to the increase in reflectivity in the short-wave infrared light band. From the aspect of the growth characteristics (green) of the persistent green leaf canopy of sugarcane, the persistent green leaf canopy of sugarcane is covered by dense green leaves for a long time, resulting in a continuously high value of radar backscattering coefficient. From the aspect of light competition (light) of densely planted crop groups, sugarcane is a typical densely planted crop, and the dense sugarcane leaves lead to poor light conditions for the lower leaves of the group, and the chlorophyll absorption rate and photosynthetic rate of the group are reduced. The application focuses on the multi-dimensional properties of soil, green, and light of sugarcane, uses optical and radar time series data sets, and sequentially designs event detection methods for ridge cultivation, persistent green leaf canopy, and group light competition, and then constructs a method and system suitable for large-scale automatic identification of sugarcane according to the time series linkage characteristics of soil-green-light in the growth cycle of sugarcane.

[0142] Compared with the prior art, the characteristics and advantages of the embodiments of the application at least include:

[0143] (1) Based on the radar shape fingerprint index to eliminate time series shift, the sugarcane phenology self-adaptive monitoring is realized. The static threshold method of the traditional technology cannot adapt to the mutation of sugarcane phenology (the difference in growth period is 2-3 months), resulting in the breakage of time series characteristics. Based on the polarization scattering time series function, the radar shape fingerprint index is designed to analyze the scattering law in the whole growth period, adaptively match the phenology phase, eliminate the time shift interference of cross-year planting, and the phenology difference caused by regional climate difference.

[0144] (2) Through soil-green-light linkage diagnosis, the high biomass advantage of sugarcane is quantified. Most of the current research methods use single spectral features, resulting in serious mixed division of sugarcane and crops such as corn and sorghum. The application sequentially designs the event detection methods for sugarcane canopy steady-state scattering index, time series radar shape fingerprint index, green leaf canopy steady-state index, soil deficit event, group light competition event, and canopy steady-state-light competition linkage discriminant index, to realize the inversion of the physiological specificity of sugarcane.

[0145] (3) Give play to the agronomic mechanism, and construct a mechanism universal remote sensing model of sugarcane. The black box model lacks agronomic explanation, and has high regional dependence. The application starts from the agronomic mechanisms of ridge tillage, soil, and canopy, reveals the causes of spectral reflection of sugarcane, gives the model strong agronomic explanation, constructs a mechanism universal model, and gets rid of the limitation of regional sample parameters. The application supports seamless expansion from plot level (10 meters) to provincial level.

[0146] Based on the same inventive concept, the present application further provides a computer device, comprising: one or more processors, and a memory for storing one or more computer programs; the program comprises program instructions, and the processor is configured to execute the program instructions stored in the memory. The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, and are configured to implement one or more instructions, and are specifically configured to load and execute one or more instructions in the computer storage medium to implement the above method.

[0147] It needs to be further explained that, based on the same inventive concept, the present application further provides a computer storage medium, which stores a computer program, and the computer program is executed by the processor to perform the above method. The storage medium can adopt any combination of one or more computer readable media. The computer readable medium can be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium may, for example, be but is not limited to an electrical, magnetic, optical, infrared, or semiconductor system, device or apparatus, or any combination of the above. More specific examples (non-exhaustive list) of the computer readable storage medium include: an electrical connection with one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or apparatus.

[0148] It should be noted that, unless otherwise defined, technical terms or scientific terms used in the present application shall have the common meaning understood by one of ordinary skill in the art to which the present application pertains. The terms "first", "second", and similar terms used in the present application do not denote any order, quantity, or importance, but are used to distinguish different components. The terms "comprise", "comprising", and similar terms mean that the elements or objects before the term encompass the elements or objects listed after the term and their equivalents, and do not exclude other elements or objects. The terms "connected" or "connected" and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The terms "upper", "lower", "left", "right", and the like are used only to represent relative positional relationships, and when the absolute positions of the described objects are changed, the relative positional relationships can also be changed accordingly.

[0149] The above description is only the preferred embodiments of the present application, and is not intended to limit the present application in other forms. Any person skilled in the art can modify or change the above-mentioned disclosed technology into equivalent embodiments with equivalent changes. However, any simple modification, equivalent change, and modification of the above-mentioned embodiments made without departing from the technical solution of the present application, and in accordance with the technical essence of the present application, shall still fall within the protection scope of the present application.

[0150] The present application is not limited to the above-mentioned preferred embodiments, and anyone can derive other various forms of a synergistic soil-green-light timing linkage automatic sugarcane recognition method under the inspiration of the present application. Any equivalent change and modification made in accordance with the scope of the present application shall fall within the scope of the present application.

Claims

1. A sugarcane automatic identification method with coordinated soil-green-light time sequence linkage, characterized in that: include: Tillage disturbance event detection: quantifying dynamic changes in soil structure during spring plowing season based on radar texture heterogeneity induced by sugarcane ridge tillage; Canopy steady-state event detection: Based on the stability of the scattering mechanism in the closed canopy of sugarcane, the steady-state of biomass during the elongation and maturity period was quantified by normalized variance; Detection of group light competition events: Based on the competitive changes in chlorophyll absorption caused by dense sugarcane planting, through time series extreme value positioning and cumulative intensity analysis; Multi-source collaborative decision-making: coupling radar structural stability and optical physiological response characteristics, fusing event detection results through linkage discrimination index; Spatial distribution of exported sugarcane; Among them, the detection of tillage disturbance events, canopy steady-state events, and group light competition events is based on the linkage analysis of time series data throughout the growth period of sugarcane.

2. The sugarcane automatic identification method using soil-green-light time sequence linkage according to claim 1 is characterized by: During the tillage disturbance event detection process: The texture heterogeneity is quantified by the sudden change in radar echo intensity induced by tillage disturbance; The mutation intensity meets the disturbance discrimination threshold specific to sugarcane cultivation; The discrimination threshold is dynamically determined by the rate of change of texture feature entropy within the tillage period window.

3. The sugarcane automatic identification method using soil-green-light time sequence linkage according to claim 1 is characterized by: During the canopy steady-state event detection process: The normalized variance is achieved through steady-state analysis of sugarcane closed canopy scattering; A variance calculation model with demeaned variance was used; The steady-state judgment threshold is in the range of 0.3-0.55, which is related to the radar signal stability caused by the saturation of canopy biomass during the elongation and maturity period of sugarcane.

4. The sugarcane automatic identification method using soil-green-light time sequence linkage according to claim 1 is characterized by: In the multi-source collaborative decision-making process: The linkage discrimination index is quantified by the antagonistic relationship between radar structural stability and optical physiological response; The antagonistic relationship adopts the ratio operation model of radar and optical indicators; The antagonistic relationship is manifested as a negative correlation between enhanced sugarcane canopy structural stability and weakened chlorophyll absorption competition.

5. The sugarcane automatic identification method using soil-green-light time sequence linkage according to claim 1 is characterized by: Before the tillage disturbance event detection, canopy steady-state event detection and group light competition event detection, a data preprocessing step is also included: Perform geocoding and terrain correction on radar time series data to ensure spatial alignment with optical data, and perform spatial mosaicking and speckle reduction processing; Perform cloud masking and smooth reconstruction on optical time series data.

6. The sugarcane automatic identification method using soil-green-light time sequence linkage according to claim 1 is characterized by: In the tillage disturbance event detection and canopy stability event detection: Based on the JM distance separability metric, the vertical-horizontal polarization ratio and the cross-polarization scattering entropy are selected to form the optimal polarization feature combination.

7. The method for automatic sugarcane identification using soil-green-light time sequence linkage according to claim 1 is characterized by: The tillage disturbance event detection is associated with the soil moisture deficiency event: When the bare soil index peak is greater than zero and the cumulative bare soil index exceeds the set threshold within the spring plowing period, the soil stress effect induced by sugarcane planting is determined.

8. The sugarcane automatic identification method using soil-green-light time sequence linkage according to claim 1 is characterized by: In the tillage disturbance event detection and canopy stability event detection: The radar shape fingerprint index is used to analyze the phenological response morphology of the entire growth period, including: the declining pattern of the tillage period signal to support the detection of tillage disturbance events; It is used to support the stable morphology of the elongation signal during the detection of canopy steady-state events; The radar shape fingerprint index is generated by a logical operation of the first-order difference of the time series signal.

9. The sugarcane automatic identification method with coordinated soil-green-light time sequence linkage according to claim 1 is characterized by: The time windows for detecting the tillage disturbance event and the group light competition event are dynamically determined by: Tillage disturbance window: ±50 days based on the bare soil index peak day; Light competition window: locks the interval of the double-peak extreme points of chlorophyll absorption rate, and partially overlaps with the canopy steady-state event detection window.

10. A system for implementing the method according to claims 1-9, characterized in that: include: Radar event perception module: used to perform tillage disturbance event detection and canopy stability event detection; Optical physiological inversion module: used to perform group light competition event detection; Multi-dimensional decision engine: used to execute multi-source collaborative decision-making; Spatial distribution generation module: used to output the spatial distribution map of sugarcane planting areas.

Images