Method for measuring forest age, terminal device and storage medium

Abstract

The application discloses a forest age estimation method, a terminal device and a storage medium. Image segmentation is performed on a normalized difference vegetation index (NDVI) time series dataset at a pixel level to obtain a normalized difference vegetation index time series dataset at a patch level. Then, sampling data is sorted according to forest age growth order to obtain a tree species normalized difference vegetation index growth feature sequence of each type of tree species. The forest age of each type of tree species corresponding to a patch with forest disturbance is estimated by identifying a forest disturbance time. The forest age of each type of tree species corresponding to a patch without forest disturbance is obtained by calculating a part in the tree species normalized difference vegetation index growth feature sequence that is most similar to the patch characteristics. Finally, the similarity between the normalized difference vegetation index time series dataset at the patch level and the tree species normalized difference vegetation index growth feature sequence of each type of tree species is calculated, the most similar tree species is selected as the tree species to which the patch belongs, and the forest age of the tree species corresponding to the patch is determined as the real forest age of the patch.

Description

Technical Field

[0001] This application relates to the field of prediction technology, and in particular to a method for calculating forest age, a terminal device, and a storage medium. Background Technology

[0002] Maintaining forest ecosystem services relies on the diversity of forest age composition, and forest age structure is crucial for maintaining species diversity and habitat protection. Forest age estimation information holds immense potential for monitoring and managing forest resources. Estimating the age of large areas of forest can provide vital information about forest dynamics and restoration processes, helping researchers and managers understand forest succession, disturbance types and frequencies, and the ecological functions of forests of different ages. Remote sensing technology can directly perform functions such as land cover classification, disturbance change monitoring, and forest structure mapping, providing forest age estimations at various spatial scales when combined with other environmental and forest distribution data.

[0003] Currently, since the time series of spectral values ​​from remote sensing satellite data can accurately detect the timing of forest disturbances within the satellite records, forest age is usually determined by calculating the timing of forest disturbances. Although this can improve the accuracy of forest age calculation, it cannot be used to calculate forest age when there is no forest disturbance within the satellite records, i.e., when there is no disturbance in the current calculation area. Therefore, a new method for forest age calculation needs to be proposed.

[0004] The above content is only used to help understand the technical solution of this application and does not imply that the above content is prior art. Summary of the Invention

[0005] This application provides a method, terminal device, and computer-readable storage medium for calculating forest age, aiming to solve the technical problem that forest age cannot be calculated when the current calculation area is undisturbed.

[0006] To achieve the above objectives, this application provides a method for calculating forest age, which includes the following:

[0007] Acquire remote sensing image data of the measurement area within a preset time range, calculate the Normalized Difference Vegetation Index (NDVI) for each image in the remote sensing image data, arrange the NDVI in chronological order, and construct a pixel-level normalized vegetation index time series dataset.

[0008] Image segmentation is performed on the pixel-level normalized vegetation index time series dataset to construct a patch-level normalized vegetation index time series dataset.

[0009] Obtain the tree species normalized vegetation index growth characteristic sequence corresponding to each type of tree species in the measurement area;

[0010] If there is no forest disturbance in the patch-level normalized vegetation index time series dataset, the age of each patch corresponding to each type of tree species is determined according to the position of each patch in the growth characteristic sequence of the tree species normalized vegetation index of each type of tree species in the patch-level normalized vegetation index time series dataset.

[0011] Based on the similarity between the patch-level normalized vegetation index time series dataset and the target local tree species normalized vegetation index growth feature sequences of each type of tree species corresponding to the location, the tree species to which each patch belongs is determined.

[0012] The forest age corresponding to the tree species to which each patch belongs is determined as the true forest age of the patch.

[0013] Optionally, the step of performing image segmentation on the pixel-level normalized vegetation index time series dataset to construct a patch-level normalized vegetation index time series dataset includes:

[0014] Using the normalized vegetation index of the first cell in each row of the pixel-level normalized vegetation index time series dataset as the starting cell, the dynamic time-normalized distance of the target cells in the same row as the starting cell is calculated sequentially.

[0015] If the dynamic temporal regularization distance of the target pixel is greater than the image segmentation threshold, the target pixel is taken as the starting pixel, and the step of sequentially calculating the dynamic temporal regularization distance of the target pixels in the same row as the starting pixel is continued; if the dynamic temporal regularization distance of the target pixel is less than the image segmentation threshold, the target pixel and the starting pixel are merged to obtain each row of patches.

[0016] Calculate the dynamic time-normalized distances of the target row patches adjacent to the row patch in sequence;

[0017] If the dynamic time-normalized distance of the target row patch is less than the image segmentation threshold, the target row patch is merged with the row patch, and the merged patches constitute the patch-level normalized vegetation index time series dataset.

[0018] Optionally, before the step of taking the target pixel as the starting pixel and continuing to calculate the dynamic time-warped distance of target pixels in the same row as the starting pixel if the dynamic time-warped distance of the target pixel is greater than the image segmentation threshold, the method includes:

[0019] The sample data in the measurement area at the current time is obtained. The temporally similar pixels corresponding to each image in the sample data are determined by visual interpretation method. The temporally similar pixels corresponding to each image are merged to obtain the corresponding initial patches.

[0020] Calculate the mean dynamic temporal regular distance between the pixels on both sides of the common boundary of each initial patch in each image, and determine the mean dynamic temporal regular distance as the image segmentation threshold.

[0021] Optionally, after the step of obtaining the tree species normalized vegetation index growth characteristic sequence of each type of tree species corresponding to the measurement area, the method further includes:

[0022] Obtain the patch normalized vegetation index growth curve after fitting the patch-level normalized vegetation index time series dataset;

[0023] Based on the rate of change curve formula, the abrupt change point of the normalized vegetation index growth curve of the patch is calculated to determine whether forest disturbance exists. The rate of change curve formula is as follows:

[0024] y'=-a1ωsin(ωx)+b1ωcos(ωx)-2a2ωsin(2ωx)+2b2ωcos(2ωx)

[0025] Where x represents the time length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index growth curve, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index growth curve, where a2 corresponds to the cosine component of the second harmonic and b2 corresponds to the sine component of the second harmonic.

[0026] Optionally, the tree species normalized vegetation index growth characteristic sequence includes the total sample distribution area of ​​each tree species at each age within all normalized vegetation index intervals, the proportion of the number of samples of each tree species at each age within all normalized vegetation index intervals to the total number of samples of that age, and the mean of the normalized vegetation index of each tree species at each age within all normalized vegetation index intervals.

[0027] Optionally, the step of determining the age of each patch corresponding to each type of tree species based on the position of each patch in the patch-level normalized vegetation index time series dataset within the tree species normalized vegetation index growth characteristic sequence of each tree species includes:

[0028] If the tree species normalized vegetation index growth feature sequence is segmented according to the time length corresponding to the preset time range, the local tree species normalized vegetation index growth feature sequence corresponding to each segment is obtained, wherein the local tree species normalized vegetation index growth feature sequence has the same time length as the patch-level normalized vegetation index time series dataset.

[0029] Obtain the mean normalized vegetation index of all pixels that make up each patch in the patch-level normalized vegetation index time series dataset within all normalized vegetation index intervals.

[0030] Based on the mean of the normalized vegetation index (NVI) of the patch within all NVI intervals, the target local growth feature sequence corresponding to the patch is determined.

[0031] The age of the patch is determined by the forest age corresponding to the normalized vegetation index growth characteristic sequence of the target local tree species.

[0032] Optionally, the step of determining the target local growth feature sequence corresponding to the patch based on the mean of the normalized vegetation index within all normalized vegetation index intervals includes:

[0033] Calculate the Euclidean distance between the mean normalized vegetation index (NVC) of the patch within all NVC intervals and the mean NVC of each age group in the growth characteristic sequence of each local tree species within all NVC intervals, and calculate the mean of the Euclidean distances. Using the mean NVC of the patch within all NVC intervals as a mask, calculate the cumulative proportion of the number of samples of each age group in each local growth characteristic sequence within all NVC intervals to the total number of samples of that age group. Using the mean NVC of the patch within all NVC intervals as a mask, calculate the cumulative sample distribution area of ​​each age group in each local growth characteristic sequence within all NVC intervals.

[0034] Principal component analysis is performed on the mean Euclidean distance, the cumulative percentage, and the cumulative area to determine the principal components, wherein the principal components are the largest eigenvalues ​​calculated based on the covariance matrix between the mean Euclidean distance, the cumulative percentage, and the cumulative area.

[0035] The principal components, the mean Euclidean distance, the cumulative percentage, and the cumulative area are input into the scoring formula to determine the score of the patch in the normalized vegetation index growth characteristic sequence of each local tree species, wherein the scoring formula F i for:

[0036] N part =1+max(Age) tree )-Length patch

[0037] Among them, Age tree Indicates forest age, Length patch This indicates the duration of the preset time range;

[0038] F i =[EDM,CP,CA] i Vector i T

[0039] Where i = 0, 1, ..., N part 'i' represents the segment number of each tree species, Vector i T Let represent the transpose of the eigenvectors corresponding to the principal components in segment i, EDM represent the mean Euclidean distance, CP represent the cumulative percentage, and CA represent the cumulative area.

[0040] Determine the maximum value among the various scores, and determine the local tree species normalized vegetation index growth characteristic sequence corresponding to the maximum value as the target local growth characteristic sequence.

[0041] Optionally, the step of determining the tree species to which each patch belongs based on the similarity between the patch-level normalized vegetation index time series dataset and the target local tree species normalized vegetation index growth feature sequences corresponding to each type of tree species at the location includes:

[0042] The patch-level normalized vegetation index time series dataset is fitted using a preset second-order Fourier function to obtain the patch-level normalized vegetation index growth curve, wherein the second-order Fourier function is:

[0043] y=a0+a1 cos(ωx)+b1 sin(ωx)+a2 cos(2ωx)+b2 sin(2ωx);

[0044] Where x represents the time length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficient a0 is the average normalized vegetation index of the patch-level normalized vegetation index time series dataset, representing the overall growth status of the tree species; coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index time series dataset, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index time series dataset, where a2 corresponds to the cosine component of the second harmonic, and b2 corresponds to the sine component of the second harmonic.

[0045] Obtain the pixel-level normalized vegetation index time series dataset corresponding to the sample data of each type of tree species corresponding to the location in the measurement area, and use the Fourier second-order function to fit the pixel-level normalized vegetation index time series dataset corresponding to the sample data to obtain the target tree species growth fitting curve corresponding to the location.

[0046] Calculate the dynamic time-normalized distance between the fitted normalized vegetation index growth curve of the patch and the target tree species growth fitted curve of each type of tree species corresponding to the location.

[0047] The tree species corresponding to the minimum value of the dynamic time-normalized distance is determined as the tree species to which the patch belongs.

[0048] Optionally, after the step of obtaining the normalized vegetation index growth characteristic sequence of each type of tree species corresponding to the measurement area, the method further includes:

[0049] If the patch-level time series dataset contains forest disturbance, obtain the disturbance time when the forest disturbance occurred and the end time of the patch-level time series dataset;

[0050] The difference between the end time and the disturbance time is determined as the forest age of each patch corresponding to each type of tree species;

[0051] Based on the similarity between the patch-level normalized vegetation index time series dataset and the tree species normalized vegetation index growth characteristic sequences of each type of tree species corresponding to the time length of the preset time range, the tree species to which each patch belongs is determined.

[0052] The forest age corresponding to the tree species to which each patch belongs is determined as the true forest age of the patch.

[0053] Optionally, the step of determining the tree species to which each patch belongs based on the similarity of the patch-level normalized vegetation index time series dataset with the tree species normalized vegetation index growth characteristic sequences corresponding to the time length of the preset time range includes:

[0054] The patch-level normalized vegetation index time series dataset is fitted using a preset second-order Fourier function to obtain the patch-level normalized vegetation index growth curve, wherein the second-order Fourier function is:

[0055] y=a0+a1 cos(ωx)+b1 sin(ωx)+a2 cos(2ωx)+b2 sin(2ωx);

[0056] Where x represents the time length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficient a0 is the average normalized vegetation index of the patch-level normalized vegetation index time series dataset, representing the overall growth status of the tree species; coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index time series dataset, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index time series dataset, where a2 corresponds to the cosine component of the second harmonic, and b2 corresponds to the sine component of the second harmonic.

[0057] Obtain the pixel-level normalized vegetation index time series dataset corresponding to the sample data of each type of tree species in the measurement area corresponding to the time length, and use the Fourier second-order function to fit the pixel-level normalized vegetation index time series dataset corresponding to the sample data to obtain the target tree species growth fitting curve corresponding to the time length.

[0058] Calculate the dynamic time-normalized vegetation index growth curve of the patch after fitting and the target tree species growth fitting curve of each type of tree species corresponding to the time length.

[0059] The tree species corresponding to the minimum value of the dynamic time-normalized distance is determined as the tree species to which the patch belongs.

[0060] In addition, to achieve the above objectives, this application also provides a terminal device, which includes: a memory, a processor, and a forest age calculation program stored in the memory and executable on the processor. When the forest age calculation program is executed by the processor, it implements the steps of the forest age calculation method as described above.

[0061] In addition, to achieve the above objectives, this application also provides a computer-readable storage medium storing a forest age calculation program, which, when executed by a processor, implements the steps of the forest age calculation method as described above.

[0062] This application proposes a method for calculating forest age, using a terminal device and a computer-readable storage medium. The method acquires remote sensing image data of the measurement area within a preset time range, calculates the normalized vegetation index (NDI) of each image in the remote sensing image data, arranges the NDI in chronological order to construct a pixel-level NDI time series dataset, and then performs image segmentation on the pixel-level NDI time series dataset to construct a patch-level NDI time series dataset. This method obtains the normalized vegetation index (NVI) growth characteristic sequences of various tree species corresponding to the measurement area. If the patch-level NVI time series dataset does not exhibit forest disturbance, the age of each patch corresponding to each tree species is determined based on its position within the NVI growth characteristic sequences of each tree species in the patch-level NVI time series dataset. The tree species to which each patch belongs is determined based on the similarity between the patch-level NVI time series dataset and the target local NVI growth characteristic sequences of each tree species corresponding to that position. The age of the tree species to which each patch belongs is then determined as the true age of the patch. In this application, for patches without forest disturbance, the tree species to which the patch belongs is determined by the similarity between the patch and the growth characteristics of various tree species, thereby determining the age of the tree species corresponding to that location as the age of the patch. Therefore, the forest age calculation method provided in this application is not limited by whether the calculation area has been disturbed, which can improve the accuracy of forest age calculation and also improve the applicability of forest age calculation. Attached Figure Description

[0063] Figure 1 This is a flowchart illustrating the first embodiment of the forest age calculation method of this application;

[0064] Figure 2 This is a detailed flowchart of step S2 in the second embodiment of the forest age calculation method of this application;

[0065] Figure 3 This is a detailed flowchart of step S2 in the third embodiment of the forest age calculation method of this application;

[0066] Figure 4 This is a schematic diagram of the terminal structure of the hardware operating environment involved in the embodiments of the present invention.

[0067] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0068] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0069] Currently, since the time series of spectral values ​​from remote sensing satellite data can accurately detect the timing of forest disturbances within the satellite records, forest age is usually determined by calculating the timing of forest disturbances. Although this can improve the accuracy of forest age calculation, it cannot be used to calculate forest age when there is no forest disturbance within the satellite records, i.e., when there is no disturbance in the current calculation area. Therefore, a new method for forest age calculation needs to be proposed.

[0070] To address the aforementioned deficiencies in related technologies, this application proposes a method for calculating forest age, the main steps of which include the following:

[0071] By acquiring remote sensing image data of the measurement area within a preset time range, the normalized vegetation index (NDI) of each image in the remote sensing image data is calculated. The NDIs are arranged in chronological order to construct a pixel-level NDI time series dataset. Then, the pixel-level NDI time series dataset is segmented to construct a patch-level NDI time series dataset. This method obtains the normalized vegetation index (NVI) growth characteristic sequences of various tree species corresponding to the measurement area. If the patch-level NVI time series dataset does not exhibit forest disturbance, the age of each patch corresponding to each tree species is determined based on its position within the NVI growth characteristic sequences of each tree species in the patch-level NVI time series dataset. The tree species to which each patch belongs is determined based on the similarity between the patch-level NVI time series dataset and the target local NVI growth characteristic sequences of each tree species corresponding to that position. The age of the tree species to which each patch belongs is then determined as the true age of the patch. In this application, for patches without forest disturbance, the tree species to which the patch belongs is determined by the similarity between the patch and the growth characteristics of various tree species, thereby determining the age of the tree species corresponding to that location as the age of the patch. Therefore, the forest age calculation method provided in this application is not limited by whether the calculation area has been disturbed, which can improve the accuracy of forest age calculation and also improve the applicability of forest age calculation.

[0072] To better understand the above technical solutions, exemplary embodiments of this application will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of this application and to fully convey the scope of this application to those skilled in the art.

[0073] Example 1

[0074] Reference Figure 1 In the first embodiment of the forest age calculation method of this application, the forest age calculation method includes the following steps:

[0075] Step S1: Obtain remote sensing image data of the measurement area within a preset time range, calculate the normalized vegetation index of each image in the remote sensing image data, arrange the normalized vegetation index in chronological order, and construct a pixel-level normalized vegetation index time series dataset.

[0076] In this embodiment, the Normalized Difference Vegetation Index (NDVI) corresponding to each pixel of each image is calculated using the near-infrared band and the red band.

[0077] The formula for calculating the normalized vegetation index is as follows:

[0078]

[0079] Wherein, NDVI represents the Normalized Difference Vegetation Index, NIR represents the near-infrared band, and R represents the red band.

[0080] It should be noted that the time length of the preset time range corresponding to the remote sensing image data of the measurement area provided in this application for forest age calculation does not need to be consistent with the time length of the growth cycle of each type of tree species.

[0081] For example, if the growth cycle of tree species in a region is 60 years (0-60 years), then the preset time range can be set to 30 years (1991-2021), thus obtaining remote sensing image data for nearly 30 years from 1991 to 2021.

[0082] Because the acquisition of remote sensing images is sometimes limited by environmental factors such as weather, there may be some pixels where effective data cannot be obtained, i.e., there are cloud and cloud shadow pixels. If cloud and cloud shadow pixels are used, it will have a certain impact on the quality of remote sensing images and analysis results, leading to inaccurate forest age calculation results.

[0083] Optionally, as an alternative implementation, after acquiring remote sensing image data within a preset time range, this application can construct a pixel-level normalized vegetation index time series dataset by performing median synthesis on the pixels of the target remote sensing image data within a preset time period each year, that is, by taking the median of multiple data points throughout the year. It should be noted that the preset time period can be set to a period with good weather conditions and a small number of cloud and cloud shadow pixels, thereby improving the accuracy of forest age calculation results.

[0084] For example, the measurement area experiences better weather conditions in autumn and winter, with fewer cloud and cloud shadow pixels. Therefore, the median of vegetation index pixels from June to December each year can be synthesized to construct a pixel-level normalized vegetation index time series dataset.

[0085] Step S2: Perform image segmentation on the pixel-level normalized vegetation index time series dataset to construct a patch-level normalized vegetation index time series dataset;

[0086] In this embodiment, this application merges adjacent normalized vegetation index (NVC) pixels with high temporal similarity into a single patch through image segmentation, thereby constructing a patch-level NVC time series dataset. After image segmentation, the basic unit for tree species classification and stand age estimation also changes from the pixel to the patch. By combining adjacent pixels into patches, the spatial distribution pattern of vegetation can be taken into account, and stand age calculation based on patches can capture more spatial information, thus providing more accurate stand age calculation results.

[0087] Step S3: Obtain the tree species normalized vegetation index growth characteristic sequence for each type of tree species corresponding to the measurement area;

[0088] In this embodiment, sampling data of tree species at different altitudes in the measurement area are obtained, as well as official survey data, which includes at least the geographical location, distribution area, forest age, dominant tree species, and tree height of the measurement area. Then, based on the sampling data and the official survey data, the normalized vegetation index growth characteristic sequence of each type of tree species is determined.

[0089] Furthermore, the normalized vegetation index (NVI) growth characteristic sequence for each type of tree species includes: the total area of ​​samples of each forest age in each interval (TASEI), the proportion of samples of each forest age in each interval to the total number of samples of that forest age (PSED), and the mean of samples of each forest age in each interval (NMSEI).

[0090] It should be noted that the normalized vegetation index (NVI) values ​​of various tree species change with the increase of forest age, but the characteristics of NVI changes in tree species during the growth cycle show interspecific differences. This is the basis for tree species classification in this invention.

[0091] Optionally, we divide the distribution range of the normalized vegetation index (NVI) of all sample data for each type of tree species in the measurement area into different intervals. For each tree species and each age, we statistically analyze the relevant NVI indicators within each interval and arrange the NVI indicators of each interval in ascending order of age. Finally, we construct three NVI growth characteristic sequences for each type of tree species: First, based on the geographical location and distribution area of ​​each tree species in the official forest survey data, we calculate the total sample distribution area of ​​each tree species and each age within all NVI intervals. The analysis includes three main aspects: First, the distribution of tree species by age across different normalized vegetation index (NVC) levels. Second, the proportion of the number of samples of each tree species by age within all NVC intervals to the total number of samples for that age, reflecting the clustering range of NVC values ​​for each tree species by age and the variation characteristics of the NVC clustering range during the growth cycle, thus visually representing the differences in growth characteristics among tree species. Third, the mean of the NVC for each tree species by age within all NVC intervals, reflecting the variation characteristics of the mean of the NVC during the growth cycle.

[0092] Step S4: If there is no forest disturbance in the patch-level normalized vegetation index time series dataset, determine the age of each patch corresponding to each type of tree species based on the position of each patch in the growth characteristic sequence of the tree species normalized vegetation index of each type of tree species in the patch-level normalized vegetation index time series dataset.

[0093] Since the patch-level normalized vegetation index time series dataset does not contain patches with forest disturbance whose forest age is longer than the preset time range, the forest disturbance time cannot be used to determine the forest age of the patches.

[0094] In this embodiment, the application segments the tree species normalized vegetation index (NVM) growth feature sequence according to the time length corresponding to a preset time range, obtaining the local tree species NVM growth feature sequence corresponding to each segment. This ensures that the local tree species NVM growth feature sequence has the same time length as the patch-level NVM time series dataset. Then, the mean NVM of all pixels constituting each patch in the patch-level NVM time series dataset is obtained within all NVM intervals. Furthermore, based on the mean NVM of the patch within all NVM intervals, the target local growth feature sequence corresponding to the patch is determined. Finally, the forest age corresponding to the target local tree species NVM growth feature sequence is determined as the forest age of the patch. It should be noted that the target local growth feature sequence is the position of the patch within the tree species NVM growth feature sequence of each tree species.

[0095] It is understandable that the number of local growth characteristic sequences obtained after segmenting each type of tree species is the same as the number of normalized vegetation index (NMSE) growth characteristic sequences for each type of tree species. These include the total area of ​​sample distribution for each age of local tree species within all NMSE intervals (Partial TASEI, PTASEI), the proportion of the number of samples for each age of local tree species within all NMSE intervals to the total number of samples for that age (Partial PSEI, PPSEI), and the mean of the NMSE for each age of local tree species within all NMSE intervals (Partial NMSEI, PNMSEI).

[0096] Optionally, the step of determining the target local growth feature sequence corresponding to the patch based on the mean normalized vegetation index (NVC) of the patch within all NVC intervals includes: calculating the Euclidean distance between the mean NVC of the patch within all NVC intervals and the mean NVC of samples of each age in the growth feature sequence of each local tree species within all NVC intervals, and calculating the Euclidean distance mean (EDM); using the mean NVC of the patch within all NVC intervals as a mask, calculating the cumulative ratio of the number of samples of each age in each local growth feature sequence within all NVC intervals to the total number of samples of that age. The method uses the mean normalized vegetation index (NVI) of a patch across all NVI intervals as a mask to calculate the cumulative area (CA) of each age group within each local growth characteristic sequence across all NVI intervals. Principal Component Analysis (PCA) is then performed on the mean Euclidean distance, cumulative proportion, and cumulative area to determine the principal components. The principal components, along with the mean Euclidean distance, cumulative proportion, and cumulative area, are then input into a scoring formula to determine the patch's score in each local tree species NVI growth characteristic sequence. The maximum value among all scores for each tree species type is then determined, and the local tree species NVI growth characteristic sequence corresponding to the maximum value in each tree species type is identified as the target local growth characteristic sequence. The age corresponding to the target local growth characteristic sequence is then determined as the age of the patch corresponding to that tree species.

[0097] In this embodiment, principal component analysis (PCA) is a commonly used data dimensionality reduction method. By retaining the most important principal components, the dimensionality of the data can be reduced while preserving as much information as possible. In this application, PCA is used to reduce the dimensionality of three factors: the mean Euclidean distance, the cumulative percentage, and the cumulative area. The principal component is the largest eigenvalue calculated from the covariance matrix between the mean Euclidean distance, the cumulative percentage, and the cumulative area. Eigenvectors and eigenvalues ​​are calculated from the covariance matrix between the variables. The eigenvectors represent the direction of the component, while the eigenvalues ​​represent the importance of the component.

[0098] The scoring formula Fi is:

[0099] N part =1+max(Age) tree )-Length patch

[0100] Among them, Agetree Indicates forest age, Length patch This indicates the length of time within the preset time range;

[0101] F i =[EDM,CP,CA] i Vector i T

[0102] Where i = 0, 1, ..., N part 'i' represents the segment number of each tree species, Vector i T Let F represent the transpose of the eigenvectors corresponding to the principal components in segment i, EDM represent the mean Euclidean distance, CP represent the cumulative proportion, CA represent the cumulative area, and F... i Indicates the rating.

[0103] Optionally, before performing principal component analysis, the mean correlation coefficients among the three variables—mean Euclidean distance, cumulative proportion, and cumulative area—of all patches can be calculated. Since a unified measure of the correlation between variables is needed, the Pearson correlation coefficient method is used to calculate the correlation coefficients. The formula for calculating the Pearson correlation coefficient is as follows:

[0104]

[0105] To determine if there is a correlation among the three variables—mean Euclidean distance, cumulative percentage, and cumulative area—if a correlation exists, principal component analysis (PCA) can be performed on them, with one principal component. The eigenvectors (Equation 2) and eigenvalues ​​(Equation 3) are calculated using the covariance matrix (Equation 1) between the variables. The eigenvectors represent the direction of the components, while the eigenvalues ​​represent the importance of the components. The component with the largest eigenvalue is the principal component, and this principal component is used to explain the information of the three variables: mean Euclidean distance, cumulative percentage, and cumulative area.

[0106] The covariance matrices of the mean Euclidean distance, cumulative percentage, and cumulative area are as follows:

[0107]

[0108]

[0109]

[0110] For example, the correlation coefficients among the three variables—mean Euclidean distance, cumulative proportion, and cumulative area—of all patches in the image can be calculated and averaged. The correlation coefficient between the mean Euclidean distance and the cumulative proportion is -0.9553, the mean Euclidean distance and the cumulative area are -0.6404, and the cumulative proportion and the cumulative area are 0.4657. The correlation coefficient between any two variables is >0.3. Meanwhile, the mean KMO test value for all patches is 0.6084 (>0.5), and the mean Bartlett's test of sphericity (p) for all patches is <0.05, indicating a correlation among the variables, making principal component analysis suitable.

[0111] Step S5: Based on the similarity between the patch-level normalized vegetation index time series dataset and the target local tree species normalized vegetation index growth characteristic sequence of each type of tree species corresponding to the location, determine the tree species to which each patch belongs;

[0112] Step S6: Determine the forest age corresponding to the tree species to which each patch belongs as the true forest age of the patch.

[0113] In this embodiment, a preset second-order Fourier function is used to fit the patch-level normalized vegetation index (NDVI) time series dataset to obtain the patch NDVI growth fitting curve (PNGFC). A pixel-level NDVI time series dataset corresponding to sample data of each tree species at the location corresponding to the measurement area is obtained, and a second-order Fourier function is used to fit the pixel-level NDVI time series dataset corresponding to the sample data to obtain the target tree species NDVI growth fitting curve (TSNGFC) for each tree species at the location. By calculating the dynamic time warping distance between the fitted patch NDVI growth curve and the target tree species growth fitting curve for each tree species at the location, the tree species corresponding to the minimum value of the dynamic time warping distance is determined as the tree species to which the patch belongs.

[0114] The second-order Fourier function is:

[0115] y=a0+a1 cos(ωx)+b1 sin(ωx)+a2 cos(2ωx)+b2 sin(2ωx);

[0116] Where x represents the length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficient a0 is the average normalized vegetation index of the patch-level normalized vegetation index time series dataset, characterizing the overall growth status of the tree species; coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index time series dataset, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index time series dataset, where a2 corresponds to the cosine component of the second harmonic, and b2 corresponds to the sine component of the second harmonic. Among these, the six initial value coefficients [a...] 0, [a1,b1,a2,b2,ω] is set to [0,0,0,0,0,0.2094].

[0117] It should be noted that the normalized vegetation index data used for fitting the curve in this application comes from the mean normalized vegetation index of each age of all sample points in the preset time range of each year. Fitting TSNGFC does not use all normalized vegetation index data within the time range, but requires screening normalized vegetation index data that can reflect the tree species growth characteristics within the time range of this segment.

[0118] Optionally, a continuous interval exceeding 60% of the PSEI for each stand age can be selected as the screening range for the Normalized Difference Vegetation Index (NDVI) value, and the TSNGFC is fitted using the NDVI data within the screening range. The tree species with the minimum dynamic time-normalized distance is the tree species classification result of the patch, and the principal component analysis result of this tree species is the true stand age of the patch.

[0119] In the technical solution disclosed in this embodiment, remote sensing image data of the measurement area within a preset time range is acquired, the normalized vegetation index of each image in the remote sensing image data is calculated, the normalized vegetation index is arranged in chronological order, and a pixel-level normalized vegetation index time series dataset is constructed. Then, the pixel-level normalized vegetation index time series dataset is segmented to construct a patch-level normalized vegetation index time series dataset. If the patch-level normalized vegetation index (NZVI) time series dataset does not exhibit forest disturbance, the NZVI growth characteristic sequences of each tree species corresponding to the measurement area are obtained. Based on the position of each patch in the NZVI growth characteristic sequences of each tree species in the patch-level NZVI time series dataset, the forest age of each patch corresponding to each tree species is determined. Based on the similarity between the patch-level NZVI time series dataset and the target local NZVI growth characteristic sequences of each tree species corresponding to that position, the tree species to which each patch belongs is determined, and the forest age corresponding to the tree species to which each patch belongs is determined as the true forest age of the patch. This application, for patches without forest disturbance, determines the tree species to which the patch belongs by the similarity between the patch and the growth characteristics of each tree species, thereby determining the forest age of the corresponding location as the forest age of the patch. Therefore, the forest age calculation method provided in this application is not limited by whether the calculation area is disturbed, which can improve the accuracy of forest age calculation and also improve the applicability of forest age calculation.

[0120] Example 2

[0121] Reference Figure 2 In the first embodiment, based on the above-described first embodiment, step S2 includes:

[0122] Step S21: Using the normalized vegetation index of the first cell in each row of the pixel-level normalized vegetation index time series dataset as the starting cell, calculate the dynamic time-normalized distance of the target cells in the same row as the starting cell.

[0123] In this embodiment, dynamic temporal warping is a method for comparing the similarity between two sequences. It performs non-linear alignment on the time axis and calculates the distance between the two sequences by finding the optimal matching path. This application uses the mean of the dynamic temporal rule distance as the image segmentation threshold to determine whether the pixels on both sides of the common boundary of each initial patch need to be merged, which can improve the accuracy and effectiveness of the merged patches, thereby improving the accuracy of forest age measurement.

[0124] Step S22: If the dynamic temporal regularization distance of the target pixel is greater than the image segmentation threshold, the target pixel is taken as the starting pixel, and the step of sequentially calculating the dynamic temporal regularization distance of the target pixels in the same row as the starting pixel is continued; if the dynamic temporal regularization distance of the target pixel is less than the image segmentation threshold, the target pixel and the starting pixel are merged to obtain each row of patches;

[0125] In this embodiment, a dynamic temporal warping distance greater than the image segmentation threshold indicates low similarity between the target pixel and the starting pixel, while a dynamic temporal warping distance less than the image segmentation threshold indicates high similarity between the target pixel and the starting pixel.

[0126] Step S23: Calculate the dynamic time-normalized distance of the target row patches adjacent to the row patch in sequence;

[0127] Step S24: If the dynamic time-normalized distance of the target row patch is less than the image segmentation threshold, the target row patch is merged with the row patch, and the merged patches constitute the patch-level normalized vegetation index time series dataset.

[0128] In this embodiment, after merging all pixels to obtain patches, adjacent patches with high similarity are merged to enrich the constructed patch-level normalized vegetation index time series dataset.

[0129] Optionally, to remove noise data from the pixel-level normalized vegetation index (NZVI) time series dataset and improve the accuracy of forest age calculation, before the step of using the NZVI of the first pixel in each row of the pixel-level NZVI time series dataset as the starting pixel and sequentially calculating the dynamic time-normalized distance of target pixels in the same row as the starting pixel, the method includes: fitting the pixel-level NZVI time series dataset with a preset Fourier second-order function to remove noise data from the pixel-level NZVI time series dataset; then continuing with the step of using the NZVI of the first pixel in each row of the pixel-level NZVI time series dataset as the starting pixel and sequentially calculating the dynamic time-normalized distance of target pixels in the same row as the starting pixel. The Fourier second-order function can be found in Embodiment 1 above, and will not be repeated here.

[0130] Optionally, before proceeding to the step of taking the target pixel as the starting pixel and continuing to calculate the dynamic time regularization distance of the target pixels in the same row as the starting pixel if the dynamic time regularization distance of the target pixel is greater than the image segmentation threshold, the method includes: acquiring sample data in the measurement area at the current time; determining the temporally similar pixels corresponding to each image in the sample data through a visual interpretation method; merging the temporally similar pixels corresponding to each image to obtain the corresponding initial patches; calculating the average dynamic time regularization distance of the pixels on both sides of the common boundary of each initial patch in each image; and determining the average dynamic time regularization distance as the image segmentation threshold.

[0131] In the technical solution provided in this embodiment, the normalized vegetation index (NDI) of the first pixel in each row of the pixel-level normalized vegetation index (NDV) time series dataset is used as the starting pixel. The dynamic temporal regularization distance of target pixels in the same row as the starting pixel is calculated sequentially. If the dynamic temporal regularization distance of the target pixel is greater than the image segmentation threshold, the target pixel is used as the starting pixel, and the step of sequentially calculating the dynamic temporal regularization distance of target pixels in the same row as the starting pixel is continued. If the dynamic temporal regularization distance of the target pixel is less than the image segmentation threshold, the target pixel and the starting pixel are... The process involves merging patches to obtain individual rows, and then sequentially calculating the dynamic time-warped distance of the target row patches adjacent to each row patch. If the dynamic time-warped distance of the target row patch is less than the image segmentation threshold, the target row patch is merged with the row patch. The merged patches constitute a patch-level normalized vegetation index time series dataset. Using the dynamic time-warped distance as the image segmentation threshold can preserve the temporal order of the data during the segmentation process, which helps to better capture the evolution process and dynamic features of the data. This can improve the accuracy and effectiveness of the merged patches, thereby improving the accuracy of forest age measurement.

[0132] Example 3

[0133] Reference Figure 3 In the third embodiment, based on any of the above embodiments, after step S2, the following is included:

[0134] Step S7: If the patch-level time series dataset has forest disturbance, obtain the disturbance time when the forest disturbance occurred and the end time of the patch-level time series dataset;

[0135] In this embodiment, forest age can be estimated by calculating the time of disturbance in the patch-level normalized vegetation index time series dataset. The forest age estimate is the time length from the time node of forest disturbance to the end time node of the normalized vegetation index time series.

[0136] Step S8: Determine the difference between the end time and the disturbance time as the forest age of each patch corresponding to each type of tree species;

[0137] Step S9: Based on the similarity between the patch-level normalized vegetation index time series dataset and the tree species normalized vegetation index growth characteristic sequences of each type of tree species corresponding to the time length of the preset time range, determine the tree species to which each patch belongs.

[0138] Step S10: Determine the forest age corresponding to the tree species to which each patch belongs as the true forest age of the patch.

[0139] In this embodiment, a preset second-order Fourier function is used to fit the patch-level normalized vegetation index (NDVI) time series dataset to obtain the patch NDVI growth fitting curve (PNGFC). A pixel-level NDVI time series dataset corresponding to sample data of each tree species at the location corresponding to the measurement area is obtained, and a second-order Fourier function is used to fit the pixel-level NDVI time series dataset corresponding to the sample data to obtain the target tree species NDVI growth fitting curve (TSNGFC) for each tree species at the location. By calculating the dynamic time warping distance between the fitted patch NDVI growth curve and the target tree species growth fitting curve for each tree species at the location, the tree species corresponding to the minimum value of the dynamic time warping distance is determined as the tree species to which the patch belongs.

[0140] It is understood that the aforementioned location refers to the local tree species normalized vegetation index (NVI) growth characteristic sequence within the tree species normalized vegetation index (NVI) growth characteristic sequence of each tree species that coincides with the time interval of the preset time range. For example, if the preset time range is 1991-2021, then the portion of the tree species normalized vegetation index growth characteristic sequence of each tree species corresponding to the time interval of 1991-2021 is the local tree species normalized vegetation index growth characteristic sequence.

[0141] The second-order Fourier function is:

[0142] y=a0+a1 cos(ωx)+b1 sin(ωx)+a2 cos(2ωx)+b2 sin(2ωx);

[0143] Where x represents the length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficient a0 is the average normalized vegetation index of the patch-level normalized vegetation index time series dataset, representing the overall growth status of the tree species; coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index time series dataset, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index time series dataset, where a2 corresponds to the cosine component of the second harmonic, and b2 corresponds to the sine component of the second harmonic. The six initial coefficients [a0, a1, b1, a2, b2, ω] are set to [0, 0, 0, 0, 0, 0.2094].

[0144] Optionally, the timing of forest disturbance can be obtained by calculating the abrupt change point of the rate of change of the spectral time series fitting curve, as shown in the formula.

[0145] y'=-a1ωsin(ωx)+b1ωcos(ωx)-2a2ωsin(2ωx)+2b2ωcos(2ωx)

[0146] Where x represents the time length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index growth curve, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index growth curve, where a2 corresponds to the cosine component of the second harmonic and b2 corresponds to the sine component of the second harmonic.

[0147] It should be noted that if forest disturbance exists, the time range of the normalized vegetation index growth characteristic sequence of the selected tree species is the same as the time range of PNGFC, which is from 0 to the estimated forest age.

[0148] In the technical solution provided in this embodiment, if there is forest disturbance in the patch-level time series dataset, the disturbance time and the end time of the patch-level time series dataset are obtained. Then, the difference between the end time and the disturbance time is determined as the forest age of each patch corresponding to each type of tree species. Then, based on the similarity between the patch-level normalized vegetation index time series dataset and the tree species normalized vegetation index growth characteristic sequences of each type of tree species corresponding to the time length of the preset time range, the tree species to which each patch belongs are determined. The forest age corresponding to the tree species to which each patch belongs is determined as the true forest age of the patch, thereby improving the efficiency of determining the true forest age of the patch.

[0149] Reference Figure 4 , Figure 4This is a schematic diagram of the terminal structure of the hardware operating environment involved in the embodiments of the present invention.

[0150] like Figure 4 As shown, the terminal may include: a processor 1001, such as a CPU; a network interface 1004; a user interface 1003; a memory 1005; and a communication bus 1002. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen, an input unit such as a keyboard, and a mouse; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be high-speed RAM or non-volatile memory, such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0151] Those skilled in the art will understand that Figure 4 The terminal structure shown does not constitute a limitation on the terminal and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0152] like Figure 4 As shown, the memory 1005, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and a forest age calculation program.

[0153] exist Figure 4 In the terminal shown, the network interface 1004 is mainly used to connect to the backend server and communicate with it; the processor 1001 can be used to call the forest age calculation program stored in the memory 1005 and perform the following operations:

[0154] Acquire remote sensing image data of the measurement area within a preset time range, calculate the normalized vegetation index of each image in the remote sensing image data, arrange the normalized vegetation index in chronological order, and construct a pixel-level normalized vegetation index time series dataset.

[0155] Image segmentation is performed on the pixel-level normalized vegetation index time series dataset to construct a patch-level normalized vegetation index time series dataset.

[0156] Obtain the tree species normalized vegetation index growth characteristic sequence corresponding to each type of tree species in the measurement area;

[0157] If there is no forest disturbance in the patch-level normalized vegetation index time series dataset, the age of each patch corresponding to each type of tree species is determined according to the position of each patch in the growth characteristic sequence of the tree species normalized vegetation index of each type of tree species in the patch-level normalized vegetation index time series dataset.

[0158] Based on the similarity between the patch-level normalized vegetation index time series dataset and the target local tree species normalized vegetation index growth feature sequences of each type of tree species corresponding to the location, the tree species to which each patch belongs is determined.

[0159] The forest age corresponding to the tree species to which each patch belongs is determined as the true forest age of the patch.

[0160] Furthermore, the processor 1001 can call the forest age calculation program stored in the memory 1005 and also perform the following operations:

[0161] Using the normalized vegetation index of the first cell in each row of the pixel-level normalized vegetation index time series dataset as the starting cell, the dynamic time-normalized distance of the target cells in the same row as the starting cell is calculated sequentially.

[0162] If the dynamic temporal regularization distance of the target pixel is greater than the image segmentation threshold, the target pixel is taken as the starting pixel, and the step of sequentially calculating the dynamic temporal regularization distance of the target pixels in the same row as the starting pixel is continued; if the dynamic temporal regularization distance of the target pixel is less than the image segmentation threshold, the target pixel and the starting pixel are merged to obtain each row of patches.

[0163] Calculate the dynamic time-normalized distances of the target row patches adjacent to the row patch in sequence;

[0164] If the dynamic time-normalized distance of the target row patch is less than the image segmentation threshold, the target row patch is merged with the row patch, and the merged patches constitute the patch-level normalized vegetation index time series dataset.

[0165] The sample data in the measurement area at the current time is obtained. The temporally similar pixels corresponding to each image in the sample data are determined by visual interpretation method. The temporally similar pixels corresponding to each image are merged to obtain the corresponding initial patches.

[0166] Calculate the mean dynamic temporal regular distance between the pixels on both sides of the common boundary of each initial patch in each image, and determine the mean dynamic temporal regular distance as the image segmentation threshold.

[0167] Furthermore, the processor 1001 can call the forest age calculation program stored in the memory 1005 and also perform the following operations:

[0168] Obtain the patch normalized vegetation index growth curve after fitting the patch-level normalized vegetation index time series dataset;

[0169] Based on the rate of change curve formula, the abrupt change point of the normalized vegetation index growth curve of the patch is calculated to determine whether forest disturbance exists. The rate of change curve formula is as follows:

[0170] y'=-a1ωsin(ωx)+b1ωcos(ωx)-2a2ωsin(2ωx)+2b2ωcos(2ωx)

[0171] Where x represents the time length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index growth curve, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index growth curve, where a2 corresponds to the cosine component of the second harmonic and b2 corresponds to the sine component of the second harmonic.

[0172] Furthermore, the processor 1001 can call the forest age calculation program stored in the memory 1005 and also perform the following operations:

[0173] If the tree species normalized vegetation index growth feature sequence is segmented according to the time length corresponding to the preset time range, the local tree species normalized vegetation index growth feature sequence corresponding to each segment is obtained, wherein the local tree species normalized vegetation index growth feature sequence has the same time length as the patch-level normalized vegetation index time series dataset.

[0174] Obtain the mean normalized vegetation index of all pixels that make up each patch in the patch-level normalized vegetation index time series dataset within all normalized vegetation index intervals.

[0175] Based on the mean of the normalized vegetation index (NVI) of the patch within all NVI intervals, the target local growth feature sequence corresponding to the patch is determined.

[0176] The age of the patch is determined by the forest age corresponding to the normalized vegetation index growth characteristic sequence of the target local tree species.

[0177] Calculate the Euclidean distance between the mean normalized vegetation index (NVC) of the patch within all NVC intervals and the mean NVC of each age group in the growth characteristic sequence of each local tree species within all NVC intervals, and calculate the mean of the Euclidean distances. Using the mean NVC of the patch within all NVC intervals as a mask, calculate the cumulative proportion of the number of samples of each age group in each local growth characteristic sequence within all NVC intervals to the total number of samples of that age group. Using the mean NVC of the patch within all NVC intervals as a mask, calculate the cumulative sample distribution area of ​​each age group in each local growth characteristic sequence within all NVC intervals.

[0178] Principal component analysis is performed on the mean Euclidean distance, the cumulative percentage, and the cumulative area to determine the principal components, wherein the principal components are the largest eigenvalues ​​calculated based on the covariance matrix between the mean Euclidean distance, the cumulative percentage, and the cumulative area.

[0179] The principal components, the mean Euclidean distance, the cumulative percentage, and the cumulative area are input into the scoring formula to determine the score of the patch in the normalized vegetation index growth characteristic sequence of each local tree species, wherein the scoring formula F i for:

[0180] N part =1+max(Age) tree )-Length patch

[0181] Among them, Age tree Indicates forest age, Length patch This indicates the duration of the preset time range;

[0182] F i =[EDM,CP,CA] i Vector i T

[0183] Where i = 0, 1, ..., N part 'i' represents the segment number of each tree species, Vector i T Let represent the transpose of the eigenvectors corresponding to the principal components in segment i, EDM represent the mean Euclidean distance, CP represent the cumulative percentage, and CA represent the cumulative area.

[0184] Determine the maximum value among the various scores, and determine the local tree species normalized vegetation index growth characteristic sequence corresponding to the maximum value as the target local growth characteristic sequence.

[0185] Furthermore, the processor 1001 can call the forest age calculation program stored in the memory 1005 and also perform the following operations:

[0186] The patch-level normalized vegetation index time series dataset is fitted using a preset second-order Fourier function to obtain the patch-level normalized vegetation index growth curve, wherein the second-order Fourier function is:

[0187] y=a0+a1 cos(ωx)+b1 sin(ωx)+a2 cos(2ωx)+b2 sin(2ωx);

[0188] Where x represents the time length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficient a0 is the average normalized vegetation index of the patch-level normalized vegetation index time series dataset, representing the overall growth status of the tree species; coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index time series dataset, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index time series dataset, where a2 corresponds to the cosine component of the second harmonic, and b2 corresponds to the sine component of the second harmonic.

[0189] Obtain the pixel-level normalized vegetation index time series dataset corresponding to the sample data of each type of tree species corresponding to the location in the measurement area, and use the Fourier second-order function to fit the pixel-level normalized vegetation index time series dataset corresponding to the sample data to obtain the target tree species growth fitting curve corresponding to the location.

[0190] Calculate the dynamic time-normalized distance between the fitted normalized vegetation index growth curve of the patch and the target tree species growth fitted curve of each type of tree species corresponding to the location.

[0191] The tree species corresponding to the minimum value of the dynamic time-normalized distance is determined as the tree species to which the patch belongs.

[0192] Furthermore, the processor 1001 can call the forest age calculation program stored in the memory 1005 and also perform the following operations:

[0193] If the patch-level time series dataset contains forest disturbance, obtain the disturbance time when the forest disturbance occurred and the end time of the patch-level time series dataset;

[0194] The difference between the end time and the disturbance time is determined as the forest age of each patch corresponding to each type of tree species;

[0195] Based on the similarity between the patch-level normalized vegetation index time series dataset and the tree species normalized vegetation index growth characteristic sequences of each type of tree species corresponding to the time length of the preset time range, the tree species to which each patch belongs is determined.

[0196] The forest age corresponding to the tree species to which each patch belongs is determined as the true forest age of the patch.

[0197] The patch-level normalized vegetation index time series dataset is fitted using a preset second-order Fourier function to obtain the patch-level normalized vegetation index growth curve, wherein the second-order Fourier function is:

[0198] y=a0+a1 cos(ωx)+b1 sin(ωx)+a2 cos(2ωx)+b2 sin(2ωx);

[0199] Where x represents the time length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficient a0 is the average normalized vegetation index of the patch-level normalized vegetation index time series dataset, representing the overall growth status of the tree species; coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index time series dataset, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index time series dataset, where a2 corresponds to the cosine component of the second harmonic, and b2 corresponds to the sine component of the second harmonic.

[0200] Obtain the pixel-level normalized vegetation index time series dataset corresponding to the sample data of each type of tree species in the measurement area corresponding to the time length, and use the Fourier second-order function to fit the pixel-level normalized vegetation index time series dataset corresponding to the sample data to obtain the target tree species growth fitting curve corresponding to the time length.

[0201] Calculate the dynamic time-normalized vegetation index growth curve of the patch after fitting and the target tree species growth fitting curve of each type of tree species corresponding to the time length.

[0202] The tree species corresponding to the minimum value of the dynamic time-normalized distance is determined as the tree species to which the patch belongs.

[0203] In addition, to achieve the above objectives, this application also provides a terminal device, which includes: a memory, a processor, and a forest age calculation program stored in the memory and executable on the processor. When the forest age calculation program is executed by the processor, it implements the steps of the forest age calculation method as described above.

[0204] In addition, to achieve the above objectives, this application also provides a computer-readable storage medium storing a forest age calculation program, which, when executed by a processor, implements the steps of the forest age calculation method as described above.

[0205] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0206] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not indicate the superiority or inferiority of the embodiments.

[0207] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which can be a computer, mobile phone, or tablet computer) to execute the methods described in the various embodiments of this application.

[0208] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for calculating forest age, characterized in that, The methods for calculating the forest age include: Acquire remote sensing image data of the measurement area within a preset time range, calculate the normalized vegetation index of each image in the remote sensing image data, arrange the normalized vegetation index in chronological order, and construct a pixel-level normalized vegetation index time series dataset. Image segmentation is performed on the pixel-level normalized vegetation index time series dataset to construct a patch-level normalized vegetation index time series dataset. Obtain the tree species normalized vegetation index growth characteristic sequence corresponding to each type of tree species in the measurement area; If there is no forest disturbance in the patch-level normalized vegetation index time series dataset, the age of each patch corresponding to each type of tree species is determined according to the position of each patch in the growth characteristic sequence of the tree species normalized vegetation index of each type of tree species in the patch-level normalized vegetation index time series dataset. Based on the similarity between the patch-level normalized vegetation index time series dataset and the target local tree species normalized vegetation index growth feature sequences of each type of tree species corresponding to the location, the tree species to which each patch belongs is determined. The forest age corresponding to the tree species to which each patch belongs is determined as the true forest age of the patch; The tree species normalized vegetation index growth characteristic sequence includes the total sample distribution area of ​​each tree species at each age within all normalized vegetation index intervals, the proportion of the number of samples of each tree species at each age within all normalized vegetation index intervals to the total number of samples of that age, and the mean of the normalized vegetation index of each tree species at each age within all normalized vegetation index intervals.

2. The method as described in claim 1, characterized in that, The step of performing image segmentation on the pixel-level normalized vegetation index time series dataset to construct a patch-level normalized vegetation index time series dataset includes: Using the normalized vegetation index of the first cell in each row of the pixel-level normalized vegetation index time series dataset as the starting cell, the dynamic time-normalized distance of the target cells in the same row as the starting cell is calculated sequentially. If the dynamic temporal regularization distance of the target pixel is greater than the image segmentation threshold, the target pixel is taken as the starting pixel, and the step of sequentially calculating the dynamic temporal regularization distance of the target pixels in the same row as the starting pixel is continued; if the dynamic temporal regularization distance of the target pixel is less than the image segmentation threshold, the target pixel and the starting pixel are merged to obtain each row of patches. Calculate the dynamic time-normalized distances of the target row patches adjacent to the row patch in sequence; If the dynamic time-normalized distance of the target row patch is less than the image segmentation threshold, the target row patch is merged with the row patch, and the merged patches constitute the patch-level normalized vegetation index time series dataset.

3. The method as described in claim 2, characterized in that, Before the step of taking the target pixel as the starting pixel and continuing to calculate the dynamic time-normalized distance of target pixels in the same row as the starting pixel if the dynamic time-normalized distance of the target pixel is greater than the image segmentation threshold, the following steps are included: The sample data in the measurement area at the current time is obtained. The temporally similar pixels corresponding to each image in the sample data are determined by visual interpretation method. The temporally similar pixels corresponding to each image are merged to obtain the corresponding initial patches. Calculate the mean dynamic temporal warping distance between the pixels on both sides of the common boundary of each initial patch in each image, and determine the mean dynamic temporal warping distance as the image segmentation threshold.

4. The method as described in claim 1, characterized in that, After the step of obtaining the tree species normalized vegetation index growth characteristic sequence of each type of tree species corresponding to the measurement area, the method further includes: Obtain the patch normalized vegetation index growth curve after fitting the patch-level normalized vegetation index time series dataset; Based on the rate of change curve formula, the abrupt change point of the normalized vegetation index growth curve of the patch is calculated to determine whether forest disturbance exists. The rate of change curve formula is as follows: ； Where x represents the time length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index growth curve, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index growth curve, where a2 corresponds to the cosine component of the second harmonic and b2 corresponds to the sine component of the second harmonic.

5. The method as described in claim 1, characterized in that, The step of determining the age of each patch corresponding to each tree species based on the position of each patch in the patch-level normalized vegetation index time series dataset within the tree species normalized vegetation index growth characteristic sequence of each tree species includes: Based on the time length corresponding to the preset time range, the normalized vegetation index growth feature sequence of the tree species is segmented to obtain the local normalized vegetation index growth feature sequence of the tree species corresponding to each segment. The local normalized vegetation index growth feature sequence of the tree species has the same time length as the patch-level normalized vegetation index time series dataset. Obtain the mean normalized vegetation index of all pixels that make up each patch in the patch-level normalized vegetation index time series dataset within all normalized vegetation index intervals. Based on the mean of the normalized vegetation index (NVI) of the patch within all NVI intervals, the target local growth feature sequence corresponding to the patch is determined. The age of the patch is determined by the forest age corresponding to the normalized vegetation index growth characteristic sequence of the target local tree species.

6. The method as described in claim 5, characterized in that, The step of determining the target local growth feature sequence corresponding to the patch based on the mean of the normalized vegetation index within all normalized vegetation index intervals includes: Calculate the Euclidean distance between the mean normalized vegetation index (NVC) of the patch within all NVC intervals and the mean NVC of each age group in the growth characteristic sequence of each local tree species within all NVC intervals, and calculate the mean of the Euclidean distances. Using the mean NVC of the patch within all NVC intervals as a mask, calculate the cumulative proportion of the number of samples of each age group in each local growth characteristic sequence within all NVC intervals to the total number of samples of that age group. Using the mean NVC of the patch within all NVC intervals as a mask, calculate the cumulative sample distribution area of ​​each age group in each local growth characteristic sequence within all NVC intervals. Principal component analysis is performed on the mean Euclidean distance, the cumulative percentage, and the cumulative area to determine the principal components, wherein the principal components are the largest eigenvalues ​​calculated based on the covariance matrix between the mean Euclidean distance, the cumulative percentage, and the cumulative area. The principal components, the mean Euclidean distance, the cumulative percentage, and the cumulative area are input into the scoring formula to determine the score of the patch in the normalized vegetation index growth characteristic sequence of each local tree species. The scoring formula F... i for: ； Among them, Age tree Indicates forest age, Length patch This indicates the length of time within the preset time range; ； Where i = 0, 1, ..., N part 'i' represents the segment number of each tree species, Vectori T Let represent the transpose of the eigenvectors corresponding to the principal components in segment i, EDM represent the mean Euclidean distance, CP represent the cumulative percentage, and CA represent the cumulative area. Determine the maximum value among the various scores, and determine the local tree species normalized vegetation index growth characteristic sequence corresponding to the maximum value as the target local growth characteristic sequence.

7. The method as described in claim 1, characterized in that, The step of determining the tree species to which each patch belongs based on the similarity between the patch-level normalized vegetation index time series dataset and the target local tree species normalized vegetation index growth feature sequences corresponding to each type of tree species at the location includes: The patch-level normalized vegetation index time series dataset is fitted using a preset second-order Fourier function to obtain the patch-level normalized vegetation index growth curve, wherein the second-order Fourier function is: ； Where x represents the time length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficient a0 is the average normalized vegetation index of the patch-level normalized vegetation index time series dataset, representing the overall growth status of the tree species; coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index time series dataset, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index time series dataset, where a2 corresponds to the cosine component of the second harmonic, and b2 corresponds to the sine component of the second harmonic. Obtain the pixel-level normalized vegetation index time series dataset corresponding to the sample data of each type of tree species corresponding to the location in the measurement area, and use the Fourier second-order function to fit the pixel-level normalized vegetation index time series dataset corresponding to the sample data to obtain the target tree species growth fitting curve corresponding to the location. Calculate the dynamic time-normalized distance between the fitted normalized vegetation index growth curve of the patch and the target tree species growth fitted curve of each type of tree species corresponding to the location. The tree species corresponding to the minimum value of the dynamic time-normalized distance is determined as the tree species to which the patch belongs.

8. The method as described in claim 1, characterized in that, After the step of obtaining the tree species normalized vegetation index growth characteristic sequence of each type of tree species corresponding to the measurement area, the following steps are included: If the patch-level time series dataset contains forest disturbance, obtain the disturbance time when the forest disturbance occurred and the end time of the patch-level time series dataset; The difference between the end time and the disturbance time is determined as the forest age of each patch corresponding to each type of tree species; Based on the similarity between the patch-level normalized vegetation index time series dataset and the tree species normalized vegetation index growth characteristic sequences of each type of tree species corresponding to the time length of the preset time range, the tree species to which each patch belongs is determined. The forest age corresponding to the tree species to which each patch belongs is determined as the true forest age of the patch.

9. The method as described in claim 8, characterized in that, The step of determining the tree species to which each patch belongs based on the similarity of the normalized vegetation index (NVI) growth characteristic sequences of each type of tree species corresponding to the time length of the patch-level NVI time series dataset and the preset time range includes: The patch-level normalized vegetation index time series dataset is fitted using a preset second-order Fourier function to obtain the patch-level normalized vegetation index growth curve, wherein the second-order Fourier function is: ； Where x represents the time length of the preset time range, ω represents the coefficient of the fitting period, approximately 0.2094 (2π / 30), coefficient a0 is the average normalized vegetation index of the patch-level normalized vegetation index time series dataset, representing the overall growth status of the tree species; coefficients a1 and b1 represent the amplitude of the fundamental frequency component of the patch-level normalized vegetation index time series dataset, a1 corresponds to the cosine component, b1 corresponds to the sine component, and coefficients a2 and b2 represent the amplitude of the second harmonic component of the patch-level normalized vegetation index time series dataset, where a2 corresponds to the cosine component of the second harmonic, and b2 corresponds to the sine component of the second harmonic. Obtain the pixel-level normalized vegetation index time series dataset corresponding to the sample data of each type of tree species in the measurement area corresponding to the time length, and use the Fourier second-order function to fit the pixel-level normalized vegetation index time series dataset corresponding to the sample data to obtain the target tree species growth fitting curve corresponding to the time length. Calculate the dynamic time-normalized vegetation index growth curve of the patch after fitting and the target tree species growth fitting curve of each type of tree species corresponding to the time length. The tree species corresponding to the minimum value of the dynamic time-normalized distance is determined as the tree species to which the patch belongs.

10. A terminal device, characterized in that, The terminal device includes: a memory, a processor, and a forest age calculation program for the terminal device stored in the memory and executable on the processor. When the forest age calculation program for the terminal device is executed by the processor, it implements the steps of the forest age calculation method for the terminal device as described in any one of claims 1 to 9.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a forest age calculation program for the terminal device, which, when executed by a processor, implements the steps of the forest age calculation method for the terminal device as described in any one of claims 1 to 9.

Images