CN-121686271-B - Cross-region crop remote sensing image processing method, device and equipment integrating geographic information

CN121686271BCN 121686271 BCN121686271 BCN 121686271BCN-121686271-B

Abstract

The application discloses a method, a device and equipment for processing cross-regional crop remote sensing images by fusing geographic information, which relate to the technical field of crop data processing technology and remote sensing intersection, and are characterized in that a generalized data set of latitude and topography joint partitions is constructed through sample geographic information data and sample satellite remote sensing image data to obtain a geographic prototype library corresponding to geographic units, then a crop classification prediction model is trained based on the geographic prototype library, and crop classification prediction is carried out by using the crop classification prediction model, so that the composite influence of geographic multidimensional factors on crop growth can be effectively captured, the fundamental problem of poor generalization of the cross-regional model is solved, the cross-geographic processing capacity of the crop classification prediction model is improved, and the crop remote sensing classification with high precision, high robustness and low cost is realized.

Inventors

Luo Zifei
Zhao Lingyuan
JIANG YUHAN
WANG QUN
CHEN ZHANGJIE

Assignees

环天智慧科技股份有限公司

Dates

Publication Date: 20260508
Application Date: 20260210

Claims (9)

1. A method for processing a cross-region crop remote sensing image fusing geographic information is characterized by comprising the following steps: Acquiring sample geographic information data and sample satellite remote sensing image data corresponding to a first target area, and preprocessing the sample geographic information data and the sample satellite remote sensing image data to obtain preprocessed data, wherein the first target area represents a region where data needs to be acquired; Dividing geographic units and binding data according to the preprocessing data to obtain a plurality of geographic units and crop sample data corresponding to each geographic unit, and establishing a geographic prototype library corresponding to the geographic units based on the crop sample data corresponding to the geographic units; Determining a target geographic unit corresponding to a second target area and a target geographic prototype library corresponding to the target geographic unit in response to a real-time crop prediction instruction input by a user through man-machine interaction, wherein the second target area represents a subarea determined in the first target area according to the real-time crop prediction instruction; Training a crop classification prediction model according to the target geographic unit and the target geographic prototype library corresponding to the target geographic unit, collecting real-time geographic information data and real-time satellite remote sensing image data corresponding to the second target area, and identifying the real-time geographic information data and the real-time satellite remote sensing image data by adopting the crop classification prediction model to obtain a crop classification prediction result; after determining the target geographic unit corresponding to the second target area and the target geographic prototype library corresponding to the second target area, the method further comprises the following steps: determining the sample number of crop sample data in a target geographic prototype library, and determining sample sufficiency according to the sample number, wherein the sample sufficiency comprises sample sufficiency or sample insufficiency; If the sample sufficiency is that the operation is not performed, and a process of training a crop classification prediction model according to the target geographic unit and the corresponding target geographic prototype library is entered; Under the condition that the sample sufficiency is insufficient, matching similar geographic prototype libraries in other geographic prototype libraries corresponding to the first target area and/or geographic prototype libraries corresponding to other areas based on geographic features corresponding to the target geographic prototype library; and supplementing the target geographic prototype library according to the similar geographic prototype library to obtain a supplemented target geographic prototype library, and entering a process of training a crop classification prediction model according to the target geographic unit and the target geographic prototype library corresponding to the target geographic unit based on the supplemented target geographic prototype library.
2. The method for processing a cross-regional crop remote sensing image fused with geographic information according to claim 1, wherein obtaining sample geographic information data and sample satellite remote sensing image data corresponding to a first target region comprises: Acquiring a multispectral satellite image corresponding to the second target area to obtain sample satellite remote sensing image data; And obtaining DEM data corresponding to the second target area to obtain sample geographic information data.
3. The method for processing the cross-regional crop remote sensing image fused with the geographic information according to claim 2, wherein preprocessing the sample geographic information data and the sample satellite remote sensing image data to obtain preprocessed data comprises the following steps: Performing image preprocessing on the sample satellite remote sensing image data to obtain preprocessed sample satellite remote sensing image data, wherein the image preprocessing comprises radiation calibration, orthographic correction, image registration and/or image fusion; Extracting elevation data corresponding to a first target area from sample geographic information data, and acquiring gradient data corresponding to the first target area according to the elevation data; carrying out partition calculation by adopting a geographic element file divided by a cultivated land boundary according to the elevation data and the gradient data corresponding to the first target area to obtain the elevation data and the gradient data corresponding to each crop sample area in the first target area; Writing the altitude data and the gradient data into an attribute table of a crop sample area in the geographic element file to obtain a geographic element file written with attributes of the crop sample area, and obtaining a first target geographic element file; dividing the preprocessed sample satellite remote sensing image data based on the smallest external square of each crop sample area to obtain a slice multispectral image corresponding to each crop sample area; And taking the first target geographic element file and the slice multispectral image corresponding to each crop sample area as preprocessing data.
4. The method for processing a cross-regional crop remote sensing image fused with geographic information according to claim 3, wherein the steps of dividing geographic units and binding data according to the preprocessing data to obtain a plurality of geographic units and crop sample data corresponding to each geographic unit comprise: extracting a latitude range corresponding to the first target area based on a space coordinate system corresponding to the preprocessed sample satellite remote sensing image data; According to the latitude range corresponding to the first target area, carrying out transverse interval division on the first target area based on unit latitude to divide the first target area into a plurality of latitude zones so as to obtain a two-dimensional network, wherein the unit latitude is set to be 1 degree; longitudinally dividing the first target area based on a unit elevation according to an elevation range corresponding to the first target area to divide the first target area into a plurality of independent geographic units, wherein the unit elevation is set to be 200m; Writing a latitude range and an altitude range corresponding to the geographic unit into a first target geographic element file according to the coordinates corresponding to the geographic unit to obtain a second target geographic element file, wherein the first target geographic element file is the same as a space coordinate system used by the preprocessed sample satellite remote sensing image data; Determining target crop sample areas in a latitude range and an elevation range corresponding to the geographic units according to a second target geographic element file, and constructing slice multispectral images, latitude data, elevation data, gradient data, data sampling time and corresponding crop label data corresponding to the target crop sample areas into crop sample data together to obtain crop sample data corresponding to the geographic units, wherein the crop label data is a real crop label input by man-machine interaction; and traversing all the geographic units to obtain crop sample data corresponding to each geographic unit.
5. The method for processing the cross-regional crop remote sensing image fused with the geographic information as claimed in claim 4, wherein after obtaining the plurality of geographic units and the crop sample data corresponding to each geographic unit, the method further comprises the steps of performing time consistency check on the crop sample data, determining crop sample data failing to check, and marking the crop sample data failing to check.
6. The method for processing a cross-regional crop remote sensing image fused with geographic information according to claim 4, wherein training a crop classification prediction model according to the target geographic unit and the target geographic prototype library corresponding to the target geographic unit comprises: Respectively extracting latitude data, altitude data, gradient data and latitude vectors, altitude vectors, gradient vectors and time vectors corresponding to data sampling time in target crop sample data according to target crop sample data in a target geographic prototype library corresponding to the target geographic unit; Splicing the slice multispectral image in the target crop sample data with a latitude vector, an altitude vector, a gradient vector and a time vector to obtain sample fusion characteristic data; and training a crop classification prediction model by taking the sample fusion characteristic data as actual input and taking crop label data in the target crop sample data as expected output.
7. The method for processing the cross-regional crop remote sensing image fused with the geographic information according to claim 6, wherein the steps of collecting the real-time geographic information data and the real-time satellite remote sensing image data corresponding to the second target region, identifying the real-time geographic information data and the real-time satellite remote sensing image data by using a crop classification prediction model, and obtaining a crop classification prediction result comprise the following steps: Collecting real-time geographic information data corresponding to the second target area and real-time satellite remote sensing image data; determining a real-time slice multispectral image, real-time latitude data, real-time elevation data, real-time gradient data and real-time data sampling time corresponding to the second target area according to the real-time geographic information data and the real-time satellite remote sensing image data; Acquiring real-time fusion characteristic data according to the real-time slice multispectral image, the real-time latitude data, the real-time elevation data, the real-time gradient data and the real-time data sampling time; And identifying the real-time fusion characteristic data by adopting the crop classification prediction model to obtain a crop classification prediction result.
8. A geographical information-fused cross-region crop remote sensing image processing apparatus for performing the geographical information-fused cross-region crop remote sensing image processing method according to any one of claims 1 to 7, characterized by comprising: the system comprises a sample data preprocessing module, a data acquisition module and a data acquisition module, wherein the sample data preprocessing module is used for acquiring sample geographic information data and sample satellite remote sensing image data corresponding to a first target area and preprocessing the sample geographic information data and the sample satellite remote sensing image data to obtain preprocessed data; The prototype library construction module is used for dividing geographic units and binding data according to the preprocessing data to obtain a plurality of geographic units and crop sample data corresponding to each geographic unit, and establishing a geographic prototype library corresponding to the geographic units based on the crop sample data corresponding to the geographic units; The prediction instruction response module is used for responding to a real-time crop prediction instruction input by a user through man-machine interaction, and determining a target geographic unit corresponding to a second target area and a target geographic prototype library corresponding to the second target area, wherein the second target area represents an area determined in the first target area according to the real-time crop prediction instruction; And the crop classification prediction module is used for training a crop classification prediction model according to the target geographic unit and the target geographic prototype library corresponding to the target geographic unit, collecting real-time geographic information data and real-time satellite remote sensing image data corresponding to the second target area, and identifying the real-time geographic information data and the real-time satellite remote sensing image data by adopting the crop classification prediction model to obtain a crop classification prediction result.
9. An electronic device comprising a memory for storing a set of computer instructions and a processor which, when executing the set of computer instructions, performs the steps of the method of any of claims 1 to 7.

Description

Cross-region crop remote sensing image processing method, device and equipment integrating geographic information Technical Field The application relates to the technical field of crop data processing technology and remote sensing intersection, in particular to a method, a device and equipment for processing cross-region crop remote sensing images by fusing geographic information. Background With the increase of global food safety demands, remote sensing technology has become a core means of crop classification and monitoring. The remote sensing data provides earth observation spanning multiple dimensions, including critical spatial and temporal information, which is critical to coping with macroscopic monitoring tasks, and enables accurate identification of large-scale crop types in combination with artificial intelligence. In macro-monitoring tasks, geographical information is of paramount importance, such as crop classification tasks. The crop growth characteristics and the planting structure are obviously influenced by geographical environments (such as latitude, altitude and topography), the crop spectrum characteristics and microclimate caused by illumination, accumulation temperature difference, altitude and gradient in the climatic period are jointly influenced, and the generalization performance of a model trained in a single area is drastically reduced when the model is applied across areas. The prior art does not sufficiently quantify the coupling effect of the geographic factors, resulting in difficulty in adapting the model to complex geographic environments. And secondly, the field sampling cost in high-altitude remote areas is high, and the cross-region image data distribution difference is obvious, so that the crop information processing difficulty based on the image is increased. The prior art generally relies on remote sensing data of a single geographical area to train a crop classification model, relies on local sample labeling, does not consider the influence of latitude differences on crop growth characteristics, and the model cannot be suitable for other areas. Disclosure of Invention The application aims to provide a method, a device and equipment for processing cross-regional crop remote sensing images fused with geographic information, which solve the problems that the prior art generally relies on remote sensing data of a single geographic region to train a crop classification model, relies on local sample labeling, does not consider the influence of latitude difference on crop growth characteristics, and the model cannot be applied to other regions. The application is realized by the following technical scheme: the first aspect of the application provides a cross-region crop remote sensing image processing method fusing geographic information, which comprises the following steps: Acquiring sample geographic information data and sample satellite remote sensing image data corresponding to a first target area, and preprocessing the sample geographic information data and the sample satellite remote sensing image data to obtain preprocessed data, wherein the first target area represents a region where data needs to be acquired; Dividing geographic units and binding data according to the preprocessing data to obtain a plurality of geographic units and crop sample data corresponding to each geographic unit, and establishing a geographic prototype library corresponding to the geographic units based on the crop sample data corresponding to the geographic units; Determining a target geographic unit corresponding to a second target area and a target geographic prototype library corresponding to the target geographic unit in response to a real-time crop prediction instruction input by a user through man-machine interaction, wherein the second target area represents a subarea determined in the first target area according to the real-time crop prediction instruction; and training a crop classification prediction model according to the target geographic unit and the target geographic prototype library corresponding to the target geographic unit, collecting real-time geographic information data and real-time satellite remote sensing image data corresponding to the second target area, and identifying the real-time geographic information data and the real-time satellite remote sensing image data by adopting the crop classification prediction model to obtain a crop classification prediction result. In one possible implementation manner, after determining the target geographic unit corresponding to the second target area and the target geographic prototype library corresponding to the second target area, the method further includes: determining the sample number of crop sample data in a target geographic prototype library, and determining sample sufficiency according to the sample number, wherein the sample sufficiency comprises sample sufficiency or sample insufficiency; If the sample sufficiency i