Planting structure classification method, device, computer equipment and storage medium

By obtaining comprehensive image information and land use data sets, and using a random forest classification model to classify and process agricultural planting structures, the problem of low classification accuracy in the existing technology is solved, and more efficient planting structure classification is achieved.

CN117437475BActive Publication Date: 2024-11-26TSINGHUA UNIVERSITY
View PDF 11 Cites 0 Cited by

Patent Information

Application Number
CN202311453734.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-11-02
Publication Date
2024-11-26
Estimated Expiration
2043-11-02

AI Technical Summary

Technical Problem

The prior art has low classification accuracy in the classification of complex agricultural planting structures, mainly due to low manual identification efficiency and poor accuracy caused by seasonal influence.

Method used

By obtaining comprehensive image information, land use data sets and historical classification information, the land image information is time series fitted and classified using a random forest classification model to determine the target land supply range and improve the classification accuracy of multiple vegetation categories.

Benefits of technology

The inefficiency problem of manual identification is avoided, the classification accuracy of complex agricultural planting structures is improved, and more accurate classification of planting structures is achieved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117437475B_ABST
    Figure CN117437475B_ABST
Patent Text Reader

Abstract

The present application relates to a planting structure classification method, device, computer equipment and storage medium. The method includes: obtaining comprehensive image information of the target planting structure, a land use data set, and historical classification information of the target planting structure; based on the land use data set, determining the target land supply range of the target planting structure, and based on the historical classification information, obtaining vegetation sample information of different vegetation types; in the comprehensive image information, extracting each land image information of the target land supply range, and performing time series fitting processing on the land image information to obtain each target land image information; based on the vegetation sample information of each vegetation type, classifying and processing each target land image information through a random forest classification model to obtain each classification information of the target planting structure. The use of this method can improve the classification accuracy of complex agricultural planting structures.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of satellite remote sensing technology, and in particular to a planting structure classification method, device, computer equipment and storage medium. Background Art

[0002] Agricultural planting structure is a combination of multiple cropping index and crop type, which plays a key role in accurately calculating irrigation water consumption and crop yield estimation. The classification of planting structure is also of great significance to agricultural land management. However, due to historical and socio-economic reasons, agricultural planting structure data are usually counted according to counties and cities, with only the planting area of ​​various crops within the county and city, but lack of spatial distribution information, which makes the classification accuracy of complex agricultural planting structure low. Therefore, how to improve the classification accuracy of complex agricultural planting structure is the current research focus.

[0003] The traditional classification method of complex planting structures is to obtain image information of complex agricultural planting areas based on traditional image acquisition equipment, and then classify the complex agricultural planting structures through manual recognition to obtain classification information of the complex agricultural planting structures. However, the efficiency of manual recognition is low, and due to the influence of seasons on vegetation, the classification accuracy of manual classification based on image information is poor, resulting in low classification accuracy of complex agricultural planting structures. Summary of the invention

[0004] Based on this, it is necessary to provide a planting structure classification method, device, computer equipment, computer readable storage medium and computer program product to address the above technical problems.

[0005] In a first aspect, the present application provides a method for classifying planting structures. The method comprises:

[0006] Obtaining comprehensive image information of a target planting structure, a land use dataset, and historical classification information of the target planting structure;

[0007] Based on the land use data set, determining the target land supply range of the target planting structure, and based on the historical classification information, obtaining vegetation sample information of different vegetation types;

[0008] Extracting the land image information of the target land supply range from the comprehensive image information, and performing time series fitting processing on the land image information to obtain the target land image information;

[0009] Based on the vegetation sample information of each vegetation type, each target land image information is classified and processed through a random forest classification model to obtain each classification information of the target planting structure.

[0010] Optionally, determining a target land supply range for the target planting structure based on the land use dataset includes:

[0011] Identify the location information corresponding to the target planting structure, and based on the location information, query the land use data set for the cultivated land range corresponding to the location information as the land supply range of the target planting structure;

[0012] Based on the preset land range extension strategy and the cultivated land range, the non-cultivated land range of the target planting structure is screened in the land use data set, and the set of the cultivated land range and the non-cultivated land range is used as the target land supply range of the target planting structure.

[0013] Optionally, the obtaining vegetation sample information of different vegetation types based on the historical classification information includes:

[0014] Based on the historical classification information, identifying vegetation feature information of each vegetation type in different seasons, and identifying vegetation map information of different seasons in the historical classification information;

[0015] For each vegetation type, based on the vegetation characteristic information of the vegetation type in different seasons, in the vegetation map of each season, based on the vegetation sample screening strategy, the sub-vegetation sample information of each season is screened, and the sub-vegetation sample information of all seasons is used as the vegetation sample information of the vegetation type.

[0016] Optionally, extracting the land image information of the target land supply range from the comprehensive image information includes:

[0017] Performing interference image removal processing on each image information in the comprehensive image information to obtain vegetation image information corresponding to each image information;

[0018] Filtering target vegetation image information in the target land supply range from each of the vegetation image information, and respectively calculating a vegetation index of each target vegetation image information and a water body index of each target vegetation image information;

[0019] The target vegetation image information, the water body index of the target vegetation image information, and the vegetation index of the target vegetation image information are used as the land image information of the target land supply range.

[0020] Optionally, performing time series fitting processing on the land image information to obtain each target land image information includes:

[0021] Obtaining the acquisition time point of the image information corresponding to each land image information, and sorting the vegetation index of each land image information in a descending order of the time points to obtain a vegetation index sequence;

[0022] The vegetation index is fitted by a harmonic fitting strategy to obtain vegetation fitting indexes corresponding to all land image information, and the vegetation fitting indexes are used to replace the vegetation indexes in each land image information to obtain each target land image information.

[0023] Optionally, the vegetation sample information based on each of the vegetation types is classified and processed by a random forest classification model to obtain classification information of the target planting structure, including:

[0024] Based on the vegetation sample information of each of the vegetation types, training the classification parameters of the random forest classification model to obtain a target random forest classification model;

[0025] Input each of the target land image information into the target random forest classification model to perform vegetation classification processing to obtain classified image information of each vegetation type corresponding to each target land image information;

[0026] The classified image information of each same vegetation type is aggregated and processed to obtain the classified information of the target planting structure.

[0027] In a second aspect, the present application also provides a planting structure classification device. The device comprises:

[0028] An acquisition module, for acquiring comprehensive image information of a target planting structure, a land use data set, and historical classification information of the target planting structure;

[0029] A determination module, used to determine the target land supply range of the target planting structure based on the land use data set, and to obtain vegetation sample information of different vegetation types based on the historical classification information;

[0030] An extraction module, used to extract the land image information of the target land supply range from the comprehensive image information, and perform time series fitting processing on the land image information to obtain the target land image information;

[0031] The processing module is used to classify the target land image information based on the vegetation sample information of each vegetation type through a random forest classification model to obtain the classification information of the target planting structure.

[0032] Optionally, the determining module is specifically used to:

[0033] Identify the location information corresponding to the target planting structure, and based on the location information, query the land use data set for the cultivated land range corresponding to the location information as the land supply range of the target planting structure;

[0034] Based on the preset land range extension strategy and the cultivated land range, the non-cultivated land range of the target planting structure is screened in the land use data set, and the set of the cultivated land range and the non-cultivated land range is used as the target land supply range of the target planting structure.

[0035] Optionally, the determining module is specifically used to:

[0036] Based on the historical classification information, identifying vegetation feature information of each vegetation type in different seasons, and identifying vegetation map information of different seasons in the historical classification information;

[0037] For each vegetation type, based on the vegetation characteristic information of the vegetation type in different seasons, in the vegetation map of each season, based on the vegetation sample screening strategy, the sub-vegetation sample information of each season is screened, and the sub-vegetation sample information of all seasons is used as the vegetation sample information of the vegetation type.

[0038] Optionally, the extraction module is specifically used to:

[0039] Performing interference image removal processing on each image information in the comprehensive image information to obtain vegetation image information corresponding to each image information;

[0040] Filtering target vegetation image information in the target land supply range from each of the vegetation image information, and respectively calculating a vegetation index of each target vegetation image information and a water body index of each target vegetation image information;

[0041] The target vegetation image information, the water body index of the target vegetation image information, and the vegetation index of the target vegetation image information are used as the land image information of the target land supply range.

[0042] Optionally, the extraction module is specifically used to:

[0043] Obtaining the acquisition time point of the image information corresponding to each land image information, and sorting the vegetation index of each land image information in a descending order of the time points to obtain a vegetation index sequence;

[0044] The vegetation index is fitted by a harmonic fitting strategy to obtain vegetation fitting indexes corresponding to all land image information, and the vegetation fitting indexes are used to replace the vegetation indexes in each land image information to obtain each target land image information.

[0045] Optionally, the processing module is specifically used to:

[0046] Based on the vegetation sample information of each of the vegetation types, training the classification parameters of the random forest classification model to obtain a target random forest classification model;

[0047] Input each of the target land image information into the target random forest classification model to perform vegetation classification processing to obtain classified image information of each vegetation type corresponding to each target land image information;

[0048] The classified image information of each same vegetation type is aggregated and processed to obtain the classified information of the target planting structure.

[0049] In a third aspect, the present application provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented.

[0050] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0051] In a fifth aspect, the present application provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0052] The above-mentioned planting structure classification method, device, computer equipment and storage medium obtain comprehensive image information of the target planting structure, a land use data set, and historical classification information of the target planting structure; based on the land use data set, determine the target land supply range of the target planting structure, and based on the historical classification information, obtain vegetation sample information of different vegetation types; in the comprehensive image information, extract each land image information of the target land supply range, and perform time series fitting processing on the land image information to obtain each target land image information; and based on the vegetation sample information of each vegetation type, classify the each target land image information through a random forest classification model to obtain each classification information of the target planting structure. By intelligently determining the target land supply range, the comprehensive image information is extracted to obtain the target land image information within the target land supply range, and then based on the vegetation sample information of each vegetation type and the target land image information, classification processing is performed through the random deep forest classification model to obtain the classification information of the target planting structure, thereby avoiding the inefficiency of manual identification and manual classification. In addition, classification processing is performed through the random forest classification model to improve the classification accuracy of multiple vegetation categories in the target planting structure, thereby improving the classification accuracy of complex agricultural planting structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A schematic diagram of a process of a planting structure classification method in one embodiment;

[0054] Figure 2 A schematic diagram of a process of an example of planting structure classification in an embodiment;

[0055] Figure 3 It is a structural block diagram of a planting structure classification device in one embodiment;

[0056] Figure 4 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0058] The planting structure classification method provided in the embodiment of the present application can be applied to application environments such as agricultural planting structure, agricultural land conversion, land use type, satellite remote sensing technology, and cloud data processing. Among them, the method can be applied to a terminal, a server, and a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, etc. Among them, the terminal intelligently determines the target land supply range, thereby extracting the image information of each target land in the target land supply range of the comprehensive image information, and then based on the vegetation sample information of each vegetation type and the image information of each target land, the classification processing is performed through the random deep forest classification model to obtain the classification information of the target planting structure, thereby avoiding the inefficiency of manual identification and manual classification, and the classification processing is performed through the random forest classification model, thereby improving the classification accuracy of multiple vegetation categories in the target planting structure, thereby improving the classification accuracy of complex agricultural planting structures.

[0059] In one embodiment, Figure 1 As shown, a planting structure classification method is provided, and the method is applied to a terminal as an example for explanation, including the following steps:

[0060] Step S101, obtaining comprehensive image information of a target planting structure, a land use data set, and historical classification information of the target planting structure.

[0061] In this embodiment, the terminal obtains the comprehensive image information of the target planting structure based on the image information of different years and seasons in the area where the target planting structure is located, which is transmitted by satellite remote sensing technology and drone aerial photography technology. Among them, the target planting structure is a planting structure with diverse vegetation varieties, irregular vegetation planting intervals, and land and water areas combined. Among them, the diversity of vegetation varieties is characterized by a number of vegetation varieties greater than the number preset by the staff, and the irregular planting intervals are characterized by the planting dispersion of a single variety of vegetation greater than the dispersion preset by the staff. Then, the terminal obtains a land use data set based on the cloud, wherein the land use data set is the China Life Cycle Basic Database (China Land CoverDataset, CLCD), which is a year-by-year land use and land cover data set within the region, the time range of the data set is between ten years, and the data resolution is 30m. The overall accuracy of the data set is 79.30% ± 1.99%. After that, the terminal obtains the classified image information of each vegetation category in the area where the target planting structure is located based on the historical classification land classification database. The vegetation category is the plant category, which includes but is not limited to wheat-corn category, other double-season categories, single-season rice category, single-season corn category, single-season cotton category, single-season peanut category, other single-season categories, greenhouse crop category, economic forest category and other vegetation categories.

[0062] Step S102, based on the land use data set, determining the target land supply range of the target planting structure, and based on the historical classification information, obtaining vegetation sample information of different vegetation types.

[0063] In this embodiment, the terminal determines the target land supply range of the target planting structure based on the land use data set, and obtains vegetation sample information of different vegetation types based on historical classification information. The target land supply range includes cultivated land range and non-cultivated land range. Due to the data resolution problem of the land use data set, the non-cultivated land range is screened to ensure that the full amount of planted vegetation of the target planting structure can be identified. The non-cultivated land range is determined by the boundary of the non-cultivated land range determined by the buffer algorithm. Then, the terminal obtains vegetation sample information of different vegetation types based on the historical classification information. The vegetation sample information includes range image information of vegetation types in the land in different seasons. The specific acquisition process will be described in detail later.

[0064] Step S103: extracting the land image information of the target land supply range from the comprehensive image information, and performing time series fitting processing on the land image information to obtain the target land image information.

[0065] In this embodiment, the terminal selects the image information within the target land supply range from each image information of the comprehensive image information as the land image information. The specific extraction process will be described in detail later. Then, the terminal performs time series fitting processing on each land image information by superimposing a multi-order sine and cosine function to obtain each target land image information. The specific fitting process will be described in detail later.

[0066] Step S104, based on the vegetation sample information of each vegetation type, the target land image information is classified and processed by a random forest classification model to obtain classification information of the target planting structure.

[0067] In this embodiment, the terminal classifies each target land image information based on the vegetation sample information of each vegetation type through a random forest classification model to obtain classification information of the target planting structure.

[0068] Based on the above scheme, the target land supply range is determined intelligently, so as to extract the comprehensive image information of each target land image information within the target land supply range, and then based on the vegetation sample information of each vegetation type and each target land image information, classification processing is performed through the random deep forest classification model to obtain the classification information of the target planting structure, thereby avoiding the inefficiency of manual identification and manual classification, and classifying and processing through the random forest classification model, thereby improving the classification accuracy of multiple vegetation categories in the target planting structure, thereby improving the classification accuracy of complex agricultural planting structures.

[0069] Optionally, based on the land use data set, a target land supply range of the target planting structure is determined, including: identifying location information corresponding to the target planting structure, and based on the location information, querying the cultivated land range corresponding to the location information in the land use data set as the land supply range of the target planting structure; based on a preset land range extension strategy and cultivated land range, screening the non-cultivated land range of the target planting structure in the land use data set, and taking the set of cultivated land range and non-cultivated land range as the target land supply range of the target planting structure.

[0070] In this embodiment, the terminal identifies the location information corresponding to the target planting structure. The location information of the target planting structure is the area information to which the target planting structure belongs, for example, the target planting structure A is located in the B province C city D town E village F. The location information can also be based on the longitude and latitude information corresponding to the earth coordinate system.

[0071] Based on the location information, the terminal queries the cultivated land range corresponding to the location information in the land use data set as the land supply range of the target planting structure. Then, based on the preset land range extension strategy and cultivated land range, the terminal screens the non-cultivated land range of the target planting structure in the land use data set, and uses the cultivated land range and the set of non-cultivated land range as the target land supply range of the target planting structure. The land range extension strategy is a cultivated land boundary determination strategy based on a buffer algorithm.

[0072] Based on the above scheme, the cultivated land range and non-cultivated land range of the target planting structure are determined through the land use data set, thereby determining the target land supply range, which improves the comprehensiveness and accuracy of the determined target land supply range.

[0073] Optionally, based on historical classification information, vegetation sample information of different vegetation types is obtained, including: based on the historical classification information, identifying the vegetation characteristic information of each vegetation type in different seasons, and identifying the vegetation map information of different seasons in the historical classification information; for each vegetation type, based on the vegetation characteristic information of the vegetation type in different seasons, in the vegetation map of each season, based on the vegetation sample screening strategy, filtering the sub-vegetation sample information of each season, and using the sub-vegetation sample information of all seasons as the vegetation sample information of the vegetation type.

[0074] In this embodiment, the terminal identifies the vegetation feature information of each vegetation type in different seasons based on the historical classification information. The vegetation feature information is the image feature information of each vegetation type. The image feature information is the feature image of each vegetation type in the historical classification information obtained by the terminal in response to the upload operation of the staff. The feature image is obtained by the staff based on the driving survey method, in conjunction with the mobile phone software (containing positioning and marking functions), going deep into the farmland, recording the longitude, latitude, and current crop types of the farmland visited, and collecting the historical experience of professional planting personnel to obtain the image feature information of the crop types planted in previous years.

[0075] The terminal identifies vegetation map information of different seasons in the historical classification information, and for each vegetation type, based on the vegetation feature information of the vegetation type in different seasons, in the vegetation map of each season, based on the vegetation sample screening strategy, screens the sub-vegetation sample information of each season, and uses the sub-vegetation sample information of all seasons as the vegetation sample information of the vegetation type. The vegetation screening strategy is a strategy for determining the range image information of the vegetation type in the land. The determination strategy includes a Sentinel screening strategy and a Landsat screening strategy. The determination strategy, for example, (1) the planting structure around the point where the sub-vegetation sample information is located is consistent. Specifically, when the spatial resolution of the classification is Xm, the same type of crop should be planted within a range of 3Xm around the sub-vegetation sample information; (2) the number of category samples of each planting structure needs to meet certain standards. (In this embodiment, the number of vegetation in the single sub-vegetation sample information screened by the Sentinel screening strategy shall not be less than 400, and the number of vegetation in each category in the Landsat classification sample shall not be less than 1500).

[0076] Based on the above scheme, vegetation sample information of different vegetation types is screened through different determination strategies and based on multiple screening methods, thereby improving the accuracy of the screened vegetation sample information.

[0077] Optionally, from the comprehensive image information, each piece of land image information within the target land supply range is extracted, including: performing interference image removal processing on each piece of image information in the comprehensive image information to obtain vegetation image information corresponding to each piece of image information; screening target vegetation image information within the target land supply range from each piece of vegetation image information, and respectively calculating the vegetation index of each target vegetation image information and the water body index of each target vegetation image information; using each piece of target vegetation image information, the water body index of each target vegetation image information, and the vegetation index of each target vegetation image information as each piece of land image information within the target land supply range.

[0078] In this embodiment, the terminal performs interference image removal processing on each image information in the comprehensive image information to obtain vegetation image information corresponding to each image information. Wherein, when the image information is image information corresponding to the Sentinel-2 image, the interference image removal processing method is to use the "COPERNICUS / S2_CLOUD_PROBABILITY" technology provided by the GEE platform to remove pixels with a cloud probability of more than 65%. When the image information is image information corresponding to the Sentinel-8 image, the interference image removal processing method is to use the 'QA_PIXEL' band contained in the image to remove pixels blocked by clouds.

[0079] The vegetation image information in the target land supply range is screened from each vegetation image information, and then each vegetation image is superimposed and arranged according to the position of the vegetation image information in the target land supply range, and the overlapping parts are removed to obtain the target vegetation image information in the target land supply range.

[0080] The terminal calculates the vegetation index of each target vegetation image information and the water index of each target vegetation image information respectively. Finally, the terminal uses each target vegetation image information, the water index of each target vegetation image information, and the vegetation index of each target vegetation image information as each land image information of the target land supply range. Among them, the vegetation index is the normalized difference vegetation index (NDVI), and the water index is the automatic water extraction index (AWEI).

[0081] The calculation formula of vegetation index is:

[0082]

[0083] In the above formula, G is the green band of the image, R is the red band of the image, and NIR is the near-infrared band of the image.

[0084] The calculation formula of water index is:

[0085] AWEI=4×(G-SWIR1)-(0.25×NIR+2.75×SWIR2)

[0086] In the above formula, SWIR1 is the shortwave infrared band 1 of the image, and SWIR2 is the shortwave infrared band 2 of the image.

[0087] Based on the above solution, the comprehensiveness and accuracy of land image information recognition are improved by removing interference, overlapping, and calculating vegetation index and water body index.

[0088] Optionally, the land image information is subjected to time series fitting processing to obtain each target land image information, including: obtaining the acquisition time point of the image information corresponding to each land image information, and sorting the vegetation index of each land image information in order from the front to the back of the time point to obtain a vegetation index sequence; fitting the vegetation index through a harmonic fitting strategy to obtain the vegetation fitting index corresponding to all the land image information, and replacing the vegetation index in each land image information with the vegetation fitting index to obtain each target land image information.

[0089] In this embodiment, the terminal obtains the acquisition time point of the image information corresponding to each land image information, and sorts the vegetation index of each land image information in the order of the time point from the front to the back to obtain the vegetation index sequence; the vegetation index is fitted by the harmonic fitting strategy to obtain the vegetation fitting index corresponding to all land image information, and the vegetation fitting index replaces the vegetation index in each land image information to obtain each target land image information. Among them, the vegetation index sequence is NDVI(t)

[0090] Specifically, the fitting formula corresponding to the harmonic fitting strategy is:

[0091]

[0092] where a i and b i is the fitting parameter, and ω i =i×ω 0 refers to the angular frequency, ω 0 =2π is the fundamental frequency. The above harmonic fitting strategy can be equivalently expressed as:

[0093]

[0094] in refers to the amplitude of each order, For phase.

[0095] Among them, the amplitude and phase configurations of harmonic fitting strategies of different orders in the time domain can represent different characteristics, such as the number of agricultural seasons per year, the length of the growing season, and the peak value of NDVI in the prosperous period. Based on the comprehensive consideration of factors such as the amount of effective data, data interval, and data period, the harmonic function order that can be used is m=3, 4, and 5.

[0096] Based on the above scheme, the sorted vegetation index is subjected to harmonic fitting processing through the harmonic fitting strategy, ensuring that the obtained target land image information can be input into the random forest classification model.

[0097] Optionally, based on the vegetation sample information of each vegetation type, each target land image information is classified and processed through a random forest classification model to obtain classification information of the target planting structure, including: based on the vegetation sample information of each vegetation type, the classification parameters of the random forest classification model are trained to obtain a target random forest classification model; each target land image information is input into the target random forest classification model to perform vegetation classification processing to obtain classification image information of each vegetation type corresponding to each target land image information; the classification image information of each same vegetation type is aggregated and processed to obtain classification information of the target planting structure.

[0098] In this embodiment, the terminal trains the classification parameters of the random forest classification model based on the vegetation sample information of each vegetation type to obtain a target random forest classification model. Then, the terminal inputs each target land image information into the target random forest classification model to perform vegetation classification processing to obtain the classified image information of each vegetation type corresponding to each target land image information. Finally, the terminal aggregates the classified image information of each same vegetation type to obtain the classification information of the target planting structure. The evaluation indicators of the random forest classification model can use overall accuracy, Kappa coefficient (to measure classification accuracy) and F1 score (F1 Score), producer accuracy of various planting structure types, and user accuracy as evaluation indicators.

[0099] When there are too many types of vegetation in the target planting structure or the area of ​​the target planting structure is too large, each target land image can be divided into multiple sub-land images, and the above-mentioned classification processing steps of each target land image information through the random forest classification model are performed on each sub-land image, thereby obtaining each sub-classification information of each sub-land image. Finally, the sub-classification information of all sub-land images is used as the classification information of the target planting structure.

[0100] Based on the above solution, classification is performed through the random deep forest classification model, which improves the classification efficiency and accuracy.

[0101] This application also provides an example of planting structure classification, such as Figure 2 As shown, the specific processing process includes the following steps:

[0102] Step S201, obtaining comprehensive image information of a target planting structure, a land use data set, and historical classification information of the target planting structure.

[0103] Step S202, identifying the location information corresponding to the target planting structure, and based on the location information, querying the cultivated land range corresponding to the location information in the land use data set as the land supply range of the target planting structure.

[0104] Step S203, based on the preset land range extension strategy and cultivated land range, the non-cultivated land range of the target planting structure is screened in the land use data set, and the set of cultivated land range and non-cultivated land range is used as the target land supply range of the target planting structure.

[0105] Step S204: extracting the land image information of the target land supply range from the comprehensive image information, and performing time series fitting processing on the land image information to obtain the target land image information.

[0106] Step S205 , based on the vegetation sample information of each vegetation type, each target land image information is classified and processed by a random forest classification model to obtain each classification information of the target planting structure.

[0107] Step S206, based on the historical classification information, identifying vegetation feature information of each vegetation type in different seasons, and identifying vegetation map information of different seasons in the historical classification information.

[0108] Step S207, for each vegetation type, based on the vegetation characteristic information of the vegetation type in different seasons, in the vegetation map of each season, based on the vegetation sample screening strategy, screen the sub-vegetation sample information of each season, and use the sub-vegetation sample information of all seasons as the vegetation sample information of the vegetation type.

[0109] Step S208: performing interference image removal processing on each image information in the comprehensive image information to obtain vegetation image information corresponding to each image information.

[0110] Step S209: Filter target vegetation image information in the target land supply range from each vegetation image information, and respectively calculate the vegetation index of each target vegetation image information and the water body index of each target vegetation image information.

[0111] Step S210: taking each target vegetation image information, the water body index of each target vegetation image information, and the vegetation index of each target vegetation image information as each land image information of the target land supply range.

[0112] Step S211, obtaining the acquisition time point of the image information corresponding to each land image information, and sorting the vegetation index of each land image information in the order of the time points from front to back to obtain a vegetation index sequence.

[0113] Step S212, fitting the vegetation index by harmonic fitting strategy to obtain vegetation fitting index corresponding to all land image information, and replacing the vegetation index in each land image information with the vegetation fitting index to obtain each target land image information.

[0114] Step S213, based on the vegetation sample information of each vegetation type, training the classification parameters of the random forest classification model to obtain a target random forest classification model.

[0115] Step S214, inputting each target land image information into the target random forest classification model to perform vegetation classification processing to obtain classified image information of each vegetation type corresponding to each target land image information;

[0116] Step S215, summarizing and processing the classified image information of each same vegetation type to obtain each classified information of the target planting structure.

[0117] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0118] Based on the same inventive concept, the embodiment of the present application also provides a planting structure classification device for implementing the planting structure classification method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more planting structure classification device embodiments provided below can refer to the limitations of the planting structure classification method above, and will not be repeated here.

[0119] In one embodiment, Figure 3 As shown, a planting structure classification device is provided, including: a post-removal module 310, a determination module 320, an extraction module 330 and a processing module 340, wherein:

[0120] An acquisition module 310, for acquiring comprehensive image information of a target planting structure, a land use data set, and historical classification information of the target planting structure;

[0121] A determination module 320 is used to determine the target land supply range of the target planting structure based on the land use data set, and obtain vegetation sample information of different vegetation types based on the historical classification information;

[0122] The extraction module 330 is used to extract the land image information of the target land supply range from the comprehensive image information, and perform time series fitting processing on the land image information to obtain the target land image information;

[0123] The processing module 340 is used to classify the target land image information based on the vegetation sample information of each vegetation type through a random forest classification model to obtain classification information of the target planting structure.

[0124] Optionally, the determining module 320 is specifically configured to:

[0125] Identify the location information corresponding to the target planting structure, and based on the location information, query the land use data set for the cultivated land range corresponding to the location information as the land supply range of the target planting structure;

[0126] Based on the preset land range extension strategy and the cultivated land range, the non-cultivated land range of the target planting structure is screened in the land use data set, and the set of the cultivated land range and the non-cultivated land range is used as the target land supply range of the target planting structure.

[0127] Optionally, the determining module 320 is specifically configured to:

[0128] Based on the historical classification information, identifying vegetation feature information of each vegetation type in different seasons, and identifying vegetation map information of different seasons in the historical classification information;

[0129] For each vegetation type, based on the vegetation characteristic information of the vegetation type in different seasons, in the vegetation map of each season, based on the vegetation sample screening strategy, the sub-vegetation sample information of each season is screened, and the sub-vegetation sample information of all seasons is used as the vegetation sample information of the vegetation type.

[0130] Optionally, the extraction module 330 is specifically used to:

[0131] Performing interference image removal processing on each image information in the comprehensive image information to obtain vegetation image information corresponding to each image information;

[0132] Filtering target vegetation image information in the target land supply range from each of the vegetation image information, and respectively calculating a vegetation index of each target vegetation image information and a water body index of each target vegetation image information;

[0133] The target vegetation image information, the water body index of the target vegetation image information, and the vegetation index of the target vegetation image information are used as the land image information of the target land supply range.

[0134] Optionally, the extraction module 330 is specifically used to:

[0135] Obtaining the acquisition time point of the image information corresponding to each land image information, and sorting the vegetation index of each land image information in a descending order of the time points to obtain a vegetation index sequence;

[0136] The vegetation index is fitted by a harmonic fitting strategy to obtain vegetation fitting indexes corresponding to all land image information, and the vegetation fitting indexes are used to replace the vegetation indexes in each land image information to obtain each target land image information.

[0137] Optionally, the processing module 340 is specifically configured to:

[0138] Based on the vegetation sample information of each of the vegetation types, training the classification parameters of the random forest classification model to obtain a target random forest classification model;

[0139] Input each of the target land image information into the target random forest classification model to perform vegetation classification processing to obtain classified image information of each vegetation type corresponding to each target land image information;

[0140] The classified image information of each same vegetation type is aggregated and processed to obtain the classified information of the target planting structure.

[0141] Each module in the above-mentioned planting structure classification device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0142] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for classifying implant structures is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0143] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0144] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented.

[0145] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0146] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the steps of any one of the methods in the first aspect.

[0147] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0148] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0149] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0150] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for classifying planting structures, characterized in that: The method comprises: Obtaining comprehensive image information of a target planting structure, a land use dataset, and historical classification information of the target planting structure; Determining a target land supply range for the target planting structure based on the land use data set; Based on the historical classification information, identifying vegetation feature information of each vegetation type in different seasons, and identifying vegetation map information of different seasons in the historical classification information; For each vegetation type, based on the vegetation characteristic information of the vegetation type in different seasons, in the vegetation map of each season, based on the vegetation sample screening strategy, the sub-vegetation sample information of each season is screened, and the sub-vegetation sample information of all seasons is used as the vegetation sample information of the vegetation type; the vegetation sample screening strategy is a determination strategy for the range image information of the vegetation type in the land, and the determination strategy includes a Sentinel screening strategy and a Landsat screening strategy; Performing interference image removal processing on each image information in the comprehensive image information to obtain vegetation image information corresponding to each image information; The target vegetation image information in the target land supply range is screened from each of the vegetation image information, and the vegetation index of each target vegetation image information and the water body index of each target vegetation image information are calculated respectively; the calculation formula of the vegetation index is: In the above formula, R is the red band of the target vegetation image information, and NIR is the near infrared band of the target vegetation image information; The calculation formula of the water index is: AWEI=4×(G-SWIR1)-(0.25×NIR+2.75×SWIR2) In the above formula, G is the green band of the target vegetation image information, SWIR1 is the shortwave infrared band 1 of the target vegetation image information, and SWIR2 is the shortwave infrared band 2 of the target vegetation image information; Using each of the target vegetation image information, the water body index of each of the target vegetation image information, and the vegetation index of each of the target vegetation image information as each of the land image information of the target land supply range; Performing time series fitting processing on the land image information to obtain each target land image information; Based on the vegetation sample information of each vegetation type, each target land image information is classified and processed through a random forest classification model to obtain each classification information of the target planting structure.

2. The method according to claim 1, characterized in that: Determining the target land supply range of the target planting structure based on the land use data set includes: Identify the location information corresponding to the target planting structure, and based on the location information, query the land use data set for the cultivated land range corresponding to the location information as the land supply range of the target planting structure; Based on the preset land range extension strategy and the cultivated land range, the non-cultivated land range of the target planting structure is screened in the land use data set, and the set of the cultivated land range and the non-cultivated land range is used as the target land supply range of the target planting structure.

3. The method according to claim 1, characterized in that The performing time series fitting processing on the land image information to obtain each target land image information includes: Obtaining the acquisition time point of the image information corresponding to each land image information, and sorting the vegetation index of each land image information in a descending order of the time points to obtain a vegetation index sequence; The vegetation index is fitted by a harmonic fitting strategy to obtain vegetation fitting indexes corresponding to all land image information, and the vegetation fitting indexes are used to replace the vegetation indexes in each land image information to obtain each target land image information.

4. The method according to claim 1, characterized in that: The vegetation sample information based on each of the vegetation types is classified and processed by a random forest classification model to obtain classification information of the target planting structure, including: Based on the vegetation sample information of each of the vegetation types, training the classification parameters of the random forest classification model to obtain a target random forest classification model; Input each of the target land image information into the target random forest classification model to perform vegetation classification processing to obtain classified image information of each vegetation type corresponding to each target land image information; The classified image information of each same vegetation type is aggregated and processed to obtain the classified information of the target planting structure.

5. A planting structure classification device, characterized in that: The device comprises: An acquisition module, used to acquire comprehensive image information of a target planting structure, a land use data set, and historical classification information of the target planting structure; A determination module, for determining the target land supply range of the target planting structure based on the land use data set; identifying the vegetation feature information of each vegetation type in different seasons based on the historical classification information, and identifying the vegetation map information of different seasons in the historical classification information; for each vegetation type, based on the vegetation feature information of the vegetation type in different seasons, in the vegetation map of each season, based on the vegetation sample screening strategy, screening the sub-vegetation sample information of each season, and using the sub-vegetation sample information of all seasons as the vegetation sample information of the vegetation type; the vegetation sample screening strategy is a determination strategy for the range image information of the vegetation type in the land, and the determination strategy includes a Sentinel screening strategy and a Landsat screening strategy; The extraction module is used to perform interference image removal processing on each image information in the comprehensive image information to obtain vegetation image information corresponding to each image information; select the target vegetation image information in the target land supply range from each vegetation image information, and calculate the vegetation index of each target vegetation image information and the water body index of each target vegetation image information respectively; the calculation formula of the vegetation index is: In the above formula, R is the red band of the target vegetation image information, and NIR is the near infrared band of the target vegetation image information; the calculation formula of the water body index is: AWEI=4×(G-SWIR1)-(0.25×NIR+2.75×SWIR2) In the above formula, G is the green band of the target vegetation image information, SWIR1 is the shortwave infrared band 1 of the target vegetation image information, and SWIR2 is the shortwave infrared band 2 of the target vegetation image information; each of the target vegetation image information, the water body index of each of the target vegetation image information, and the vegetation index of each of the target vegetation image information are used as each of the land image information of the target land supply range; the land image information is subjected to time series fitting processing to obtain each of the target land image information; The processing module is used to classify the target land image information based on the vegetation sample information of each vegetation type through a random forest classification model to obtain the classification information of the target planting structure.

6. The device according to claim 5, characterized in that The determination module is specifically used for: Identify the location information corresponding to the target planting structure, and based on the location information, query the land use data set for the cultivated land range corresponding to the location information as the land supply range of the target planting structure; Based on the preset land range extension strategy and the cultivated land range, the non-cultivated land range of the target planting structure is screened in the land use data set, and the set of the cultivated land range and the non-cultivated land range is used as the target land supply range of the target planting structure.

7. The device according to claim 5, characterized in that The extraction module is specifically used for: Obtaining the acquisition time point of the image information corresponding to each land image information, and sorting the vegetation index of each land image information in a descending order of the time points to obtain a vegetation index sequence; The vegetation index is fitted by a harmonic fitting strategy to obtain vegetation fitting indexes corresponding to all land image information, and the vegetation fitting indexes are used to replace the vegetation indexes in each land image information to obtain each target land image information.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Citation Information

Patent Citations

  • Land cover classification method based on MODIS time series data

    CN104318270A

  • Crop classification method and device, computer equipment and storage medium

    CN108596029A

  • Planting structure classification method and device

    CN111814563A

  • Construction method of land cover classification model

    CN112597870A

  • Ecological area land classification method and device and electronic equipment

    CN113688909A