Afforestation time identification method, device and equipment, storage medium and product

Abstract

The invention discloses an afforestation time identification method, device and equipment, a storage medium and a product, and relates to the technical field of artificial intelligence. According to the method, the corresponding feature extraction sub-modules are set for different data types in the preset classification model, the advantage features of different data types in the multi-source data can be extracted during feature extraction, and fusion is performed through the feature fusion module, so that the earth surface features of the target area can be reflected more comprehensively, and the user experience is improved. Therefore, the classification result of the preset classification model is more accurate and credible; on the basis of a more accurate classification result, the afforestation time of the target area is obtained by comparing the classification results in different periods, and the accuracy of newly-added afforestation time recognition can be remarkably improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and particularly to a method, device, equipment, storage medium and product for identifying afforestation time. Background Art

[0002] In the field of forestry carbon sinks, especially for newly added afforestation activities (defined as afforestation behaviors carried out on planned afforestation lands with a canopy density lower than 0.2), the accurate identification and compliance verification of afforestation time are crucial.

[0003] In related technologies, the commonly used methods at home and abroad include relying on engineering documents such as design reports and acceptance reports, and using the construction and acceptance time therein as the basis for judging the afforestation time. However, these documents may have problems such as missing information and incomplete data.

[0004] Therefore, how to achieve more accurate identification of afforestation time is an urgent problem to be solved currently. Summary of the Invention

[0005] The main purpose of this application is to provide a method, device, equipment, storage medium and product for identifying afforestation time, aiming to solve the technical problem of how to achieve more accurate identification of afforestation time.

[0006] To achieve the above purpose, this application proposes a method for identifying afforestation time, which includes: Obtain multi-source data of the target area in multiple periods, where the multi-source data includes land use planning data, high-resolution satellite image data, and hyperspectral satellite image data; For each of the periods, input the multi-source data corresponding to each period into a preset classification model to obtain multiple sub-areas obtained by classifying the target area according to tree species types by the preset classification model; wherein, the preset classification model includes a feature extraction module, a feature fusion module, and a classification module, the feature extraction module includes multiple feature extraction sub-modules, the number of the feature extraction sub-modules is the same as the number of data types of the multi-source data and they correspond to each other one by one, each of the feature extraction sub-modules is respectively used to extract features from the multi-source data of the corresponding data type, the feature fusion module is used to fuse the features extracted by all the feature extraction sub-modules, and the classification module is used to use the data output by the feature fusion module to classify the target area according to tree species types to obtain multiple sub-areas; Compare and analyze the output contents of the preset classification models corresponding to different periods to obtain the afforestation time of the target area.

[0007] In some embodiments, before obtaining the multi-source data of the target area in multiple periods, the method for identifying afforestation time further includes For a sample area, obtain corresponding original data, where the original data includes corresponding national land planning data, multi-temporal high-resolution satellite image data, and multi-temporal hyperspectral satellite image data; Perform transformation processing on the original data to obtain sample data, where the sample data includes multiple forest areas, and each forest area has a corresponding forest type label; Build a deep learning model, and train the deep learning model based on the sample data to obtain the preset classification model.

[0008] In some embodiments, the performing transformation processing on the original data to obtain sample data includes: Perform data cleaning, data standardization, and data cropping on the original data to obtain the sample data; Among them, the data cleaning includes: removing pixel values outside the set regional pixel value range in the original data; The data standardization includes: converting the coordinate values of each pixel in the original data in the original coordinate system to the preset coordinate system to obtain the corresponding coordinate values in the preset coordinate system; The data cropping includes: splitting the original data into multiple data sets; all data in each data set corresponds to the same coordinate range and the central grid points correspond one by one.

[0009] In some embodiments, the building a deep learning model, and training the deep learning model based on the sample data to obtain the preset classification model includes: Obtain the aerial high-definition image data corresponding to the sample area, where for each forest area in the aerial high-definition image data, a corresponding forest type label is manually added; Build a deep learning model, input the sample data into the deep learning model to obtain a sample output; Based on the sample output and the aerial high-definition image data, construct a loss function, and train the deep learning model to obtain the preset classification model.

[0010] In some embodiments, the sample data includes multiple data subsets, and the training the deep learning model based on the sample data to obtain the preset classification model includes: Train the deep learning model based on at least some of the data subsets in the sample data to obtain a model to be optimized; Based on the remaining data subsets in the sample data, perform cross-validation on the model to be optimized, and optimize to obtain the preset classification model according to the validation results.

[0011] In addition, to achieve the above object, the present application also provides a forestation time recognition device, which includes: A data acquisition module, configured to acquire multi-source data of a target area in multiple periods, where the multi-source data includes land use planning data, high-resolution satellite image data, and hyperspectral satellite image data; A result generation module, for each of the periods, inputting the multi-source data corresponding to each period into a preset classification model, and obtaining a plurality of sub-areas obtained by classifying the target area according to tree species types by the preset classification model; wherein, the preset classification model includes a feature extraction module, a feature fusion module, and a classification module, the feature extraction module includes a plurality of feature extraction sub-modules, the number of the feature extraction sub-modules is the same as the number of data types of the multi-source data and they are in one-to-one correspondence with each other, each of the feature extraction sub-modules is respectively configured to extract features from the multi-source data of the corresponding data type, the feature fusion module is configured to fuse the features extracted by all the feature extraction sub-modules, and the classification module is configured to use the data output by the feature fusion module to classify the target area according to tree species types to obtain a plurality of sub-areas; and is further configured to compare the sub-areas of different periods to obtain the forestation time of the target area.

[0012] In addition, to achieve the above object, the present application also provides a forestation time recognition device, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, where the computer program is configured to implement the steps of the forestation time recognition method as described above.

[0013] In addition, to achieve the above object, the present application also provides a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium, and when the computer program is executed by a processor, it implements the steps of the forestation time recognition method as described above.

[0014] In addition, to achieve the above object, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the forestation time recognition method as described above.

[0015] One or more technical solutions proposed by the present application have at least the following technical effects: By setting corresponding feature extraction sub - modules for different data types in the preset classification model, when performing feature extraction, the dominant features of different data types in multi - source data can be extracted and fused through the feature fusion module, which can more comprehensively reflect the surface features of the target area, thus making the classification result of the preset classification model more accurate and reliable; based on the more accurate classification result, by comparing the classification results of different periods, the afforestation time of the target area can be obtained, which can significantly improve the accuracy of identifying the new afforestation time. Brief Description of the Drawings

[0016] The drawings herein are incorporated into the specification and form a part of this specification, showing embodiments consistent with the present application and, together with the specification, are used to explain the principles of the present application.

[0017] To more clearly illustrate the technical solutions in the embodiments of the present application or in the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 Shows a schematic flow chart of the afforestation time identification method provided by an embodiment of the present application; Figure 2 Shows a schematic diagram of dividing forest areas in high - definition image data provided by an exemplary embodiment of the present application; Figures 3 to 5 Shows a schematic structural diagram of the preset classification model provided by an exemplary embodiment of the present application; Figure 6 Shows a schematic diagram of the afforestation time of the target area provided by an exemplary embodiment of the present application; Figure 7 Shows a schematic structural diagram of the afforestation time identification device provided by an embodiment of the present application; Figure 8 Shows a schematic diagram of the structure of the afforestation time identification device provided by an embodiment of the present application.

[0019] The realization of the purpose, functional features and advantages of the present application will be further described with reference to the embodiments and the drawings. Detailed Embodiments

[0020] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0021] To better understand the technical solutions of the present application, the following will be described in detail in combination with the drawings of the specification and the specific embodiments.

[0022] The main solution of the embodiments of this application is: Obtain multi-source image data of the target area, where the multi-source image data is obtained based on multiple data sources; Input the multi-source image data into a preset classification model to obtain the areas of different forest types in the target area and the corresponding afforestation times; among them, the preset classification model is trained based on sample data with preset labels, and the sample data is obtained based on the same multiple data sources.

[0023] In the related art, in the field of forestry carbon sinks, especially for newly added afforestation activities (defined as afforestation behaviors carried out on planned afforestation lands with a canopy density lower than 0.2), the accurate identification and compliance verification of the afforestation time are crucial. Currently, the commonly used methods at home and abroad include relying on engineering documents such as design reports and acceptance reports, and using the construction and acceptance times therein as the basis for judging the afforestation time. In addition, in 2024, Google, Meta, and Microsoft jointly released an innovative newly added afforestation identification system. This system accurately identifies land use types by analyzing multi-temporal satellite multi-spectral remote sensing data, and determines the afforestation time based on changes in land type. This system particularly uses Sentinel-2 and Landsat satellite data to achieve a spatial resolution of nearly 10 meters.

[0024] However, when using the former method (i.e., obtaining the afforestation time through engineering documents), the authenticity of the obtained results is difficult to verify because these documents often have problems such as missing historical materials and incomplete data, resulting in the authenticity being difficult to be independently verified, thus making the authenticity of the results questioned. The results obtained by using the latter identification system have insufficient accuracy because in the aspect of land type identification using satellite remote sensing technology, the spatial resolution of multi-spectral satellites such as Sentinel-2 and Landsat-8 / 9 used by Google, etc., is greater than 10 meters. Although it has effectively identified large-area forested lands, for newly added forest lands, especially those with an afforestation time of less than 10 years, mostly non-forested lands with a low canopy density, the existing remote sensing technology is difficult to accurately identify, and it is unable to effectively distinguish forest lands from wastelands, grasslands, etc., resulting in confused results and low accuracy.

[0025] In summary, how to achieve more accurate identification of the afforestation time is an urgent problem to be solved currently.

[0026] Based on this, this application provides a solution. By extracting various features from multi-source data for model training, including national land planning data, multi-temporal satellite remote sensing data, and hyperspectral satellite image data, etc., a more accurate preset classification model is used for regional division, and based on comparison, accurate identification of the newly added afforestation scope and afforestation time is achieved.

[0027] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication, and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an afforestation time recognition device that can implement the above functions. Hereinafter, taking the afforestation time recognition device as an example, this embodiment and the following embodiments will be described.

[0028] Referring to Figure 1 , Figure 1 shows a schematic flow chart of an afforestation time recognition method provided by an embodiment of the present application. The afforestation time recognition method can be applied to an afforestation time recognition device, including the following steps S110 to step S120: Step S110, obtaining multi-source data of the target area in multiple periods.

[0029] Among them, the multi-source data may include national land planning data, multi-temporal high-resolution satellite image data, and multi-temporal hyperspectral satellite image data. It can be understood that the national land planning data, multi-temporal high-resolution satellite image data, and multi-temporal hyperspectral satellite image data are obtained through different acquisition sources.

[0030] In this embodiment, the target area refers to the area where the afforestation time needs to be recognized currently. In some implementation manners, satellites equipped with different relevant devices can be used to separately photograph the target area, so as to obtain the multi-temporal high-resolution satellite image data and multi-temporal hyperspectral satellite image data in the multi-source data.

[0031] The national land planning data as described above includes, but is not limited to, land use data, land planning data, and topographic and geomorphic data. The national land planning data is usually collected and sorted out by government departments or professional institutions for guiding land development, ecological protection, resource management, and urban and rural planning. It can be understood that the target area can have corresponding national land planning data.

[0032] The multi-temporal high-resolution satellite image data refers to the image data with relatively high resolution obtained at different time points (or time periods). The resolution of the image can be approximately understood as the clarity of the image, which will not be elaborated in this embodiment.

[0033] The multi-temporal hyperspectral satellite image data refers to the satellite image data with high spectral resolution obtained at different time points (multi-temporal). Among them, hyperspectral means that the image data has high spectral resolution, that is, the image can capture the spectral information of multiple narrow bands. Compared with traditional images (usually only having several bands, such as red, green, blue, near infrared), the hyperspectral image contains dozens or even hundreds of bands, and the spectral range of each band is relatively narrow (usually less than 10 nanometers). This high spectral resolution enables the image to more accurately identify the spectral characteristics of surface substances.

[0034] Step S120: For each period, input the multi-source data corresponding to each period into a preset classification model to obtain multiple sub-regions obtained by classifying the target area according to forest types by the preset classification model.

[0035] Among them, the preset classification model is trained based on sample data with preset forest labels. The preset classification model can adopt a model with a self-attention mechanism, which is not limited in this embodiment.

[0036] In some embodiments, before step S110, the preset classification model can be pre-trained, which specifically includes the following steps S210 to S230.

[0037] Step S210: For the sample area, obtain the corresponding original data.

[0038] Among them, the original data includes the corresponding land use planning data, multi-temporal high-resolution satellite image data, and multi-temporal hyperspectral satellite image data. In this embodiment, it can refer to having land use planning data, multi-temporal high-resolution satellite image data, and multi-temporal hyperspectral satellite image data.

[0039] In this embodiment, the sample area refers to any area with corresponding land use planning data, multi-temporal high-resolution satellite image data, and multi-temporal hyperspectral satellite image data. It can be understood that for relevant institutions (such as map service providers), the above-mentioned relevant data of relevant regions in different historical time periods can be saved.

[0040] Step S220: Perform transformation processing on the original data to obtain sample data.

[0041] Among them, the transformation processing operations can include data cleaning, data standardization, data cropping, and obtaining forest type label operations. For further explanation, step S220 will be split into the following ① to ③ for description.

[0042] ① In this embodiment, data cleaning refers to removing the data outside the set range in the original data. Among them, the set range can be set by the staff according to the actual scenario. For example, the data outside the set range can be obvious noise points in the satellite image, such as random high-brightness or low-brightness pixels outside the set pixel value range; it can also be blurred or ghosted areas caused by other reasons, which are not limited in this embodiment.

[0043] ②In this embodiment, data standardization refers to converting the coordinates of each pixel in the original data with respect to the coordinate system corresponding to its original acquisition to the same preset coordinate system, and then obtaining the corresponding longitude and latitude coordinate values. Specifically, the geographic coordinate systems of different data can be converted to WGS-84 longitude and latitude coordinates, and then the different data can be reprojected so that they are projected to the same central point grid position.

[0044] ③In this embodiment, data cropping refers to splitting the data after the processing in ② into a data set with the same spatial range according to the original resolution and coordinate values corresponding to the data source to which the data belongs. Exemplarily, it can be splitting the spatial range corresponding to the data into a data set of 0.0003°×0.0003° (about 30 meters×30 meters) according to the original resolutions of different data.

[0045] In addition, in this embodiment, aerial high-definition image data can be collected for the sample area by a drone. First, the area can be divided based on the aerial high-definition image data, and forest type labels can be generated for each forest area. Specifically, based on the aerial high-definition image data, the staff can clearly divide the forest types and the boundaries of the forest areas, so as to form forest type labels for each divided forest area, which is convenient for parameter fine-tuning during the model training process. Exemplarily, as Figure 2 shown, Figure 2 it shows that after dividing the sample area in the aerial high-definition image data according to the forest types, multiple forest areas are formed. These forest areas can include arbors, forests, and non-forest areas. For example, the area within the white wireframe can be a forest, the area within the black wireframe can be an arbor, and the other areas are non-forest areas.

[0046] Step S230, build a deep learning model, and train the deep learning model based on the sample data to obtain a preset classification model.

[0047] After obtaining the sample data, the deep learning model can be trained based on at least part of the data subsets in the sample data to obtain a model to be optimized; based on the remaining part of the data subsets in the sample data, cross-validation is performed on the model to be optimized to optimize and obtain a preset classification model.

[0048] Exemplarily, the sample data (a total of ten copies) can be split to obtain multiple data subsets. Among them, eight copies are used for subsequent model training (as the training set), one copy is used for model testing (as the test set), and one copy is used for model verification (as the verification set). It can be understood that using cross-validation can provide a more stable and reliable estimate of the model performance.

[0049] During the training process of the model, first, characteristic information mining is carried out according to different types of input data. For land data, its characteristic type data is accurately extracted, and key elements such as land use and topography are deeply mined to obtain land use characteristics and geomorphic characteristics; for high-resolution satellite image data, spatial texture information is efficiently extracted, and details such as the shape and edge of ground objects are accurately captured to obtain texture characteristics; for multi-temporal and multi-spectral data, surface spectral characteristic information is accurately extracted, and the reflectance changes in different bands are fully analyzed to obtain spectral characteristics; different data sources have different emphases on the characteristics that can be extracted. Based on the feature extraction from multiple dimensions, it can provide a rich and accurate data basis for the subsequent processing and analysis of the model.

[0050] After that, the extracted features are input into the feature fusion module to fuse the extracted features. Among them, the feature fusion module can include a fully connected layer, a convolutional layer, and a fusion layer. Specifically, the extracted features can be added through the fully connected layer and then input into the fusion layer after feature fusion through a 1×1 convolutional layer.

[0051] In this embodiment, the preset classification model may further include a classification module. The classification module aims to divide regions according to tree types and performs classification operations based on the fused features to obtain multiple tree region data in the sample region.

[0052] In some embodiments, a reconstructor can be set. After size change, a two-dimensional image is obtained. The two-dimensional image includes the sample region, the boundaries of the tree regions divided according to tree types, and the tree type labels corresponding to each tree region. In this embodiment, a combined loss function is set. By comparing the output two-dimensional image with the aforementioned aerial high-definition image data (which has been manually partitioned and labeled with tree types), the parameters in the preset classification model are trained and adjusted by minimizing the loss function.

[0053] Specifically, the multiple loss functions can include bounding box loss (Lbox), confidence loss (Lconf), classification loss (Lcls), spatial continuity loss (Lspa), and the composed total loss function (Ltotal).

[0054] The bounding box loss function is as follows:

[0055] Among them, i and j represent grid cells and the bounding boxes within each cell respectively, is the total number of grid cells (usually S*S), B is the number of bounding boxes predicted for each grid cell; x, y, w, h represent the center coordinates (x,y) and width (w) and height (h) of the real bounding box; correspond to x, y, w, h respectively and represent the predicted bounding box values. Refers to the existence parameter of the target. When the target exists in grid cell i, the value is 1, otherwise it is 0.

[0056] The confidence loss function is as follows:

[0057] where Ci represents the probability of the existence of the target in grid cell i, is the predicted value of this probability by the model.

[0058] The classification loss function is as follows:

[0059] where k represents the class index, is the probability of the k-th class predicted by the model, and is an indicator variable, which is 1 if the target belongs to class k, otherwise it is 0.

[0060] The spatial continuity loss function is as follows:

[0061] Xuav is the vector result obtained by one-dimensional expansion of the labeled data, and Xp is the vector result of one-dimensional expansion of the predicted result. This loss is used to measure the spatial continuity loss.

[0062] The total loss function is as follows: Ltotal = λ1Lbox + λ2Lconf + λ3Lcls + λ4Lspa λ1 to λ4 are preset loss function weight coefficients.

[0063] Through the above loss functions, the difference between the model's predicted value and the true value can be measured, and the model optimization can be guided.

[0064] For easy understanding, the above model training process can be referred to Figures 3 - 5 .

[0065] It can be understood that compared with a single data source, by introducing multi-source data, the respective feature advantages of multi-source data are utilized, such as the resolution and texture recognition advantages of high-resolution satellite images, the specific spectral response advantages of multi-temporal and multi-spectral satellite remote sensing for different surface features, etc. Compared with a single criterion, the classification result is more scientific and reliable through the multi-source data feature judgment.

[0066] After obtaining the preset classification model based on the above S210 to S230, in step S120, for each of multiple periods, the multi-source data obtained in the previous step S110 can be input into the preset classification model to obtain corresponding outputs, that is, multiple sub-regions classified according to forest tree types.

[0067] Step S130, compare and analyze the output contents of the preset classification models corresponding to different periods to obtain the afforestation time of the target area.

[0068] It can be understood that the model outputs corresponding to multiple periods are all for the same target area. That is to say, the differences between the target areas in different periods can be directly determined through comparison. This difference essentially stems from new afforestation behaviors. By comparing the sub-region distribution and the area expansion of the sub-regions corresponding to different periods, the afforestation situation between different periods can be clearly known.

[0069] As an example, assume that the first period is the nth year, and the corresponding model output indicates that there is no sub-region of arbor type in the target area; while the second period is the (n + m)th year, and the corresponding model output indicates that there is a sub-region of arbor type in the target area. Then the staff can gradually infer that within m years after the nth year, there is an afforestation event for this arbor area. Based on this, the method provided in this embodiment can be used again for comparison in other periods within these m years, so as to gradually infer the afforestation time of this arbor area.

[0070] In some embodiments, the above comparison operation can be performed by the afforestation time recognition device provided in this embodiment; in other embodiments, the above comparison operation can also be performed by the staff with the naked eye, and this embodiment does not make a limitation here.

[0071] In some embodiments, after obtaining the afforestation time of the target area, the division of the specific sub-regions of the target area and the afforestation time of each sub-region can be visually displayed by making a display interface. In this embodiment, it can be displayed through a display interface pre-designed by the staff. As an example, the display interface can be as Figure 6 shown, including the target area. Each sub-region in the target area is clearly divided by a boundary line, and each sub-region is identified with a different color. Each color corresponds to a different afforestation time (or non-forested land), and is clearly displayed in the display interface (such as the lower left corner).

[0072] In the actual application of the afforestation time recognition method provided in this embodiment, while providing the afforestation time recognition result, relevant image data and their data sources in multiple periods (such as before and after afforestation in a certain sub-region) can also be provided together, so as to construct a complete evidence chain to enhance the credibility and effectiveness of the result.

[0073] Specifically, by displaying the relevant image data before and after afforestation, the surface changes can be intuitively reflected, providing intuitive evidence for the implementation of the afforestation activity. To ensure the transparency and traceability of the evidence chain, the sources of the relevant image data can be clearly disclosed while providing the afforestation time recognition result, including but not limited to relevant satellite data and unmanned aerial vehicle (UAV) aerial photography data. In some implementation manners, national land planning data, on-site investigation data of the project site, participatory rural appraisal results, etc. can also be displayed together to further ensure credibility.

[0074] This embodiment provides an afforestation time recognition method. By setting corresponding feature extraction sub-modules for different data types in a preset classification model, when performing feature extraction, the dominant features of different data types in multi-source data can be extracted, and through the feature fusion module for fusion, the surface features of the target area can be more comprehensively reflected, so that the classification result of the preset classification model is more accurate and credible; based on the more accurate classification result, by comparing the classification results in different periods, the afforestation time of the target area can be obtained, which can significantly improve the accuracy of the recognition of the newly added afforestation time.

[0075] This application also provides an afforestation time recognition device. Please refer to Figure 7 , the afforestation time recognition device 100 includes: A data acquisition module 110, configured to acquire multi-source data of the target area in multiple periods, where the multi-source data includes national land planning data, high-resolution satellite image data, and hyperspectral satellite image data; A result generation module 120 is configured to input, for each of the periods, the multi-source data corresponding to each period into a preset classification model, and obtain the distribution of multiple sub-regions obtained by classifying the target region according to forest tree types by the preset classification model; wherein, the preset classification model includes a feature extraction module, a feature fusion module, and a classification module, the feature extraction module includes multiple feature extraction sub-modules, the number of the feature extraction sub-modules is the same as and corresponds one-to-one to the number of data types of the multi-source data, each of the feature extraction sub-modules is respectively configured to extract features from the multi-source data of the corresponding data type, the feature fusion module is configured to fuse the features extracted by all the feature extraction sub-modules, and the classification module is configured to use the data output by the feature fusion module to classify the target region according to forest tree types into multiple sub-regions; and is further configured to compare and analyze the sub-regions of different periods to obtain the afforestation time of the target region.

[0076] The afforestation time recognition device 100 provided in this application adopts the afforestation time recognition method in the above embodiment, and can solve the technical problem of how to achieve more accurate recognition of afforestation time. Compared with the prior art, the beneficial effects of the afforestation time recognition device 100 provided in this application are the same as those of the afforestation time recognition method provided in the above embodiment, and other technical features in the afforestation time recognition device 100 are the same as the features disclosed in the above embodiment method, and will not be elaborated here.

[0077] This application provides an afforestation time recognition device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the afforestation time recognition method in the first embodiment above.

[0078] Reference is made below Figure 8 , which shows a schematic structural diagram of an afforestation time recognition device suitable for implementing the embodiments of this application. The afforestation time recognition device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistant), PADs (Portable Application Description), PMPs (Portable Media Player), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 8 The afforestation time recognition device shown is only an example, and should not impose any limitation on the functions and usage scope of the embodiments of this application.

[0079] As shown Figure 8 in the figure, the afforestation time recognition device 200 may include a processing device 210 (such as a central processing unit, a graphics processing unit, etc.), which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 220 or a program loaded from a storage device 230 into a random access memory (RAM: Random Access Memory) 240. In the RAM 240, various programs and data required for the operation of the afforestation time recognition device are also stored. The processing device 210, the ROM 220, and the RAM 240 are connected to each other through a bus 250. An input / output (I / O) interface 260 is also connected to the bus. Generally, the following systems may be connected to the I / O interface 260: an input device 270 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 280 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 230 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 290. The communication device 290 may allow the afforestation time recognition device to communicate with other devices wirelessly or wiredly to exchange data. Although the figure shows an afforestation time recognition device having various systems, it should be understood that it is not required to implement or have all the shown systems. More or fewer systems may be implemented or had alternatively.

[0080] Specifically, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program codes for performing the methods shown in the flowcharts. In such an embodiment, the computer program may be downloaded and installed from a network through the communication device, or installed from the storage device 230, or installed from the ROM 220. When the computer program is executed by the processing device 210, the above functions defined in the methods of the embodiments disclosed in the present application are executed.

[0081] The afforestation time recognition device provided by the present application adopts the afforestation time recognition method in the above embodiments, and can solve the technical problem of how to achieve more accurate afforestation time recognition. Compared with the prior art, the beneficial effects of the afforestation time recognition device provided by the present application are the same as those of the afforestation time recognition method provided by the above embodiments, and other technical features in the afforestation time recognition device are the same as the features disclosed in the method of the previous embodiment, which will not be elaborated here.

[0082] It should be understood that each part disclosed in this application can be implemented by hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0083] As described above, the above are only specific embodiments of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application, and all of them should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

[0084] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., computer programs) stored thereon, and the computer-readable program instructions are used to execute the afforestation time recognition method in the above embodiments.

[0085] The computer-readable storage medium provided by this application can be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) or flash memory, optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this embodiment, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system or device. The program code contained on the computer-readable storage medium can be transmitted by any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0086] The above computer-readable storage medium can be included in the afforestation time recognition device; it can also exist separately without being assembled into the afforestation time recognition device.

[0087] The above computer-readable storage medium carries one or more programs. When the above one or more programs are executed by the afforestation time recognition device, the afforestation time recognition device can write computer program code for performing the operations of this application in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages - such as Java, Smalltalk, C++; and also include conventional procedural programming languages - such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network - including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, by connecting through an Internet service provider via the Internet).

[0088] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0089] The modules involved in the embodiments described in this application can be implemented in software or in hardware. Among them, the name of the module does not constitute a limitation to the unit itself in some cases.

[0090] The readable storage medium provided by this application is a computer-readable storage medium. The computer-readable storage medium stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned afforestation time recognition method, and can solve the technical problem of how to achieve more accurate recognition of afforestation time. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by this application are the same as those of the afforestation time recognition method provided in the above embodiments, and will not be elaborated here.

[0091] This application also provides a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the steps of the above-mentioned afforestation time recognition method.

[0092] The computer program product provided by this application can solve the technical problem of how to achieve more accurate recognition of afforestation time. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as those of the afforestation time recognition method provided in the above embodiments, and will not be elaborated here.

[0093] The above are only some embodiments of this application, and do not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application by using the content of the specification and drawings of this application, or directly / indirectly applied in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for identifying afforestation time, characterized in that, The afforestation time recognition method includes: Obtaining multi-source data of the target area in multiple periods, where the multi-source data includes land use planning data, high-resolution satellite image data, and hyperspectral satellite image data; For each of the periods, inputting the multi-source data corresponding to each period into a preset classification model to obtain multiple sub-regions obtained by classifying the target area according to forest types by the preset classification model; wherein, the preset classification model includes a feature extraction module, a feature fusion module, and a classification module. The feature extraction module includes multiple feature extraction sub-modules. The number of the feature extraction sub-modules is the same as and corresponds one by one to the number of data types of the multi-source data. Each of the feature extraction sub-modules is respectively used to extract features from the multi-source data of the corresponding data type. The feature fusion module is used to fuse the features extracted by all the feature extraction sub-modules. The classification module is used to use the data output by the feature fusion module to classify the target area according to forest types to obtain multiple sub-regions; Comparing and analyzing the output contents of the preset classification models corresponding to different periods to obtain the afforestation time of the target area.

2. The afforestation time identification method according to claim 1, characterized in that Before obtaining the multi-source data of the target area in multiple periods, the afforestation time recognition method further includes For a sample area, obtaining corresponding original data, where the original data includes corresponding land use planning data, multi-temporal high-resolution satellite image data, and hyperspectral satellite image data; Performing transformation processing on the original data to obtain sample data; Building a deep learning model and training the deep learning model based on the sample data to obtain the preset classification model.

3. The afforestation time identification method according to claim 2, wherein The performing transformation processing on the original data to obtain sample data includes: Performing data cleaning, data standardization, and data cropping on the original data to obtain the sample data; Wherein, the data cleaning includes: removing pixel values outside the pixel value range of a set area from the original data; The data standardization includes: converting the coordinate values of each pixel in the original coordinate system of the original data to a preset coordinate system to obtain the corresponding coordinate values in the preset coordinate system; The data cropping includes: splitting the original data into multiple data sets; all the data in each data set correspond to the same coordinate range and the central grid points correspond one by one.

4. The afforestation time identification method according to claim 3, characterized in that The building a deep learning model and training the deep learning model based on the sample data to obtain the preset classification model includes: Obtaining the aerial high-definition image data corresponding to the sample area, wherein each forest area in the aerial high-definition image data is artificially added with a corresponding forest type label; Building a deep learning model, inputting the sample data into the deep learning model to obtain a sample output; Constructing a loss function based on the sample output and the aerial high-definition image data, and training the deep learning model to obtain the preset classification model.

5. The afforestation time identification method according to claim 2, characterized in that, The sample data includes multiple data subsets. The training the deep learning model based on the sample data to obtain the preset classification model includes: Training the deep learning model based on at least a partial data subset of the sample data to obtain a model to be optimized; Based on the remaining partial data subset of the sample data, performing cross-validation on the model to be optimized, and optimizing to obtain the preset classification model according to the validation result.

6. An afforestation time recognition device, characterized in that, The afforestation time recognition device includes: A data acquisition module, configured to acquire multi-source data of a target area in multiple periods, where the multi-source data includes national land planning data, high-resolution satellite image data, and hyperspectral satellite image data; A result generation module, for each of the periods, inputting the multi-source data corresponding to each period into a preset classification model, and obtaining a distribution of multiple sub-areas obtained by classifying the target area according to tree species types by the preset classification model; wherein, the preset classification model includes a feature extraction module, a feature fusion module, and a classification module, the feature extraction module includes multiple feature extraction sub-modules, the number of the feature extraction sub-modules is the same as the number of data types of the multi-source data and they correspond to each other one by one, each of the feature extraction sub-modules is respectively configured to extract features from the multi-source data of the corresponding data type, the feature fusion module is configured to fuse the features extracted by all the feature extraction sub-modules, and the classification module is configured to use the data output by the feature fusion module to classify the target area according to tree species types to obtain multiple sub-areas; and is further configured to compare and analyze the sub-areas in different periods to obtain the afforestation time of the target area.

7. An afforestation time recognition device, characterized in that, The afforestation time recognition device includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, where the computer program is configured to implement the steps of the afforestation time recognition method according to any one of claims 1 to 5.

8. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium, and when the computer program is executed by a processor, it implements the steps of the afforestation time recognition method according to any one of claims 1 to 5.

9. A computer program product, characterized in that, The computer program product includes a computer program, and when the computer program is executed by a processor, it implements the steps of the afforestation time recognition method according to any one of claims 1 to 5.

Images