Multi-mode rubber plantation remote sensing identification and forest age estimation method

Abstract

The invention discloses a multi-mode rubber plantation remote sensing identification and forest age estimation method, which utilizes multi-source remote sensing data and combines an advanced image processing and analysis method to accurately monitor and estimate the distribution and forest age of different modes of rubber plantations such as field and small farmers. Firstly, through data preparation and integration, unsupervised image segmentation and classification optimization of a Gaussian mixture model are combined, and a high-precision rubber plantation distribution map is generated. And secondly, phenological characteristics are extracted based on time sequence vegetation-moisture index (such as NDWI) data, the planting year of the rubber tree is identified, the forest age is estimated, a rubber forest age distribution diagram is finally generated, and scientific support is provided for plantation management and ecological assessment.

Description

Technical Field

[0001] The present invention belongs to the field of remote sensing monitoring and image processing, and particularly relates to a multi-mode remote sensing identification and forest age estimation method for rubber plantations. Background Art

[0002] The rubber tree (Hevea brasiliensis) originally originated in the Amazon region, but in recent years, the expansion of its plantations has mainly occurred in historically unsuitable regions such as Southeast Asia, South Asia, and China. The rapid growth of global rubber cultivation is closely related to the increasing demand for rubber products in the industry. Now, rubber plantations have expanded to 25 degrees north latitude. Southeast Asia is the core of the global rubber industry, and Thailand, Indonesia, and Malaysia together produce more than 90% of the natural rubber. The sustainable development of the rubber industry has an important impact on the global economy and farmers' livelihoods. Therefore, it is crucial to systematically identify rubber plantations and their expansion patterns.

[0003] The modes of rubber plantations can be divided into smallholder plantations and industrialized plantations. Industrialized plantations are usually large in scale and have a standardized planting structure, which is conducive to the identification and analysis of remote sensing technology. Smallholder plantations, on the other hand, pose challenges to remote sensing monitoring due to their small scale and dispersion. Remote sensing technology has become an important tool for evaluating the distribution of rubber plantations and their ecological impacts. By integrating multi-source data (such as optical remote sensing, synthetic aperture radar, and lidar), the distribution, spatio-temporal changes, and forest age of rubber plantations can be efficiently monitored.

[0004] Remote sensing technology is widely used in the monitoring of rubber plantations and can extract key ecological indicators and analyze their growth conditions, especially in the growth and shedding cycles of leaves. With the development of cloud computing technology, the accuracy of remote sensing data processing and phenological analysis has been gradually improved. However, due to the heterogeneity of data sources and differences in processing methods, cross-regional comparison and long-term dynamic monitoring still face challenges, especially in smallholder plantations. Therefore, it is of great significance to deepen the application of remote sensing technology in multi-source data integration and dynamic monitoring.

[0005] Therefore, the present invention solves the limitations of the prior art in large-scale rubber plantation monitoring and forest age estimation, especially the problem that the identification ability of the existing methods is limited at a large scale under the influence of regional phenological differences and complex planting terrains. For example, the phenological changes in rubber plantations in northern Laos are significant, while the climate in the south is stable and the phenological changes are not obvious. It is difficult for the existing technology to eliminate phenological interference, resulting in poor classification consistency and insufficient accuracy. At the same time, under complex terrain conditions (such as mountains and hills), optical images are significantly affected by terrain shadows and vegetation canopy occlusion, and the existing methods have not fully integrated the advantages of SAR and optical data, limiting the monitoring ability. The present invention improves the classification accuracy and adaptability by integrating the advantages of SAR and optical images, eliminating phenological interference, and combining the Gaussian mixture model with time series analysis technology. Summary of the invention

[0006] The present invention is to solve the technical problems existing in the background technology, and the object is to provide a kind of multi-mode rubber plantation remote sensing identification and forest age estimation method, using multi-source remote sensing data, combined with advanced image processing and analysis methods, the distribution and forest age of different modes of rubber plantations such as field and small farmers are accurately monitored and estimated. First, by data preparation and integration, combined with the unsupervised image segmentation and classification optimization of Gaussian mixture model, a high-precision rubber plantation distribution map is generated. Secondly, based on time series vegetation-water index (such as NDWI) data extraction phenological characteristics, the planting year of rubber tree is identified and the forest age is estimated, and finally a rubber forest age distribution map is generated, which provides scientific support for plantation management and ecological assessment.

[0007] In order to solve the technical problem, the technical solution of the present invention is:

[0008] A multi-mode rubber plantation remote sensing identification and forest age estimation method, the method comprising:

[0009] S1: Using multi-source remote sensing data and ground sample point data, a remote sensing feature dataset of rubber plantations was constructed. Radiation correction, atmospheric correction and geometric correction were performed on the remote sensing feature dataset. The spectral characteristics and phenological changes of rubber plantations were extracted using the normalized vegetation index and normalized water index indicators to obtain pre-processed remote sensing image data.

[0010] S2: Based on the pre-processed remote sensing image data, a multi-dimensional remote sensing feature space is constructed. Through the synthesis of images throughout the year, the adaptability of the data to regional characteristics is enhanced, and multi-source remote sensing feature data that adapt to the characteristics of different regions is obtained;

[0011] S3: Use Gaussian mixture model to perform unsupervised image segmentation on multi-source remote sensing feature data, preliminarily identify the spatial distribution of rubber plantations, introduce expectation maximization algorithm to optimize the classification results, and obtain a preliminary distribution map of rubber plantations;

[0012] S4: Based on the obtained preliminary distribution map of rubber plantations, ground sample data was used for accuracy verification, the overall accuracy and Kappa coefficient were calculated, the classification errors were corrected according to the verification results, and re-verification was performed to obtain a high-precision distribution map of rubber plantations;

[0013] S5: Based on the time series NDWI data of Landsat images, the changes in the phenological characteristics of rubber tree planting were analyzed, the key turning points of the planting year were identified, and the planting year of rubber trees was estimated through the remote sensing vegetation characteristic model based on the growth law of rubber trees, and the forest age estimation results of rubber plantations were obtained;

[0014] S6: Combine the forest age estimation results with the spatial distribution data of rubber plantations to generate a forest age distribution map of rubber plantations.

[0015] Furthermore, in the step S1, the multi-source remote sensing data includes: Sentinel-1 SAR images, Sentinel-2 and Landsat optical images.

[0016] Furthermore, the step S2 includes:

[0017] S201: Sentinel-1 and Sentinel-2 data fusion, combine the VV and VH polarization SAR data of Sentinel-1 with the optical image data of Sentinel-2, and adopt the radar data fusion of ascending orbit ASC and descending orbit DSC; the fusion formula is as follows:

[0018]

[0019] Where: P ASC is the pixel value of the ascending orbit image; P DSC is the pixel value of the descending orbit image; P fused is the pixel value after fusion;

[0020] S202: Extraction of the maximum NDVI value, eliminate the interference of phenological changes through the maximum NDVI value throughout the year, extraction of the short-wave infrared band B11, and use the B11 band data of Sentinel-2 to capture the water change information of rubber plantations; eliminate the interference of phenological changes through the extraction of the maximum NDVI value throughout the year, and the NDVI calculation formula is as follows:

[0021]

[0022] To exclude the influence of seasonal changes, adopt the maximum NDVI value synthesis method throughout the year:

[0023] NDVI max = max(NDVI t )

[0024] Where: NDVI t is the NDVI value at each moment in the time series; NDVI max is the maximum NDVI value throughout the year; to capture the sensitivity of rubber plantations to water changes, extract the short-wave infrared band B11 data;

[0025] S203: Feature combination to construct a multi-dimensional feature space, splice the VV and VH polarization features of Sentinel-1 with the B11 band and the maximum NDVI value feature of Sentinel-2 to construct a multi-dimensional feature space.

[0026] Furthermore, step S3 includes: assuming that the data consists of different Gaussian distributions using the Gaussian Mixture Model (GMM), and estimating the parameters using the EM algorithm;

[0027] The mathematical expression of the Gaussian Mixture Model (GMM) is:

[0028]

[0029] where p(x) is the probability density of the sample point x, π k is the weight of the k-th Gaussian distribution, μ k and ∑ k are the mean and covariance of the k-th Gaussian distribution respectively, and K is the number of Gaussian distributions; select an appropriate mode to distinguish rubber plantations from other land cover types, and calculate the estimated values of the initial mean, covariance, and weight;

[0030] For each pixel point in the remote sensing image, calculate the probability that it belongs to each category according to the GMM model. For the case simplified to two Gaussian distributions, the GMM formula is simplified to:

[0031] p(x) = π 1 N(x|μ 1 , ∑ 1 ) + π 2 N(x|μ 2 , ∑ 2 )

[0032] where μ 1 , ∑ 1 , π 1 represent the mean, covariance, and weight of the rubber plantation category respectively, and μ 2 , ∑ 2 , π 2 represent the corresponding parameters of the farmland or oil palm plantation;

[0033] Use the Expectation-Maximization (EM) algorithm to calculate the probability that each sample point belongs to different Gaussian distributions, that is, calculate the posterior probability according to the current model parameters. Update the mean, covariance, and weight of the model according to the calculated probability. During this period, the model parameters will be gradually adjusted to maximize the likelihood of the data; repeat the steps until the parameters converge, that is, the change is very small or reaches the preset number of iterations; classify each pixel point into the category with the highest probability according to the calculated probability that it belongs to different categories.

[0034] Compared with the prior art, the advantages of the present invention are:

[0035] High-precision identification: By integrating multi-source remote sensing data and advanced image processing techniques (such as Gaussian mixture models), high-precision identification of rubber plantations can be achieved, especially in complex terrains and diverse planting patterns.

[0036] Automation and efficiency: By using unsupervised learning methods (GMM) and the expectation maximization algorithm, manual intervention is reduced, the degree of automation in classification and forest age estimation is improved, and the analysis efficiency is significantly enhanced.

[0037] Strong adaptability: The method can adapt to the characteristics of different regions, effectively handle the phenological changes between regions, and make the monitoring results equally reliable under different geographical and climatic conditions.

[0038] Construction of multi-dimensional feature space: By synthesizing annual images and fusing features, a multi-dimensional remote sensing feature space is constructed, providing rich information for subsequent analysis, and thus accurately capturing the growth dynamics of rubber plantations.

[0039] Accurate forest age estimation: By using NDWI to analyze the growth process of rubber trees, the planting year and growth stage can be accurately identified, providing a scientific basis for the calculation of forest age.

[0040] Support for sustainable agricultural management: This method provides important support for the sustainable management of rubber plantations, including farmland planning, management decision-making, and ecological environment assessment, promoting the coordinated development of agriculture and the ecosystem.

[0041] Data supports decision-making: The generated high-precision rubber plantation distribution map and forest age distribution map provide reliable information for decision-makers and farmers, supporting scientific agricultural management and policy formulation.

[0042] In summary, by integrating modern remote sensing technology and machine learning methods, the present invention not only improves the monitoring accuracy and efficiency of rubber plantations, but also provides effective data support and scientific basis for promoting sustainable agriculture and ecological management. Brief description of the drawings

[0043] Figure 1 The main flow chart of a multi-mode remote sensing identification and forest age estimation method for rubber plantations of the present invention;

[0044] Figure 2 The identification result map of a multi-mode remote sensing identification and forest age estimation method for rubber plantations of the present invention. Detailed implementation manners

[0045] The following describes the specific implementation manners of the present invention in conjunction with embodiments:

[0046] It should be noted that the structures, proportions, sizes, etc. shown in this specification are only used to cooperate with the content disclosed in the specification for those familiar with this technology to understand and read, and are not used to limit the implementation conditions of the present invention. Any modification of the structure, change of the proportional relationship, or adjustment of the size, without affecting the effects that the present invention can produce and the purposes that can be achieved, should still fall within the scope covered by the technical content disclosed in the present invention.

[0047] Meanwhile, the terms such as "upper", "lower", "left", "right", "middle", and "one" cited in this specification are only for the convenience of clear narration and are not used to limit the scope of implementation of the present invention. The change or adjustment of their relative relationship, without substantial change in the technical content, should also be regarded as the scope of implementation of the present invention.

[0048] Example 1:

[0049] The present invention aims to accurately monitor and estimate the distribution and forest age of rubber plantations in different modes such as large fields and small farms by using multi-source remote sensing data and combining advanced image processing and analysis methods. First, through data preparation and integration, combined with unsupervised image segmentation and classification optimization of the Gaussian mixture model, a high-precision rubber plantation distribution map is generated. Secondly, based on time-series vegetation-water index (such as NDWI) data, phenological characteristics are extracted, the planting year of rubber trees is identified and the forest age is estimated, and finally a rubber forest age distribution map is generated to provide scientific support for plantation management and ecological assessment. As Figure 1 shown, the specific steps are as follows:

[0050] 1. Theoretical basis and data preparation

[0051] Using multi-source remote sensing data such as Sentinel and Landsat, combined with ground sample point data, a remote sensing feature dataset of rubber plantations is constructed. Through preprocessing such as radiometric correction, atmospheric correction, and geometric correction of the remote sensing data, the quality and consistency of the data are ensured. Based on the time-series images, the normalized difference vegetation index (NDVI) and the normalized difference water index (NDWI) are extracted to dynamically analyze the spectral characteristics and phenological changes of rubber plantations, providing solid data support for subsequent classification and forest age estimation.

[0052] 2. Integration of multi-source remote sensing data

[0053] In view of the significant differences in phenological changes between the north and south regions (the phenological characteristics in the north are obvious, while those in the south are relatively gentle), through the synthesis of annual images, phenological interferences are effectively eliminated, and the adaptability of the technology to different regions is enhanced. The SAR images (VV and VH polarization characteristics) of Sentinel-1 are integrated with the optical images of Sentinel-2 and Landsat (red light band B4 and short-wave infrared band B11) data to construct a multi-dimensional remote sensing feature space. SAR images are applicable to complex terrains due to their good penetration ability, while optical images provide vegetation characteristics with high spectral resolution, enabling the precise monitoring of the spatial distribution of plantations on flat and mountainous lands simultaneously.

[0054] 3. Unsupervised Image Segmentation of Gaussian Mixture Model

[0055] The Gaussian Mixture Model (GMM) is used for unsupervised image segmentation of the feature space constructed from multi-source remote sensing data. Through statistical modeling methods, the rubber plantations in a single year are initially classified to identify their potential spatial distribution patterns. To further improve the classification accuracy, the Expectation-Maximization (EM) algorithm is introduced to iteratively optimize the initial classification results, thereby obtaining more accurate classification probability results. Finally, the distribution map of rubber plantations in a single year is output, providing a reference basis for subsequent forest age identification and accuracy verification.

[0056] 4. Accuracy Verification of Rubber Plantations

[0057] To ensure the reliability of the classification results of rubber plantations, ground sample data is used to evaluate the accuracy of the classification results, calculate the overall accuracy (OA) and Kappa coefficient, and the classification results are corrected and re-verified. The finally generated high-precision rubber plantation distribution map can provide accurate basic data support for subsequent research.

[0058] 5. Forest Age Identification and Estimation

[0059] Based on the time series NDWI data of Landsat images, by analyzing the inter-annual monthly phenological feature changes, the key inflection points in rubber tree planting are identified. Combining the planting rules of rubber trees and the remote sensing vegetation feature model, the planting year of rubber trees is deduced, and the forest age is estimated accordingly. This method makes full use of the mutations in rubber tree planting and can achieve high-precision forest age identification, providing key reference data for the production management and ecological assessment of rubber plantations.

[0060] 6. Distribution Map of Rubber Forest Age

[0061] By fusing the forest age identification results with the spatial distribution information of rubber plantations, the distribution map of rubber forest age is generated.

[0062] Example 2:

[0063] This embodiment is applied to the said Embodiment 1. Specifically, in the first step, multi-source data fusion is carried out to prepare the dataset; then the Gaussian mixture model is called to achieve unsupervised automatic classification of rubber plantations; based on the plantation map generated for a single year (such as 2024), Landsat data is used to identify the NDWI inflection point, and then its planting year is determined.

[0064] 1. Data Fusion and Feature Extraction (Based on the Feature Combination of Sentinel-1 and Sentinel-2)

[0065] Implementation Steps:

[0066] ① Sentinel-1 and Sentinel-2 Data Fusion: Combine the VV and VH polarized SAR data of Sentinel-1 with the optical image data of Sentinel-2, and adopt the radar data fusion of ascending orbit (ASC) and descending orbit (DSC) to effectively reduce the influence of mountain shadows and terrain complexity on the image..

[0067] ② Extraction of the Maximum NDVI Value: Eliminate the interference of phenological changes through the maximum NDVI value throughout the year, making the spatial distribution of rubber plantations more stable, effectively reducing phenological interference, and enhancing the representativeness and stability of vegetation recognition.

[0068] ③ Extraction of the Shortwave Infrared Band (B11): Use the B11 band data of Sentinel-2 to capture the moisture change information of rubber plantations.

[0069] The present invention combines Sentinel-1 synthetic aperture radar (SAR) data and Sentinel-2 optical image data, and uses the feature fusion method to improve the recognition accuracy of the spatial distribution of rubber plantations, especially more stable and applicable in complex terrain environments. The SAR data of Sentinel-1 includes VV and VH polarization features, which can provide surface structure and vegetation scattering information. The present invention adopts the radar data fusion of ascending orbit (ASC) and descending orbit (DSC) to effectively reduce the influence of mountain shadows and terrain complexity on the image. The fusion formula is as follows:

[0070]

[0071] Where: P ASC is the pixel value of the ascending orbit image; P DSC is the pixel value of the descending orbit image; P fusedis the fused pixel value. Through orbital fusion processing, the noise caused by terrain occlusion in single-orbital images is eliminated, and at the same time, the stability of rubber plantation recognition in complex terrain environments is improved. In the processing of Sentinel-2 optical images, the present invention eliminates the interference of phenological changes by extracting the maximum NDVI value throughout the year. The NDVI calculation formula is as follows:

[0072]

[0073] To exclude the influence of seasonal changes (such as greening and withering), the method of synthesizing the maximum NDVI value throughout the year is adopted:

[0074] NDVI max = max(NDVI t )

[0075] where: NDVI t is the NDVI value at each moment in the time series; NDVI max is the maximum NDVI value throughout the year. By extracting the maximum NDVI value throughout the year, the phenological interference can be effectively reduced, and the representativeness and stability of vegetation recognition can be enhanced. At the same time, in order to capture the sensitivity of rubber plantations to water changes, the present invention extracts the short-wave infrared band (B11) data. The B11 band can reflect the canopy water content of rubber trees and the vegetation growth status, providing important support for the accurate recognition of rubber plantations. In terms of feature combination, the present invention splices the VV and VH polarization features of Sentinel-1 with the B11 band and the maximum NDVI value feature of Sentinel-2 to construct a multi-dimensional feature space.

[0076] 2. Automatic Classification of Rubber Plantations Based on Gaussian Mixture Model (GMM)

[0077] Implementation steps:

[0078] ① GMM modeling: Use GMM to assume that the data consists of different Gaussian distributions (for example, the Gaussian distributions of rubber plantations and tropical rainforests), and use the EM algorithm to estimate the parameters.

[0079] ② Classification process: Explain how to calculate the probability of each pixel point belonging to a certain category according to GMM, so as to realize the automatic classification of rubber plantations in remote sensing images.

[0080] The Gaussian Mixture Model (GMM) is a probabilistic statistical model used to assume that a dataset is composed of multiple Gaussian distributions (normal distributions) with different parameters. Each Gaussian distribution represents a potential class or cluster in the data. By weighted combination of the data, GMM can describe the distribution of the entire dataset. In remote sensing image analysis, GMM is often used for automatic classification of land cover types, such as distinguishing rubber plantations from other vegetation or land cover types. The mathematical expression of GMM is as follows:

[0081]

[0082] where \(p(x)\) is the probability density of the sample point \(x\), \(\pi\) k is the weight of the \(k\)-th Gaussian distribution, \(\mu\) k and \(\sum\) k are the mean and covariance of the \(k\)-th Gaussian distribution respectively, and \(K\) is the number of Gaussian distributions. For complex land cover types, GMM can automatically separate different classes of land cover according to the spectral characteristics or other attributes of pixels. In practical applications, when distinguishing rubber plantations from types such as farmland or oil palm plantations, GMM can be simplified to two Gaussian distributions (i.e., \(K = 2\)). At this time, the formula of GMM is simplified as follows:

[0083] \(p(x)=\pi\) 1 N(x|\(\mu\) 1 ,\(\sum\) 1 )+\(\pi\) 2 N(x|\(\mu\) 2 ,\(\sum\) 2 )

[0084] where \(\mu\) 1 ,\(\sum\) 1 ,\(\pi\) 1 represent the mean, covariance and weight of the rubber plantation class respectively, and \(\mu\) 2 ,\(\sum\) 2 ,\(\pi\) 2 represent the corresponding parameters of farmland or oil palm plantations. Through prior knowledge (such as rubber plantations usually having different spectral reflection characteristics), rubber plantations can be assigned to clusters with different means, thus achieving effective classification. To estimate the parameters of these Gaussian distributions, this study uses the Expectation-Maximization (EM) algorithm. The EM algorithm consists of two iterative steps: the first step is to calculate the probability that each sample belongs to each Gaussian distribution (expectation step); the second step is to update the model parameters according to these probabilities (maximization step). Through multiple iterations, the EM algorithm can estimate the optimal model parameters, thus realizing the automatic classification of remote sensing images. After parameter estimation is completed, GMM can calculate the probability that each pixel belongs to a specific class, thus achieving efficient land cover classification.

[0085] 3. Determining the Planting Year Using Landsat Data

[0086] Implementation Steps:

[0087] ① Landsat Data Selection: Select Level-2 Collection-2 Tier-1 products of Landsat-5, Landsat-7, and Landsat-8 to ensure data quality and consistency.

[0088] ② NDWI Change Analysis: Use NDWI (Normalized Difference Water Index) to analyze the inflection points of land use change, and then infer the growth process and planting year of rubber plantations.

[0089] This study uses Landsat time series data (Landsat-5, Landsat-7, and Landsat-8) to estimate the planting year of rubber plantations. Since 1984, the Landsat satellite series has provided optical images with a 30-meter spatial resolution, and the revisit time of each satellite is 16 days. To ensure data quality, we used Level-2 Collection-2 Tier-1 products of Landsat-5, Landsat-7, and Landsat-8, which are atmospherically corrected and provide surface reflectance data suitable for long-term remote sensing analysis. Landsat data can accurately capture different growth stages of rubber plantations, from initial land preparation to the seedling stage after planting, and then to mature rubber forests in adulthood. For this reason, this study uses the Normalized Difference Water Index (NDWI) to monitor the water changes and growth process of rubber plantations. The calculation formula of NDWI is:

[0090]

[0091] where NIR represents the near-infrared band (B5 in Landsat-8), and SWIR represents the shortwave infrared band (B7 in Landsat-8). NDWI is sensitive to water changes and can reflect the growth process of rubber plantations. Compared with vegetation indices that rely on visible spectral bands, NDWI has less noise and lower sensitivity to atmospheric conditions (such as clouds and water vapor), and is particularly suitable for tropical regions. Taking rubber plantations as an example, during the land preparation stage, the NDWI value is low, indicating that the soil is exposed. When the rubber seedlings gradually grow after planting, the NDWI value rises steadily, reflecting an increase in vegetation water absorption. As the rubber trees enter the mature stage and form a complete canopy, the NDWI value reaches a plateau. By analyzing the NDWI changes in the Landsat time series, the planting year of rubber plantations can be clearly inferred, as Figure 2 shown.

[0092] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0093] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in one Figure 1 process or multiple processes and / or blocks Figure 1 block or multiple blocks.

[0094] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that realizes the functions specified in one Figure 1 process or multiple processes and / or blocks Figure 1 block or multiple blocks.

[0095] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in one Figure 1 process or multiple processes and / or blocks Figure 1 block or multiple blocks.

[0096] The preferred embodiments of the present invention are described in detail above, but the present invention is not limited to the above embodiments. Within the knowledge scope of those of ordinary skill in the art, various changes can be made without departing from the purpose of the present invention.

[0097] Many other changes and modifications can be made without departing from the concept and scope of the present invention. It should be understood that the present invention is not limited to specific embodiments, and the scope of the present invention is defined by the appended claims.

Claims

1. A multi-mode rubber plantation remote sensing identification and stand age estimation method, characterized in that: The method comprises: S1: Using multi-source remote sensing data and ground sample point data, a remote sensing feature dataset of rubber plantations was constructed. Radiation correction, atmospheric correction and geometric correction were performed on the remote sensing feature dataset. The spectral characteristics and phenological changes of rubber plantations were extracted using the normalized vegetation index and normalized water index indicators to obtain pre-processed remote sensing image data. S2: Based on the pre-processed remote sensing image data, a multi-dimensional remote sensing feature space is constructed. Through the synthesis of images throughout the year, the adaptability of the data to regional characteristics is enhanced, and multi-source remote sensing feature data that adapt to the characteristics of different regions is obtained; S3: Use Gaussian mixture model to perform unsupervised image segmentation on multi-source remote sensing feature data, preliminarily identify the spatial distribution of rubber plantations, introduce expectation maximization algorithm to optimize the classification results, and obtain a preliminary distribution map of rubber plantations; S4: Based on the obtained preliminary distribution map of rubber plantations, ground sample data was used for accuracy verification, the overall accuracy and Kappa coefficient were calculated, the classification errors were corrected according to the verification results, and re-verification was performed to obtain a high-precision distribution map of rubber plantations; S5: Based on the time series NDWI data of Landsat images, the changes in the phenological characteristics of rubber tree planting were analyzed, the key turning points of the planting year were identified, and the planting year of rubber trees was estimated through the remote sensing vegetation characteristic model based on the growth law of rubber trees, and the forest age estimation results of rubber plantations were obtained; S6: Combine the forest age estimation results with the spatial distribution data of rubber plantations to generate a forest age distribution map of rubber plantations.

2. A multi-mode rubber plantation remote sensing identification and stand age estimation method according to claim 1, characterized in that: In step S1, the multi-source remote sensing data includes: Sentinel-1 SAR images, Sentinel-2 and Landsat optical images.

3. A multi-mode rubber plantation remote sensing identification and stand age estimation method according to claim 2, characterized in that: The step S2 comprises: S201: Sentinel-1 and Sentinel-2 data fusion, combining the VV and VH polarimetric SAR data of Sentinel-1 with the optical image data of Sentinel-2, using the radar data fusion of the ascending orbit ASC and the descending orbit DSC; the fusion formula is as follows: Where: P ASC is the pixel value of the ascending orbit image; P DSC is the pixel value of the descending orbit image; P fused is the pixel value after fusion; S202: Extraction of the maximum NDVI value, using the annual NDVI maximum value to eliminate the interference of phenological changes, extraction of short-wave infrared band B11, using Sentinel-2's B11 band data to capture moisture change information in rubber plantations; extracting the annual NDVI maximum value to eliminate the interference of phenological changes, the NDVI calculation formula is as follows: In order to eliminate the impact of seasonal changes, the annual NDVI maximum value synthesis method is used: NDVI max =max(NDVI t ) Among them: NDVI t is the NDVI value at each moment in the time series; NDVI max is the maximum value of NDVI throughout the year; in order to capture the sensitivity of rubber plantations to moisture changes, shortwave infrared band B11 data was extracted; S203: Feature combination constructs a multidimensional feature space. The VV and VH polarization features of Sentinel-1 are spliced ​​with the B11 band and NDVI maximum value features of Sentinel-2 to construct a multidimensional feature space.

4. A multi-mode rubber plantation remote sensing identification and stand age estimation method according to claim 1, characterized in that: The step S3 comprises: using a Gaussian mixture model GMM to assume that the data is composed of different Gaussian distributions, and using an EM algorithm to estimate parameters; The mathematical expression of Gaussian mixture model GMM is: Among them, p(x) is the probability density of sample point x, π k is the weight of the kth Gaussian distribution, μ k and∑ k are the mean and covariance of the kth Gaussian distribution, respectively, K is the number of Gaussian distributions; select an appropriate mode to distinguish rubber plantations from other landform types, and calculate the initial mean, covariance and weight estimates; For each pixel in the remote sensing image, the probability of belonging to each category is calculated according to the GMM model. For the case of simplifying to two Gaussian distributions, the GMM formula is simplified to: p(x)=π1N(x|μ1,∑1)+π2N(x|μ2,∑2) Among them, μ1, ∑1, and π1 represent the mean, covariance, and weight of the rubber plantation category, respectively, and μ2, ∑2, and π2 represent the corresponding parameters of farmland or oil palm plantation; Use the expectation-maximization EM algorithm to calculate the probability that each sample point belongs to a different Gaussian distribution, that is, calculate the posterior probability based on the current model parameters. According to the calculated probability, update the mean, covariance and weight of the model. During this period, the model parameters will be gradually adjusted to maximize the likelihood of the data; repeat the steps until the parameters converge, that is, the change is very small or the pre-set number of iterations is reached; according to the calculated probability that each pixel belongs to a different category, classify it into the category with the highest probability.

Images