Method and system for calculating vegetation cover index based on double dynamic cloud optimization mechanism

Abstract

The present application relates to the technical field of ecological remote sensing monitoring, in particular to a method and system for calculating vegetation coverage index based on a double dynamic cloud optimization mechanism, which comprises the following steps: obtaining an NDVI original time series dataset of a target region, dynamically analyzing the cloud coverage degree of different time series based on a cloud coverage dynamic analysis mechanism and the NDVI original time series dataset to obtain cloud coverage grading results of different time series; processing the NDVI original time series dataset through a filtering processing model to obtain filtered NDVI time series data of different time series; setting an NDVI time series data adaptive optimization mechanism based on cloud coverage grading, correcting the NDVI time series data of different time series based on the adaptive optimization mechanism and the cloud coverage grading results to obtain a target NDVI time series dataset; analyzing the vegetation coverage index of the target region according to the target NDVI time series dataset to obtain vegetation coverage index analysis results, thereby realizing real-time dynamic monitoring of the vegetation coverage of the target region.

Description

Technical Field

[0001] This application belongs to the field of ecological remote sensing monitoring technology, specifically relating to a method and system for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism. Background Technology

[0002] In ecological remote sensing monitoring, vegetation index is a key indicator for measuring vegetation coverage and assessing regional ecological quality. However, cloud cover can interfere with vegetation index analysis, affecting data monitoring and assessment based on this index.

[0003] Currently, the main methods for addressing cloud interference include single-phase maximum value synthesis, time-series filtering reconstruction, and static threshold correction. However, all of these methods have limitations to varying degrees and are difficult to guarantee the authenticity of data and the accurate representation of vegetation coverage under complex conditions such as cloud cover.

[0004] The single-phase maximum value synthesis method relies too heavily on the maximum value assumption. In areas with continuous cloud cover or cloud cover throughout the growing season, the maximum value can lead to distortion of vegetation signals. The temporal filtering reconstruction method can play a certain role in some cloud interference scenarios, but in cases of long-term or full-season cloud cover, the continuous cloud obstruction will result in insufficient effective input data, causing the analysis results to deviate from the true vegetation growth trend. The static threshold correction method suffers from threshold rigidity and cannot be dynamically adjusted according to phenological changes, interannual variations, and extreme weather events, which can easily lead to misjudgments. Moreover, it is limited by the spatiotemporal resolution of auxiliary data.

[0005] Therefore, it is necessary to develop a dual optimization method for NDVI that combines graded correction and multi-time dynamic data fusion to effectively reduce the interference of cloud cover on vegetation index analysis results and improve the accuracy and reliability of vegetation cover monitoring results in different regions. Summary of the Invention

[0006] To address the shortcomings of existing methods and the needs of practical applications, and to reduce the interference of cloud cover factors on vegetation index analysis results, thereby further improving the accuracy and reliability of vegetation cover monitoring results in different regions, this invention provides a method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism. The method includes: acquiring the original NDVI time-series dataset of the target region; constructing a dynamic cloud cover analysis mechanism; dynamically classifying the cloud cover degree at different time series based on the dynamic cloud cover analysis mechanism and the original NDVI time-series dataset to obtain cloud cover level classification results for different time series; setting a filtering processing model to process the original NDVI time-series dataset to obtain filtered NDVI time-series data for different time series; setting an adaptive optimization mechanism for NDVI time-series data based on cloud cover level; correcting the NDVI time-series data for different time series based on the adaptive optimization mechanism and the cloud cover level classification results to obtain a target NDVI time-series dataset; and analyzing the vegetation cover index of the target region based on the target NDVI time-series dataset to obtain vegetation cover index analysis results.

[0007] The dual dynamic cloud optimization mechanism of this invention includes a dynamic cloud coverage analysis mechanism and an NDVI adaptive optimization mechanism based on cloud coverage level. It can dynamically adjust and optimize according to the cloud coverage level at different times, better adapt to complex and ever-changing cloud coverage environments, and improve the applicability and robustness of this method in different scenarios.

[0008] Optionally, the step of setting a filtering processing model, combining the cloud coverage level classification results of different time series, and processing the original NDVI time series dataset through the filtering processing model to obtain filtered NDVI time series data of different time series includes: setting filtering window parameters, the filtering window parameters including window width, window half width and relative offset within the window; and determining the filtering coefficients through the least squares objective function.

[0009] A filtering model is established by combining the filtering coefficients, the window width, the window half-width, and the relative offset; the original NDVI time series dataset is processed using the filtering model to obtain filtered NDVI time series data.

[0010] This invention uses the least squares objective function to determine the filtering coefficients, which can quickly and accurately solve for the optimal filtering coefficients. Combined with the pre-set filtering window parameters, a filtering processing model is established, which makes the calculation process highly efficient and can process large-scale NDVI raw time series datasets in a short time.

[0011] Optionally, the step of acquiring the raw NDVI time-series dataset of the target area, constructing a dynamic cloud coverage analysis mechanism, and dynamically classifying the cloud coverage level at different time series based on the dynamic cloud coverage analysis mechanism and the raw NDVI time-series dataset to obtain cloud coverage level classification results at different time series includes: acquiring sky imaging data at different time series in the target area based on the raw NDVI time-series dataset; dynamically analyzing the sky imaging data through the dynamic cloud coverage analysis mechanism to obtain new feature matrices at different time series in the target area; combining the dynamic cloud coverage analysis mechanism and the new feature matrices to analyze the sky cloud cover analysis results at different time series in the target area; and the dynamic cloud coverage analysis mechanism dynamically classifying the cloud coverage level at different time series based on the sky cloud cover analysis results to obtain cloud coverage level classification results at different time series.

[0012] This invention dynamically analyzes sky imaging data to generate a new feature matrix, which can automatically adapt to differences in environmental data and ensure that cloud coverage levels can be classified in different regions and time periods.

[0013] Optionally, the step of dynamically analyzing the sky imaging data through the cloud coverage dynamic analysis mechanism to obtain new feature matrices of different time series in the target area includes: establishing a sky imaging feature extraction function in the cloud coverage dynamic analysis mechanism, extracting features from the sky imaging data according to the sky imaging feature extraction function to obtain sky imaging feature matrices of different time series; setting feature matrix segmentation conditions in the cloud coverage dynamic analysis mechanism, segmenting the sky imaging feature matrix based on the feature matrix segmentation conditions to obtain feature matrix segmentation results of different time series; setting a pixel determination function in the cloud coverage dynamic analysis mechanism, determining the feature matrix segmentation results through the pixel determination function to obtain determination results of feature matrices of different time series; and dynamically adjusting and optimizing the sky imaging feature matrix based on the determination results to obtain new feature matrices of different time series in the target area.

[0014] The dynamic analysis mechanism of this invention can perform dynamic analysis and feature matrix adjustment and optimization, which can reflect the changes in cloud coverage in a timely manner. This enables the new feature matrix to accurately describe the cloud coverage characteristics at different time series, providing strong support for real-time monitoring and early warning of cloud coverage-related events.

[0015] Optionally, the step of combining the dynamic cloud cover analysis mechanism and the new feature matrix to analyze the sky cloud cover analysis results of different time series in the target area includes: establishing a sky cloud cover prediction model in the dynamic cloud cover analysis mechanism; and analyzing the sky cloud cover analysis results of different time series in the target area based on the sky cloud cover prediction model and the new feature matrix. This invention combines a new feature matrix for different time series, enabling the sky cloud cover prediction model to comprehensively consider the changing patterns of cloud cover over time, thereby improving the predictive ability of sky cloud cover changes.

[0016] Optionally, the step of setting an adaptive optimization mechanism for NDVI time-series data based on cloud coverage level, and correcting NDVI time-series data of different time series based on the adaptive optimization mechanism and the cloud coverage level classification results to obtain the target NDVI time-series data set, includes: constructing an adaptive optimization mechanism for NDVI time-series data based on cloud coverage level classification results; and setting an adaptive optimization mechanism for NDVI time-series data of level I, level II, and level III pixels within the adaptive optimization mechanism. This invention's adaptive optimization mechanism corrects NDVI data of different cloud coverage levels, reducing noise and uncertainty in the data, making the input model data more stable and reliable, thereby enhancing the robustness of the method of this invention.

[0017] Optionally, the step of setting up a Level I pixel NDVI time-series data adaptive optimization mechanism, a Level II pixel NDVI time-series data adaptive optimization mechanism, and a Level III pixel NDVI time-series data adaptive optimization mechanism in the NDVI time-series data adaptive optimization mechanism includes: obtaining a complete detection result of the Level I pixel NDVI time-series data based on the Level I pixel NDVI time-series data adaptive optimization mechanism; obtaining first Level I pixel NDVI time-series data based on the complete detection result; smoothing the first Level I pixel NDVI time-series data using the Level I pixel NDVI time-series data adaptive optimization mechanism to obtain second Level I pixel NDVI time-series data; and performing mean analysis on the second Level I pixel NDVI time-series data according to the Level I pixel NDVI time-series data adaptive optimization mechanism to obtain the target NDVI time-series data of the Level I pixels.

[0018] This invention performs a series of steps, including complete detection, smoothing, and mean analysis, on the NDVI time-series data of Class I pixels, which helps to ensure the quality and accuracy of the NDVI time-series data of Class I pixels.

[0019] Optionally, the step of setting up a Level I pixel NDVI time-series data adaptive optimization mechanism, a Level II pixel NDVI time-series data adaptive optimization mechanism, and a Level III pixel NDVI time-series data adaptive optimization mechanism in the NDVI time-series data adaptive optimization mechanism includes: using the Level II pixel NDVI time-series data adaptive optimization mechanism to perform periodic analysis on the Level II pixel NDVI time-series data to obtain Level II pixel NDVI time-series data at different time points; and correcting and optimizing the Level II pixel NDVI time-series data at different time points based on the variation law of Level II pixel NDVI values ​​at different time points to obtain the target NDVI time-series data of the Level II pixels.

[0020] This invention effectively identifies outliers in the NDVI values ​​of Level II pixels through periodic analysis and correction and optimization of variation patterns, thereby ensuring the reliability of Level II pixel NDVI data.

[0021] Optionally, the step of setting up a Level I, Level II, and Level III NDVI time-series data adaptive optimization mechanism in the NDVI time-series data adaptive optimization mechanism includes: obtaining optical remote sensing data and microwave remote sensing data of the target area; the Level III NDVI time-series data adaptive optimization mechanism calibrating the optical and microwave remote sensing data to obtain calibrated optical and microwave remote sensing data; analyzing the relationship between optical and microwave remote sensing data by combining the Level III NDVI time-series data adaptive optimization mechanism, the calibrated optical and microwave remote sensing data; constructing a neural network-based data fusion model in the Level III NDVI time-series data adaptive optimization mechanism; and fusing the calibrated optical and microwave remote sensing data by combining the data fusion model and the relationship to obtain the target NDVI time-series data of the Level III pixels.

[0022] This invention uses a neural network data fusion model to fuse optical remote sensing data and microwave remote sensing data, which can fully explore the complementary information between the two types of data, so that the NDVI time series data of the Class III pixel target accurately reflects the real situation of vegetation.

[0023] Secondly, to efficiently execute the method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism provided by this invention, this invention also provides a system for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism. The system includes an input device, a processor, an output device, and a memory, wherein the input device, processor, output device, and memory are interconnected. The memory includes the method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism as described in the first aspect of this invention. The memory stores a computer program, which includes program instructions, and the processor is configured to call the program instructions. The system for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism provided by this invention has a compact structure, strong applicability, and greatly improves operating efficiency. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings.

[0025] Figure 1 This is a flowchart of the method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism according to the present invention.

[0026] Figure 2 This is a schematic diagram of the NDVI time-series data adaptive optimization mechanism of the present invention;

[0027] Figure 3 This is a schematic diagram of the processing flow of the adaptive optimization mechanism for Level I pixel NDVI time-series data of the present invention;

[0028] Figure 4 This is a schematic diagram of the processing flow of the adaptive optimization mechanism for Level II pixel NDVI time-series data of the present invention.

[0029] Figure 5 This is a schematic diagram of the processing flow of the adaptive optimization mechanism for Level III pixel NDVI time-series data of the present invention;

[0030] Figure 6 This is a system structure diagram of the present invention for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism. Detailed Implementation

[0031] Specific embodiments of the present invention will now be described in detail. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other instances, well-known circuits, software, or methods have not been specifically described to avoid obscuring the invention.

[0032] Throughout this specification, references to "an embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with that embodiment or example is included in at least one embodiment of the invention. Therefore, the phrases "in an embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily refer to the same embodiment or example. Furthermore, specific features, structures, or characteristics can be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. Moreover, those skilled in the art will understand that the illustrations provided herein are for illustrative purposes and are not necessarily drawn to scale.

[0033] Please see Figure 1 To reduce the interference of cloud cover on vegetation index analysis, a dynamic hierarchical correction strategy is implemented, which can provide more reliable vegetation change information and achieve real-time monitoring and effective analysis of vegetation dynamic change trends. This invention provides a method for calculating the vegetation cover index based on a dual dynamic cloud optimization mechanism. The method includes the following steps:

[0034] S1. Obtain the original NDVI time-series dataset of the target area, construct a dynamic cloud coverage analysis mechanism, and dynamically classify the cloud coverage level at different time series based on the dynamic cloud coverage analysis mechanism and the original NDVI time-series dataset to obtain the cloud coverage level classification results at different time series. The specific steps and implementation content are as follows:

[0035] Obtain the original NDVI time-series dataset for the target region.

[0036] When acquiring the raw NDVI time-series dataset for the target region, it is first necessary to clearly define the specific scope of the target region, including its geographic coordinates and boundary information, to ensure the accuracy and relevance of the data. Simultaneously, the time span of the study should be comprehensively considered, and an appropriate time period should be determined based on the research objectives to ensure effective analysis of vegetation evolution trends.

[0037] Make full use of multiple remote sensing data sources to acquire raw NDVI time-series datasets of the target area. During data acquisition, strictly adhere to relevant data acquisition specifications and procedures. For satellite remote sensing data, understand its data format, projection method, resolution, and other parameters to ensure data compatibility and usability. Simultaneously, conduct a preliminary inspection of the acquired raw data, and promptly supplement or correct any problematic data.

[0038] Establish a dynamic analysis mechanism for cloud coverage.

[0039] First, sky imaging data at different time series in the target area are obtained based on the above-mentioned original NDVI time series data;

[0040] Based on the original NDVI time-series data of different time periods in the target area, and taking into account factors such as the geographical features, climate conditions and vegetation distribution characteristics of the target area, sky imaging data of different time periods in the target area are selected to ensure that the acquired sky imaging data has high accuracy and reliability, and to provide a data foundation for subsequent NDVI time-series data analysis of the target area.

[0041] Then, the sky imaging data is dynamically analyzed through the above-mentioned cloud coverage dynamic analysis mechanism to obtain new feature matrices of different time series in the target area.

[0042] Step 1: A sky imaging feature extraction function was established in the dynamic analysis mechanism of cloud coverage. Based on the above sky imaging feature extraction function, features were extracted from the sky imaging data to obtain sky imaging feature matrices for different time series.

[0043] The above-mentioned sky imaging feature extraction function takes sky camera imaging data as input. The sky camera imaging data can come from satellite remote control data or other base station monitoring information. The algorithm logic analyzes and mines the data, that is, it performs feature extraction operations on the sky imaging data, and finally generates sky imaging feature matrices of different time series.

[0044] The sky imaging feature extraction function established by combining the activation function satisfies the following relationship:

[0045]

[0046] in, The feature matrix representing the image captured by the sky camera. This indicates the row number of the matrix within the interval. Let a and b represent the column numbers of the matrix within the interval, where the ranges of a and b both satisfy the following condition: , This represents an element in a matrix.

[0047] Each element in the matrix Both consist of two parts, namely include and , This represents the probability that the current pixel is part of the sky background, and its value typically ranges from [value range missing]. Between these values, the closer the value is to 1, the greater the probability that the pixel is part of the sky background. This represents the probability that the current pixel is a cloud, and its value range is also within... Between 1 and 1, the closer the value is to 1, the greater the probability that the pixel is a cloud.

[0048] The second step involves setting feature matrix segmentation conditions in the dynamic analysis mechanism of cloud coverage. Based on these conditions, the sky imaging feature matrix is ​​segmented to obtain feature matrix segmentation results for different time series.

[0049] Feature matrix segmentation conditions were set in the dynamic cloud coverage analysis mechanism. Based on these segmentation conditions, the generated sky imaging feature matrix was analyzed. Segmentation is performed to obtain feature matrix segmentation results at different time sequences. Specifically, threshold segmentation is applied to the feature matrix, with the feature matrix segmentation threshold set to [value missing]. Its value range is within In other embodiments, the values ​​can be flexibly adjusted according to the actual data monitoring situation to obtain the best segmentation effect.

[0050] The above feature matrix segmentation conditions satisfy the following relationship:

[0051]

[0052] in, The feature matrix representing the image captured by the sky camera. This indicates that you can choose according to the actual situation. or ,in This indicates the probability that the current pixel is a sky background. This indicates the probability that the current pixel is a cloud.

[0053] In one optional embodiment, select (That is, the probability of a pixel being a sky background), in the above formula for ,Will Greater than the segmentation threshold of Set as ;Will Less than the segmentation threshold of .

[0054] In one optional embodiment, select (That is, the probability that a pixel is a cloud), in the above formula for ,Will Greater than the segmentation threshold of Set as ;Will Less than the segmentation threshold of Set as .

[0055] Step 3: A pixel determination function is set in the dynamic analysis mechanism of cloud coverage. The pixel determination function is used to determine the segmentation results of the feature matrix and obtain the determination results of different time-series feature matrices.

[0056] By determining the segmentation results of the feature matrix based on the pixel determination function, determination results of different temporal feature matrices can be obtained.

[0057] A matrix needs to be introduced during the pixel determination process. ,Will and Multiplication to determine each To determine whether a pixel belongs to the cloud or the sky, if the result is 0, the pixel is determined to be a sky pixel; if the result is 1, the pixel is determined to be a cloud pixel.

[0058] The pixel determination function is as follows:

[0059]

[0060] in, This indicates the result of the decision function operation. This represents each element in the characteristic matrix. This indicates the introduced matrix.

[0061] because This includes and Based on the above content, it can be seen that This indicates the probability that the current pixel is a sky background. This indicates the probability that the current pixel is a cloud.

[0062] Furthermore, the pixel determination function described above can also satisfy the following relationship:

[0063]

[0064]

[0065] in Indicates the first Line 1 The probability that a column pixel is a sky background;

[0066] in Indicates the first Line 1 The probability that a column pixel is a cloud;

[0067] When the calculation result When the value is 0, the pixel is determined to be a sky pixel;

[0068] When the calculation result When the value is 1, the pixel is determined to be a cloud pixel.

[0069] After the pixel determination function determines the value, each element in the original feature matrix is ​​converted into a scalar value (0 or 1), where 0 represents sky pixels and 1 represents cloud pixels, thus obtaining a new matrix containing only 0 and 1. This matrix is ​​beneficial for the rapid analysis and calculation of cloud cover in subsequent operations.

[0070] Step 4: Based on the judgment results, dynamically adjust and optimize the sky imaging feature matrix to obtain new feature matrices for different time series in the target area.

[0071] By dynamically adjusting and optimizing the sky imaging feature matrix based on the above determination results, new feature matrices for different time series in the target region can be obtained. It satisfies the following relationship:

[0072]

[0073] in, This represents a new feature matrix representing different time series within the target region. This represents the result of the pixel determination function. The result includes 0 and 1, where 0 represents sky pixels and 1 represents cloud pixels.

[0074] New feature matrix The inner elements are converted into a vector that is either 0 or 1, where 1 represents a cloud pixel, indicating that the pixel corresponds to a cloud and the point is in a cloudy state, and 0 represents a sky pixel, indicating that the pixel corresponds to a cloudless sky and the point is in a cloudless state.

[0075] During processing, the sky image has been processed to a size of [size missing]. The image is of a specific size, and the cloud pixels are completely contained within the sky image. Simultaneously, the cloud feature matrix generated by the sky imaging feature extraction function is also of size [size missing]. These specifications ensure the consistency and accuracy of data processing.

[0076] Next, the cloud cover dynamic analysis mechanism and the new feature matrix were combined to analyze the cloud cover analysis results of different time series in the target area.

[0077] Within the framework of a dynamic analysis mechanism for cloud cover, a sky cloud cover prediction model was established. This model can accurately capture the dynamic changes in sky cloud cover, providing strong support for subsequent analysis.

[0078] To accurately calculate the percentage of cloud pixels in the all-sky camera image, i.e., to determine the proportion of cloud pixels within the inscribed circle of the sky imaging matrix, cloud pixels are represented as elements with a value of 1 in the feature matrix. First, all cloud pixels (i.e., elements with a value of 1) are summed. Then, this sum is divided by the number of pixels in the inscribed circle of the sky imaging matrix. This yields the cloud cover result in the all-sky camera image. The cloud cover result is then used... The specific calculation formula is as follows:

[0079]

[0080] in, This indicates the results of cloud cover analysis in sky imaging.

[0081] Molecular part This involves summing all cloud pixels (elements with a value of 1) in the feature matrix and then multiplying by the corresponding parameter 4; the denominator part... This represents the area of ​​the inscribed circle in the sky image.

[0082] Based on the sky cloud cover prediction model and the newly generated feature matrix, the sky cloud cover analysis results for the target area under different time series are analyzed and evaluated. To facilitate calculation, the number of pixels corresponding to the side length of the sky image is kept consistent in the row and column directions during approximation (in practical applications, this can be flexibly adjusted according to the actual number of rows and columns of the matrix). The values ​​calculated using the above formula are... The value, which is the final cloud cover result, ranges from [value missing]. Within the range. The closer the value is to 1, the higher the degree of cloud cover in the sky; conversely, the lower the value is, the lower the degree of cloud cover.

[0083] Finally, the dynamic cloud coverage analysis mechanism dynamically classifies the cloud coverage level at different time periods based on the sky cloud volume analysis results, so as to obtain the cloud coverage level classification results at different time periods.

[0084] Cloud cover calculation results Analysis can effectively reveal cloud coverage at different time points, based on the following content:

[0085] Clear and cloudless condition: when Approaching infinitely close to 0, specifically in When the value is in the range of , it indicates that the proportion of cloud pixels in the sky image is relatively small. This means that the cloud cover in the sky is relatively sparse at this time, and most areas present a clear and cloudless scene, allowing sunlight to shine down without obstruction, and the sky is clear and bright.

[0086] Medium cloud coverage: If It is in the middle value range, specifically in When the range is specified (this specific range is not fixed and can be flexibly adjusted according to actual needs and different application scenarios), it indicates that there are a certain number of clouds in the sky. At this time, the cloud coverage is at a medium level, the sky is not completely blue, and the clouds are distributed in a sparse or dense state, sometimes blocking part of the sunlight and bringing dappled light and shadow to the earth.

[0087] Severe cloud cover status: When Gradually approaching 1, that is, at When the range is in the range, it means that cloud pixels occupy the majority of the sky image, reflecting that the cloud coverage in the sky is quite serious, the sky is almost completely covered by clouds, and the thick cloud layer completely blocks the sky, making it difficult for sunlight to penetrate.

[0088] S2. Set up a filtering model to process the original NDVI time series dataset, obtaining filtered NDVI time series data of different time series. The specific setup steps and implementation details are as follows:

[0089] The relevant process for setting up the filtering model is as follows:

[0090] Step 1: Set the filter window parameters. In this embodiment, the filter window parameters mainly include the window width, the window half-width, and the relative offset within the window. During filtering, it is necessary to set the filter window parameters, which mainly include the window width, the window half-width, and the relative offset within the window. Let the filter window width be... ( (odd number); window half width Through formula The calculation shows that the data point within the window is relative to the current data point to be processed. offset used express, The range of values ​​is from arrive All integers within.

[0091] Step 2: In this embodiment, the filter coefficients are determined using a least squares objective function. The specific values ​​of the filter coefficients determined by the least squares objective function depend on the window size. The core objective of the least squares method is to minimize the sum of squared errors between the original data and the fitted polynomial within the local window, taking into account the size and order of the polynomial used. This allows the optimal filtering coefficients to be determined, enabling the filtered data to more accurately reflect the true vegetation change trend and effectively reduce interference from noise such as cloud cover.

[0092] The objective function of the least squares method satisfies the following relationship:

[0093]

[0094] in, Let the objective function of the least squares method be represented. Represents the objective function Regarding filter coefficients Find the minimum value. Indicates the index within the local window is The original data values, Indicates the half-width of the filter window. For the index of data points being processed, Indicates relative within the window The offset, Representation and position The relevant filter coefficients, Represents the local window relative to the current data point to be processed. There is an offset The original data values.

[0095] This objective function measures the sum of squared errors between the original data and the fitted polynomial within a local window. By adjusting... The value of makes this function reach its minimum value.

[0096] Represents the objective function Regarding filter coefficients Find the minimum value, that is, find a set of values. The value of makes To reach the minimum.

[0097] It is a summation symbol, indicating summation from... arrive All integers Perform summation. The above... It is the index of the data point within a local window; the window is set to the currently pending data point. Centered on, the window half-width is Therefore, the index range of data points within the window is from arrive When processing NDVI data, This represents the NDVI observation data value (raw data value) at a certain moment within the window.

[0098] Similarly, the summation symbol indicates summation from... arrive All integers Perform summation. It is the data point within the window relative to the current data point to be processed. The offset is used to determine the contribution of data at different positions within the window to the fitting.

[0099] With position The relevant filter coefficients are the unknown parameters that need to be solved for. The values ​​of the correlation coefficients determine the shape of the local polynomial fit, and the optimal shape is determined using the least squares method. The value allows the fitted polynomial to better approximate the original data.

[0100] Represents the local window relative to the current data point to be processed. There is an offset The original data values.

[0101] During the summation process, the index used to calculate the fitted polynomial is... The value at that point is obtained by comparing it with the filter coefficients. Multiply and sum the results to obtain the fitted value, then compare it with the original data. Compare the results and calculate the sum of squared errors.

[0102] The objective function aims to find an optimal set of filter coefficients. , making the function The goal is to minimize the sum of squared errors between the original data and the fitted polynomial within a local window. This results in smoother data processing, reduced noise interference, and a better reflection of the true trend of data changes.

[0103] By taking the partial derivative of the objective function and setting it to zero, we can obtain information about... The system of linear equations is used to solve for the optimal filter coefficients. Its specific value depends on the window size and the order of the polynomial used.

[0104] Step 3: Establish a filtering processing model by combining the above filtering coefficients, window width, window half-width, and relative offset.

[0105] For the original NDVI data sequence { },in This is the index of the current data point to be processed. Relative within the window The offset, the output value after filtering. The calculation formula needs to satisfy the following relationship:

[0106]

[0107] in, This represents the output value after filtering. Indicates the half-width of the filter window. Indicates relative within the window The offset, Representation and position The relevant filter coefficients, This indicates the value of the original data within the window.

[0108] , Indicates window width (odd number), variable The range of values ​​is arrive All integers.

[0109] Then, the original NDVI time series dataset is processed using the above filtering model to obtain the filtered NDVI time series data.

[0110] By processing the raw NDVI time-series dataset using a filtering model, filtered NDVI time-series data can be obtained, improving the spatiotemporal coverage of effective data. Simultaneously, binary satellite observation data can also be collected. This data primarily covers NDVI and its associated bands, facilitating subsequent pixel information identification and analysis, and providing crucial information for subsequent cloud coverage level classification and data processing.

[0111] Furthermore, the NDVI time series data processing method in this embodiment is merely an optional condition of the present invention. In other embodiments, the NDVI time series data processing method can be adjusted according to the actual situation of the original time series dataset and data acquisition requirements, which can more effectively remove outliers, ensure the integrity and continuity of the original time series data, and thus improve data quality.

[0112] S3. Set up an adaptive optimization mechanism for NDVI time-series data based on cloud coverage level. Based on the adaptive optimization mechanism and the cloud coverage level classification results, correct NDVI time-series data of different time series to obtain the target NDVI time-series data set. The specific implementation details are as follows:

[0113] In this embodiment, an adaptive optimization mechanism for NDVI time-series data based on cloud coverage level was constructed according to the cloud coverage level classification results.

[0114] After successfully obtaining relevant data on cloud coverage, based on the cloud volume analysis results The NDVI data from the growing season are classified into different grades. Since cloud cover can affect the accuracy and reliability of NDVI data, and the more cloud cover there is, the more significant the interference with the data, the more important it is to classify the data for proper use.

[0115] Based on this, according to the cloud cover analysis results Cloud coverage levels were determined, and the NDVI data from the growing season was divided into the following three quality levels:

[0116] Level 1 pixels (high-quality data) When cloud cover falls within this range, it means there is relatively little cloud cover in the sky. At this time, the NDVI data is less affected by cloud interference, and the relevant data can reflect the true growth status of vegetation. It can be directly used as an important basis for various studies and decisions, and can provide support for the study of vegetation dynamics.

[0117] Level II pixels (partial interference) Within this range, cloud cover is relatively high, causing some cloud interference to the NDVI data. Although the data still retains a certain degree of usability, the cloud cover introduces potential errors. To improve data accuracy and ensure the reliability of subsequent analysis and application results, optimization is necessary to further eliminate information bias caused by cloud interference.

[0118] Level III pixels (severe interference) When cloud cover reaches or exceeds 0.7, it indicates severe cloud cover, resulting in significant interference with NDVI data and poor data quality. Directly using this data for analysis will lead to substantial biases in the results, affecting the accurate analysis and judgment of vegetation conditions. Therefore, for such data, it is essential to comprehensively utilize multiple data fusion techniques and model algorithms for correction.

[0119] To further optimize NDVI time-series data of different quality levels and enable it to function effectively at various quality levels, this embodiment primarily incorporates adaptive optimization mechanisms for NDVI time-series data at levels I, II, and III. These multi-level optimization mechanisms allow for personalized processing of NDVI time-series data based on the characteristics and needs of pixels at different quality levels, thereby improving the overall data quality and application effectiveness. A schematic diagram of the NDVI time-series data adaptive optimization mechanism is further illustrated; please refer to [link / reference]. Figure 2 .

[0120] In the NDVI time-series data adaptive optimization mechanism, a Level I pixel NDVI time-series data adaptive optimization mechanism is set up when... Approaching infinitely close to 0, specifically in The interval is determined to be sunny and cloudless at this time. The adaptive optimization mechanism of NDVI time series data of Class I pixels is used to adjust and optimize it, which aims to provide a solid and reliable data foundation for subsequent vegetation analysis and research.

[0121] First, based on the adaptive optimization mechanism of the NDVI time series data of the Class I pixels, the complete detection results of the NDVI time series data of the Class I pixels are obtained, and the first Class I pixel NDVI time series data is obtained based on the above complete detection results.

[0122] An adaptive optimization mechanism for Level I pixel NDVI time-series data is used to perform a data integrity check. The core objective of this step is to check for missing or outlier values ​​in the data, thereby ensuring the integrity and accuracy of the data. After the integrity check, the first Level I pixel NDVI time-series data can be obtained.

[0123] Assume the observation time number is At that time, the first NDVI data at each moment Missing data may occur. In this case, valid data from the immediately preceding and following timestamps should be used. and Perform linear interpolation calculations to fill in the gaps. Missing values.

[0124] The calculation formula for the above linear interpolation method is as follows:

[0125]

[0126] in, The linear interpolation method is used to obtain... data, express NDVI data at each time point, express NDVI data at each time point, express The time of moment, express The time of moment, express The time of a moment.

[0127] This method can effectively handle the problem of missing values ​​in the data, ensuring the continuity and integrity of time series data.

[0128] Then, the first level I pixel NDVI time series data is smoothed using the adaptive optimization mechanism of the level I pixel NDVI time series data to obtain the second level I pixel NDVI time series data.

[0129] After completing the data integrity check, the first level I pixel NDVI time series data was obtained. Then, the level I pixel NDVI time series data was smoothed using an adaptive optimization mechanism. The main purpose of the smoothing process is to reduce the impact of random noise on the level I pixel NDVI time series data, making the data curve smoother, which is more conducive to subsequent analysis and trend extraction.

[0130] In this embodiment, the following is adopted: The point moving average method is used to smooth the NDVI time-series data of Level I pixels. For multiple acquired ( The NDVI data at each of the valid observation times must satisfy the following relationship: Regarding the first Data points ,based on The point moving average method is used for smoothing, and after smoothing... The value satisfies the following calculation formula:

[0131]

[0132] in, Indicates the smoothed result , Indicates the sequence number corresponding to the observation time. To resize the movable window, This represents NDVI data at different observation times.

[0133] Indicates to Round down to the nearest integer. The size of the movable window is usually an odd number to ensure symmetry.

[0134] The above smoothing method can effectively eliminate random fluctuations in the data, making the NDVI time series data of Class I pixels more stable and reliable.

[0135] Finally, mean analysis is performed on the second level I pixel NDVI time series data based on the adaptive optimization mechanism of level I pixel NDVI time series data to obtain the target NDVI time series data of level I pixels.

[0136] Based on the adaptive optimization mechanism of NDVI time-series data for Level I pixels, mean analysis is performed on the smoothed NDVI time-series data of the second Level I pixels to obtain the target NDVI time-series data for Level I pixels. The purpose of mean analysis is to calculate the average NDVI value during the growing season, which helps to effectively reflect the vegetation cover of Level I pixels throughout the entire growing season.

[0137] Assuming that a certain pixel was acquired during the growing season... The NDVI data at each valid observation time are denoted as follows: In this embodiment, the arithmetic mean method is used to calculate the average value of the pixel during its growth season. The calculation formula is as follows:

[0138]

[0139] in, This represents the NDVI data calculated using the averaging method. Indicates the sequence number corresponding to the observation time. This represents NDVI data at different observation times.

[0140] in, This represents the sequence number corresponding to the observation time. It is calculated using this formula. It can comprehensively reflect the vegetation cover of the pixel throughout the entire growing season.

[0141] The target NDVI time-series data of Class I pixels obtained by averaging can more accurately and comprehensively reflect the vegetation cover of the pixel throughout the growing season, providing an important reference for subsequent vegetation research and analysis.

[0142] The aforementioned adaptive optimization mechanism for Level I pixel NDVI time-series data accurately identifies and processes missing values ​​in the data through integrity checks, and linear interpolation ensures the continuity of Level I pixel NDVI time-series data in the time dimension. The smoothing process effectively reduces the impact of random noise on the data, smoothing based on local characteristics of the data to eliminate random fluctuations and make the data curve smoother and more stable. Mean analysis can comprehensively reflect the vegetation cover of Level I pixels throughout the growing season, thereby more realistically reflecting the vegetation cover level and enabling the vegetation cover index to more accurately reflect the actual vegetation condition.

[0143] A further schematic diagram of the processing flow of the adaptive optimization mechanism for Level I pixel NDVI time-series data is shown below. Please refer to [link / reference] for details. Figure 3 .

[0144] In the NDVI time-series data adaptive optimization mechanism, a Level II pixel NDVI time-series data adaptive optimization mechanism is set up when... It is in the middle value range, specifically in The interval was determined to be under medium cloud coverage, and the Level II pixel NDVI time series data adaptive optimization mechanism was used to adjust and optimize it.

[0145] When the NDVI time series data of Class II pixels is partially interfered with by clouds, the clouds will block and scatter the vegetation reflection signal to a certain extent, which will cause the NDVI value to deviate. In order to eliminate or reduce the influence of clouds and improve data quality, a specific correction method for the NDVI time series data of Class II pixels is required.

[0146] First, the NDVI time series data of Level II pixels is periodically analyzed using an adaptive optimization mechanism for Level II pixel NDVI time series data to obtain Level II pixel NDVI time series data at different time points.

[0147] An adaptive optimization mechanism for Level 2 pixel NDVI time-series data is employed to perform periodic analysis on the data, thereby obtaining Level 2 pixel NDVI time-series data at different time points. Since harmonic analysis technology can decompose time-series data into harmonic components of different frequencies, it can capture the periodic changes in the data.

[0148] For NDVI time-series data of Class II pixels, the harmonic-based data analysis model can be expressed as:

[0149]

[0150] in, This represents the NDVI data at time point t. This represents the constant term in the harmonic analysis process. Indicates the number of harmonics. Indicates the order of harmonics. This represents the first harmonic coefficient corresponding to different orders. Indicates the period of the data. Represents a time variable. This represents the second harmonic coefficients corresponding to different orders. This represents the time series error term.

[0151] The Normalized Difference Vegetation Index (NDVI) values ​​at different time points are the target variables analyzed and predicted in this embodiment. They can reflect the growth status and coverage of vegetation at different times. The NDVI values ​​are usually between -1 and 1. Positive values ​​indicate vegetation coverage areas, and the larger the value, the higher the vegetation coverage. Negative values ​​generally indicate non-vegetation areas such as water bodies and snow. Values ​​close to 0 may indicate bare soil.

[0152] The average level of NDVI time series data represents a baseline value of NDVI without considering periodic variations, reflecting the overall basic state of vegetation in the region.

[0153] The number of harmonics determines the quantity of harmonic components of different frequencies in the model. Increasing the number of harmonics N allows for the capture of more complex periodic variations. In practical applications, the appropriate number of harmonics should be selected based on the data characteristics and research objectives. The optimal number of harmonics can be determined by observing the autocorrelation function, power spectrum, and other characteristics of NDVI time series data, or by using information criteria. .

[0154] Harmonic order Different orders of harmonics correspond to different frequency components, with lower-order harmonics... High-order harmonics typically reflect long-term, large-scale periodic changes in data, such as annual cycles, while they capture shorter-period, more subtle fluctuations.

[0155] First harmonic coefficients corresponding to different orders and Together, they determine the amplitude and phase of different harmonic components, affecting the degree and manner in which the harmonic contributes to the change in NDVI value.

[0156] For NDVI time series data, the data period can be one year (days, months, quarters), because vegetation growth and changes have a clear annual cycle and are influenced by the periodicity of factors such as seasons and climate. If studying other phenomena with different periodicities, the data period needs to be adjusted accordingly. .

[0157] The time variable can be a specific point in time, such as a date (in days) or a sequence number in a time series, used to determine the degree to which the NDVI value is affected by various harmonic components at a specific moment.

[0158] Second harmonic coefficients corresponding to different orders and The synergistic effect determines the amplitude and phase of different harmonics, thereby affecting the periodic variation characteristics of the NDVI time series.

[0159] The time series error term includes the impact of other environmental factors on the NDVI value, such as measurement errors, sudden natural disasters, and uncertainties in human activities.

[0160] The error term is an independent and identically distributed random variable, and follows a mean of 0 and a variance of . It follows a normal distribution.

[0161] Then, based on the variation pattern of NDVI values ​​of Class II pixels at different time points, the time series data of NDVI of Class II pixels at different time points are corrected and optimized to obtain the target NDVI time series data of Class II pixels.

[0162] Based on the variation patterns of NDVI values ​​of Level II pixels at different time points, the time-series NDVI data of Level II pixels at different time points are corrected and optimized to obtain target NDVI time-series data of Level II pixels. For time-series NDVI data, the variation patterns of NDVI values ​​at the same location at different times can be used for correction. Since vegetation growth and changes have certain periodicity and seasonality, a time-series model can be established to predict NDVI values ​​during cloud cover periods, and then the predicted values ​​are compared and corrected with the actual observed values. When the actual observed values ​​deviate due to cloud interference, the actual observed values ​​can be adjusted using methods such as weighted averaging and error correction based on the difference between the predicted and actual values, thereby obtaining more accurate target NDVI time-series data of Level II pixels that reflect the vegetation status.

[0163] Harmonic analysis technology decomposes the NDVI time-series data of Class II pixels into harmonic components of different frequencies, enabling precise capture of the periodic patterns of vegetation growth and change. Vegetation growth typically exhibits a clear annual cycle, influenced by factors such as season and climate. Harmonic analysis can clearly identify these long-term, large-scale periodic changes, as well as shorter-period, more subtle fluctuations, thus more accurately reflecting the true changes in NDVI values.

[0164] A further schematic diagram of the processing flow of the adaptive optimization mechanism for NDVI time-series data of Level II pixels is shown below. Please refer to [link / reference] for details. Figure 4 .

[0165] In the NDVI time-series data adaptive optimization mechanism, a Level III pixel NDVI time-series data adaptive optimization mechanism is set up when... Gradually approaching 1, that is, at The interval was determined to be in a state of severe cloud coverage, and the Level III pixel NDVI time series data adaptive optimization mechanism was used to adjust and optimize it.

[0166] First, obtain optical and microwave remote sensing data of the target area;

[0167] Optical and microwave remote sensing data for the target area need to be acquired. To ensure data availability and consistency, optical and microwave remote sensing data should be collected from the same area and within the same time period. Simultaneously, it should be ensured that the optical remote sensing data includes the red (R) and near-infrared (NIR) bands used to calculate NDVI, while the microwave remote sensing data requires the selection of appropriate polarization and imaging modes.

[0168] Then, the Level III pixel NDVI time-series data adaptive optimization mechanism calibrates the optical remote sensing data and the microwave remote sensing data to obtain calibrated optical remote sensing data and calibrated microwave remote sensing data. This process consists of two steps:

[0169] The first step is geometric correction: Geometric correction is performed on optical and microwave remote sensing data to give them a unified geographic coordinate system and projection method, ensuring that the two types of data can be accurately matched in space. In this embodiment, the ground control point (GCP) correction method is used, selecting a sufficient number of evenly distributed ground control points and using a polynomial correction model for correction.

[0170] The second step involves optimizing and calibrating the optical remote sensing data and the microwave remote sensing data separately:

[0171] For optical remote sensing data, the raw digital values ​​(DN) are mainly converted into radiance values ​​based on sensor parameters and calibration formulas. This data is then further converted to surface reflectance using an atmospheric correction model. In this example, Landsat 8 optical remote sensing data is used, and its radiometric calibration formula is as follows:

[0172]

[0173] in, Indicates the radiance value. Gain coefficient representing the radiance value. Represents the original numerical value of DN. The offset index represents the radiance value.

[0174] For microwave remote sensing data, calibration is mainly performed based on its imaging principle and sensor parameters, converting the raw data into backscattering coefficients. In this embodiment, Sentinel-1 microwave remote sensing data is used as an example, and its calibration formula is as follows:

[0175]

[0176] in, This represents the converted backscattering coefficient. These represent the cells in the i-th row and j-th column, Represents the grayscale value of the original image. This indicates the calibration parameters.

[0177] The calibrated optical remote sensing data and calibrated microwave remote sensing data can be obtained by following the implementation steps.

[0178] Next, the relationship between optical and microwave remote sensing was analyzed by combining the adaptive optimization mechanism of NDVI time-series data of Level III pixels, calibrated optical remote sensing data, and calibrated microwave remote sensing data.

[0179] An adaptive optimization mechanism for Level III pixel NDVI time-series data was used, combined with calibrated optical and microwave remote sensing data, to analyze the relationship between the two. The correlation between them was visually presented through methods such as scatter plotting and calculating correlation coefficients. Specifically, the Pearson correlation coefficient between optical remote sensing NDVI values ​​and microwave remote sensing correlation parameters was calculated as follows: Its calculation formula satisfies the following relationship:

[0180]

[0181] in, This represents the Pearson correlation coefficient. Indicates the number of samples This represents the i-th sample value of the optical remote sensing NDVI value. This represents the mean of the optical remote sensing NDVI values. This represents the i-th sample value of the microwave remote sensing correlation parameter. This represents the mean value of relevant microwave remote sensing parameters.

[0182] Correlation coefficient The range of values ​​is The closer the absolute value is to 1, the stronger the linear correlation between the two.

[0183] Finally, a neural network-based data fusion model is constructed in the adaptive optimization mechanism of NDVI time series data of Level III pixels; and the calibrated optical remote sensing data and calibrated microwave remote sensing data are fused together with the above data fusion model and relationship to obtain the target NDVI time series data of Level III pixels.

[0184] First, a fusion model is established: a fusion model is built based on a multilayer perceptron (MLP) neural network, which consists of an input layer, a hidden layer, and an output layer. The input layer receives microwave remote sensing features, the hidden layer transforms and extracts the input features through a nonlinear activation function, and the output layer outputs the predicted optical remote sensing NDVI value.

[0185] The input layer has m neurons, the hidden layer has h neurons, and the output layer has 1 neuron.

[0186] The weight matrix from the input layer to the hidden layer is set as follows: The bias vector is set to ;

[0187] The weight matrix from the hidden layer to the output layer is set as follows: The bias vector is set to ;

[0188] Then the output H of the hidden layer and the output of the output layer They respectively satisfy the following relations:

[0189]

[0190] in, This represents the output of the hidden layer. This represents the activation function of the hidden layer. This represents the weight matrix from the input layer to the hidden layer. It is the input microwave remote sensing feature vector. This represents the bias vector of the hidden layer.

[0191]

[0192] in, This indicates the output of the output layer. This represents the activation function of the output layer. This represents the weight matrix from the hidden layer to the output layer. This represents the output of the hidden layer. This represents the bias vector of the output layer.

[0193] and These represent the activation functions of the hidden layer and the output layer, respectively. These activation functions can be ReLU, Sigmoid, etc. Simultaneously, the neural network is trained using a training dataset, and the weights and biases are updated through backpropagation to minimize the error between the predicted and true values.

[0194] Then, the target NDVI time series data of Level III pixels is obtained by data fusion through the fusion model.

[0195] A fusion model is used to convert microwave remote sensing data into NDVI values ​​similar to optical remote sensing data. This involves inputting microwave remote sensing features into the fusion model to obtain predicted NDVI values, and then using a weighted fusion method to compare the predicted NDVI values ​​with the original optical remote sensing NDVI values.

[0196] Let the weighting coefficients be... and ,and The fused NDVI value for:

[0197]

[0198] in, This represents the data after Level III pixel fusion processing. express The weighting coefficients, This indicates the output of the output layer. express The weighting coefficients, This represents the original optical remote sensing NDVI value.

[0199] The above and Adjustments can be made based on the degree of cloud interference; for areas severely affected by cloud interference, A larger value can be selected; This indicates that the output of the output layer is the predicted NDVI value; This represents the original optical remote sensing NDVI value, the data value of the area not severely affected by cloud interference.

[0200] Based on the implementation steps, adaptive optimization processing was performed on the NDVI time-series data of Level III pixels to obtain the target NDVI time-series data of Level III pixels. Furthermore, a schematic diagram of the processing flow of the adaptive optimization mechanism for Level III pixel NDVI time-series data was drawn. Please refer to [link / reference needed] for details. Figure 5 .

[0201] In the embodiment, the NDVI time series data adaptive optimization mechanism based on cloud coverage level can obtain the target NDVI time series data set.

[0202] An adaptive optimization mechanism for NDVI time-series data based on cloud coverage levels is used to obtain the target NDVI time-series data set. Specifically, for NDVI data at different cloud coverage levels (Level I, II, and III), the adaptive optimization mechanism based on cloud coverage levels is implemented in each embodiment, enabling efficient and accurate acquisition of the target NDVI time-series data set. This mechanism fully considers the impact of cloud coverage on NDVI data quality, classifying the data into Level I, II, and III according to different cloud coverage levels.

[0203] For NDVI data at different cloud cover levels, targeted data adjustment and optimization operations were implemented. For Level I cloud cover data (where cloud cover is relatively low and has a relatively small impact on NDVI data), integrity checks, noise removal, and mean analysis were mainly performed to further improve the accuracy and stability of the data. For Level II cloud cover data (where cloud cover is moderate and causes some interference to the data), an interpolation method based on data from similar surrounding areas was used to fill in some data gaps caused by cloud cover. For Level III cloud cover data (where cloud cover is high and seriously affects data quality), multi-source data fusion and fusion models were used to restore the true surface vegetation information.

[0204] After a series of adjustments and optimizations, the corrected target NDVI time-series data was obtained. Subsequently, this data was systematically integrated according to chronological order, strictly adhering to the continuity and logic of the time series during the integration process to ensure accurate and seamless data connection at each time point. Ultimately, a complete and coherent time-series data set was constructed.

[0205] Furthermore, the method for obtaining the target NDVI time series data set in this embodiment is merely an optional condition of the present invention. In other embodiments, the method for obtaining the target NDVI time series data set can be adjusted according to the actual environmental conditions of cloud coverage and the collection requirements of NDVI time series data. It can also be combined with satellite remote sensing data, ground observation data, and meteorological data to improve the accuracy and reliability of data acquisition, and further ensure the reliability and accuracy of the target NDVI time series data set.

[0206] S4. Analyze the vegetation cover index of the target area based on the aforementioned target NDVI time-series data set to obtain the vegetation cover index analysis results, thereby achieving real-time dynamic monitoring of the vegetation cover status of the target area. The specific implementation details are as follows:

[0207] In this embodiment, a suitable vegetation cover index calculation model needs to be selected based on the topography, vegetation type, and research objectives of the target area. These models may include pixel-based binary models, hybrid pixel decomposition models, etc. The pixel-based binary model assumes that each pixel consists of a vegetated portion and a bare soil portion, and calculates the vegetation cover index through the linear relationship between NDVI values ​​and vegetation cover. The hybrid pixel decomposition model considers the mixing of multiple land cover types within a pixel, and accurately calculates the vegetation cover index by establishing an endmember spectral library and a linear mixing model.

[0208] Using the selected computational model and combining it with the target NDVI time-series data set, the vegetation cover index of the target area at different time points is calculated pixel by pixel. During the calculation process, the growth cycles and seasonal variation characteristics of different vegetation types also need to be fully considered, and the computational parameters need to be adjusted accordingly.

[0209] In one alternative embodiment, since the vegetation growth in forest areas is relatively stable, relatively fixed parameters can be used for calculation; while for farmland areas, the parameters need to be dynamically adjusted according to the planting season and growth stage of the crops to ensure the accuracy of the calculation results.

[0210] At the same time, the calculated vegetation cover index needs to be analyzed from both temporal and spatial dimensions.

[0211] In terms of time, a curve of vegetation cover index change over time in the target area is plotted to analyze its interannual and seasonal variation patterns. In the example, by comparing the vegetation cover index curves of different years, it is determined whether the vegetation cover situation in the target area shows an improving or deteriorating trend; the seasonal variation curve is analyzed to understand the growth status and cover change characteristics of vegetation in different seasons.

[0212] In the spatial dimension, using Geographic Information System (GIS) technology, the vegetation cover index is visualized in the form of thematic maps, intuitively presenting the differences in vegetation cover in different areas within the target region. Through spatial clustering analysis, hotspot analysis, and other methods, high-value and low-value areas of vegetation cover are identified, and their spatial distribution characteristics and influencing factors are analyzed.

[0213] The method for calculating the vegetation cover index based on the dual dynamic cloud optimization mechanism also includes the establishment of a visualization system.

[0214] The visualization system utilizes relevant visualization software, such as ArcGIS and ENVI, to present vegetation cover index analysis results and real-time dynamic monitoring results in an intuitive and easy-to-understand manner. It can also create dynamic charts and 3D models to present changes in vegetation cover in the target area from multiple angles and in a comprehensive manner. Specifically, dynamic charts demonstrate the dynamic changes in the vegetation cover index over time; and 3D models visually present the spatial distribution relationship between the topography and vegetation cover of the target area.

[0215] A method based on a dual dynamic cloud optimization mechanism for calculating vegetation cover index is used to periodically generate vegetation cover index analysis results and real-time dynamic monitoring reports for target areas. These reports include an overview of the monitoring area, data sources and processing methods, vegetation cover index calculation results, spatiotemporal change analysis, early warning and response measures for abnormal situations, and results visualization. This provides scientific basis and decision support for ecological protection, resource management, and agricultural planning.

[0216] The above method can make full use of the target NDVI time series data set to comprehensively and deeply analyze the vegetation cover index of the target area, realize real-time dynamic monitoring of vegetation cover in the target area, and provide strong support for the sustainable development of the regional ecological environment.

[0217] Please see Figure 6In an optional embodiment, to efficiently execute the method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism provided by this invention, this invention also provides a system for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism. The system includes a processor, an input device, an output device, and a memory, which are interconnected. The memory stores a computer program, which includes program instructions. The processor is configured to call the program instructions to execute the specific steps of the method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism and related embodiments provided by this invention. The system for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism of this invention has a complete structure and is objectively stable.

[0218] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism, characterized in that, The method includes: Obtain the original time-series NDVI dataset of the target area, construct a dynamic cloud coverage analysis mechanism, and dynamically classify the cloud coverage level of different time series based on the dynamic cloud coverage analysis mechanism and the original time-series NDVI dataset to obtain the cloud coverage level classification results of different time series. A filtering model is set up, and the original NDVI time series dataset is processed by the filtering model to obtain different time series NDVI time series data after filtering. An adaptive optimization mechanism for NDVI time series data based on cloud coverage level is set up. Based on the adaptive optimization mechanism for NDVI time series data and the cloud coverage level classification results, different time series NDVI time series data are corrected to obtain the target NDVI time series data set. The vegetation cover index of the target area is analyzed based on the target NDVI time series data set to obtain the vegetation cover index analysis results. The process of acquiring the original NDVI time-series dataset of the target area, constructing a dynamic cloud coverage analysis mechanism, and dynamically classifying the cloud coverage level at different time series based on the dynamic cloud coverage analysis mechanism and the original NDVI time-series dataset to obtain cloud coverage level classification results at different time series includes: Sky imaging data at different time series in the target area were obtained based on the NDVI raw time series dataset; The cloud coverage dynamic analysis mechanism is used to dynamically analyze the sky imaging data to obtain new feature matrices of different time series in the target area; The cloud cover dynamic analysis mechanism and the new feature matrix are combined to analyze the cloud cover analysis results at different time series in the target area; The cloud coverage dynamic analysis mechanism dynamically classifies the cloud coverage level at different time periods based on the cloud cover analysis results, so as to obtain the cloud coverage level classification results at different time periods. The dynamic analysis of the sky imaging data through the cloud coverage dynamic analysis mechanism to obtain new feature matrices of the target area at different time series includes: In the cloud coverage dynamic analysis mechanism, a sky imaging feature extraction function is established, and features are extracted from the sky imaging data based on the sky imaging feature extraction function to obtain sky imaging feature matrices at different time series. In the dynamic analysis mechanism of cloud coverage, feature matrix segmentation conditions are set, and the sky imaging feature matrix is ​​segmented based on the feature matrix segmentation conditions to obtain feature matrix segmentation results at different time series. In the dynamic analysis mechanism of cloud coverage, a pixel determination function is set, and the feature matrix segmentation result is determined by the pixel determination function to obtain the determination result of different time-series feature matrices; Based on the determination result, the sky imaging feature matrix is ​​dynamically adjusted and optimized to obtain new feature matrices for different time sequences in the target area.

2. The method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism according to claim 1, characterized in that, The setting of the filtering processing model involves processing the original NDVI time series dataset using the filtering processing model to obtain different time series NDVI time series data after filtering, including: Set the filter window parameters, which include the window width, the window half-width, and the relative offset within the window; The filter coefficients are determined using the least squares objective function. A filtering processing model is established by combining the filtering coefficients, the window width, the window half-width, and the relative offset. The original NDVI time series dataset is processed using the filtering model to obtain filtered NDVI time series data.

3. The method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism according to claim 1, characterized in that, The analysis results of cloud cover at different time series in the target area, combining the dynamic analysis mechanism of cloud cover and the new feature matrix, include: A sky cloud cover prediction model is established within the aforementioned dynamic cloud cover analysis mechanism. Based on the aforementioned sky cloud cover prediction model and the new feature matrix, the analysis results of sky cloud cover at different time series in the target area are analyzed.

4. The method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism according to claim 1, characterized in that, The aforementioned NDVI time-series data adaptive optimization mechanism based on cloud coverage level is used to correct NDVI time-series data of different time series based on the NDVI time-series data adaptive optimization mechanism and the cloud coverage level classification results, so as to obtain the target NDVI time-series data set, including: An adaptive optimization mechanism for NDVI time-series data based on cloud coverage level classification results is constructed. The NDVI time-series data adaptive optimization mechanism includes a Level I pixel NDVI time-series data adaptive optimization mechanism, a Level II pixel NDVI time-series data adaptive optimization mechanism, and a Level III pixel NDVI time-series data adaptive optimization mechanism.

5. The method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism according to claim 4, characterized in that, The setting of Level I, Level II, and Level III NDVI time-series data adaptive optimization mechanisms in the NDVI time-series data adaptive optimization mechanism includes: Based on the adaptive optimization mechanism of the Level I pixel NDVI time series data, the complete detection result of the Level I pixel NDVI time series data is obtained, and the first Level I pixel NDVI time series data is obtained based on the complete detection result. The first level I pixel NDVI time series data is smoothed using the adaptive optimization mechanism for the level I pixel NDVI time series data to obtain the second level I pixel NDVI time series data. The mean analysis of the second level I pixel NDVI time series data is performed based on the adaptive optimization mechanism of the level I pixel NDVI time series data to obtain the target NDVI time series data of the level I pixel.

6. The method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism according to claim 4, characterized in that, The setting of Level I, Level II, and Level III NDVI time-series data adaptive optimization mechanisms in the NDVI time-series data adaptive optimization mechanism includes: The adaptive optimization mechanism for Level II pixel NDVI time series data is used to perform periodic analysis on the Level II pixel NDVI time series data to obtain Level II pixel NDVI time series data at different time points; Based on the variation pattern of NDVI values ​​of Class II pixels at different time points, the time-series data of NDVI of Class II pixels at different time points are corrected and optimized to obtain the target NDVI time-series data of Class II pixels.

7. The method for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism according to claim 4, characterized in that, The setting of Level I, Level II, and Level III NDVI time-series data adaptive optimization mechanisms in the NDVI time-series data adaptive optimization mechanism includes: Obtain optical and microwave remote sensing data of the target area; The adaptive optimization mechanism for the NDVI time series data of the Class III pixels performs calibration processing on the optical remote sensing data and the microwave remote sensing data to obtain calibrated optical remote sensing data and calibrated microwave remote sensing data. The relationship between optical remote sensing and microwave remote sensing is analyzed by combining the adaptive optimization mechanism of the NDVI time series data of the Class III pixels, the calibrated optical remote sensing data, and the calibrated microwave remote sensing data. A neural network-based data fusion model is constructed in the adaptive optimization mechanism for the NDVI time-series data of the Level III pixels. The calibrated optical remote sensing data and the calibrated microwave remote sensing data are fused together using the data fusion model and the relationship to obtain the target NDVI time series data of level III pixels.

8. A system for calculating vegetation cover index based on a dual dynamic cloud optimization mechanism, characterized in that, The system includes a processor, an input device, an output device, and a memory, which are interconnected. The memory stores a computer program, which includes program instructions. The processor is configured to invoke the program instructions to execute the method for calculating the vegetation cover index based on a dual dynamic cloud optimization mechanism as described in any one of claims 1-7.

Images