CN-122023569-A - Dynamic drawing method for staple grain crops based on time sequence remote sensing image

CN122023569ACN 122023569 ACN122023569 ACN 122023569ACN-122023569-A

Abstract

The invention provides a main grain crop dynamic drawing method based on time sequence remote sensing images, which comprises the steps of obtaining a time sequence remote sensing image group of a target area, wherein the time sequence remote sensing image group comprises a plurality of remote sensing images arranged in time sequence, inputting the time sequence remote sensing image group into a trained first prediction model, obtaining a main grain crop preliminary drawing result output by the first prediction model, wherein the main grain crop preliminary drawing result comprises a space image of the target area and a predicted value of the occupied area proportion of the main grain crop in the area corresponding to each pixel point in the space image, determining a first main grain crop proportion threshold value corresponding to the target area based on the geographic position of the target area, and obtaining a main grain crop target drawing result of the target area based on the first main grain crop proportion threshold value corresponding to the target area and the main grain crop preliminary drawing result. The invention can realize the generation of accurate local ration crop proportion graph.

Inventors

LI XUECAO
HUANG JIANXI
WEN YANAN
Chen tuo
Hu Tengyun

Assignees

中国农业大学

Dates

Publication Date: 20260512
Application Date: 20241105

Claims (10)

1. A method for dynamically drawing a staple food crop based on a time sequence remote sensing image is characterized by comprising the following steps: Acquiring a time sequence remote sensing image group of a target area, wherein the time sequence remote sensing image group comprises a plurality of remote sensing images arranged in time sequence; Inputting the time sequence remote sensing image group into a trained first prediction model, and obtaining a primary drawing result of the main grain crop output by the first prediction model, wherein the primary drawing result of the main grain crop comprises a space image of the target area and a predicted value of the occupation area proportion of the main grain crop in the area corresponding to each pixel point in the space image; Determining a first main grain crop proportion threshold corresponding to the target area based on the geographic position of the target area, and obtaining a main grain crop target drawing result of the target area based on the first main grain crop proportion threshold corresponding to the target area and the main grain crop preliminary drawing result, wherein the first main grain crop proportion threshold corresponding to the target area reflects a credible value of the main grain crop proportion of the geographic position of the target area; The first prediction model is obtained based on training of multiple groups of first training data, each group of first training data comprises a sample time sequence remote sensing image group and a primary drawing result label of a main grain crop corresponding to the sample time sequence remote sensing image group, and the primary drawing result label of the main grain crop comprises labeling values of the occupation area proportion of the main grain crop in the area corresponding to each sample pixel point.
2. The method for dynamic drawing of a staple food crop based on time-series remote sensing images according to claim 1, wherein before inputting the time-series remote sensing image group into the trained first prediction model, the method comprises: and acquiring historical farmland data of the target area, and adding a mask into the remote sensing image based on the historical farmland data, wherein the mask is used for removing a non-farmland part in the remote sensing image.
3. The method for dynamically drawing a main grain crop based on a time-series remote sensing image according to claim 1, wherein before obtaining a main grain crop target drawing result of the target area based on the first main grain crop proportion threshold value and the main grain crop preliminary drawing result corresponding to the target area, the method comprises: Determining a second main grain crop proportion threshold corresponding to the target area based on a true value of the historical main grain crop occupation area proportion of the target area, wherein the second main grain crop proportion threshold corresponding to the target area reflects the accuracy degree of the occupation area proportion of the main grain crop in the primary drawing result of the main grain crop; and screening the predicted value of the floor area ratio of the main grain crop in the main grain crop target drawing result based on the second main grain crop ratio threshold.
4. The method for dynamic drawing of a main grain crop based on a time-series remote sensing image according to claim 3, wherein the determining the second main grain crop ratio threshold corresponding to the target area based on the historical main grain crop occupation area ratio data of the target area comprises: Acquiring a historical time sequence remote sensing image group corresponding to a real value of the land occupation ratio of the historical staple food crops in the target area, inputting the historical time sequence remote sensing image group into the trained first prediction model, and acquiring a predicted value of the land occupation ratio of the historical staple food crops in the target area, which is output by the first prediction model; Determining an overlapping area based on a predicted value and a true value of the area proportion of the historical staple grain crops in the target area, wherein the overlapping area is an area where the predicted value and the true value point to the staple grain crops; randomly sampling in the overlapping area to obtain a plurality of sampling pixel points; Drawing an ROC curve based on a predicted value and a true value of the land occupation ratio of the historical staple food crops corresponding to the sampling pixel points; determining the second ration crop ratio threshold based on the ROC curve.
5. The method for dynamic drawing of a staple food crop based on time-series remote sensing images according to claim 1, wherein determining a first staple food crop ratio threshold corresponding to the target area based on the geographic position of the target area comprises: Determining a target grid of the target area within a preset geographic range based on the geographic position of the target area; acquiring the first main grain crop proportion threshold corresponding to the target grid as the first main grain crop proportion threshold corresponding to the target area; the first main grain crop proportion threshold corresponding to the grid in the preset geographic range is obtained based on the following steps: Randomly selecting a part of all grids in the preset geographic range as first grids, and the rest of grids as second grids, wherein the number of the first grids is smaller than that of the second grids; Determining the first main grain crop proportion threshold corresponding to the first grid based on a true value of the historical main grain crop occupation proportion of the first grid and a predicted value of the historical main grain crop occupation proportion of the first grid obtained based on the first prediction model; Determining Gaussian distribution data of the second grid according to the predicted value of the historical staple grain crop occupation ratio of the second grid, which is obtained based on the first prediction model, wherein the Gaussian distribution data of the second grid comprises a first Gaussian distribution parameter of the predicted value of the historical staple grain crop occupation ratio of the second grid and a second Gaussian distribution parameter of the predicted value of the historical non-staple grain crop occupation ratio of the second grid; Inputting the Gaussian distribution data of the second grid into a trained second prediction model, and acquiring the first ration crop proportion threshold value corresponding to the second grid output by the second prediction model; the second prediction model is obtained through training based on a plurality of groups of second training data, and each group of second training data comprises the Gaussian distribution data of the first grid and the first ration crop proportion threshold corresponding to the first grid.
6. The method for dynamic drawing of a main grain crop based on a time-series remote sensing image according to claim 5, wherein determining the first main grain crop proportion threshold corresponding to the first grid based on the real value of the historical main grain crop occupation ratio of the first grid and the predicted value of the historical main grain crop occupation ratio of the first grid obtained by the first prediction model comprises: Acquiring a first main grain crop occupation area corresponding to a predicted value of the historical main grain crop occupation ratio of the first grid obtained based on the first prediction model; acquiring a second staple food crop occupation area corresponding to a true value of the historical staple food crop occupation ratio based on the first grid; And taking a predicted value of the historical main grain crop occupation ratio of the first grid, which enables the occupation area of the first main grain crop to be closest to the occupation area of the second main grain crop, as the first main grain crop proportion threshold corresponding to the first grid.
7. The utility model provides a staple food crop dynamic drawing device based on time sequence remote sensing image which characterized in that includes: The remote sensing image acquisition module is used for acquiring a time sequence remote sensing image group of the target area, wherein the time sequence remote sensing image group comprises a plurality of remote sensing images arranged in time sequence; the primary drawing module is used for inputting the time sequence remote sensing image group into a trained first prediction model, and obtaining a primary drawing result of the main grain crop output by the first prediction model, wherein the primary drawing result of the main grain crop comprises a space image of the target area and a predicted value of the occupation area ratio of the main grain crop in the area corresponding to each pixel point in the space image; The target drawing module is used for determining a first main grain crop proportion threshold value corresponding to the target area based on the geographic position of the target area, and obtaining a main grain crop target drawing result of the target area based on the first main grain crop proportion threshold value corresponding to the target area and the main grain crop preliminary drawing result, wherein the first main grain crop proportion threshold value corresponding to the target area reflects a credible value of the main grain crop proportion of the geographic position of the target area; The first prediction model is obtained based on training of multiple groups of first training data, each group of first training data comprises a sample time sequence remote sensing image group and a primary drawing result label of a main grain crop corresponding to the sample time sequence remote sensing image group, and the primary drawing result label of the main grain crop comprises labeling values of the occupation area proportion of the main grain crop in the area corresponding to each sample pixel point.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method for dynamic drawing of a staple food crop based on time-series remote sensing images as defined in any one of claims 1 to 6.
9. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the method for dynamic drafting of a staple food crop based on time-series remote sensing images as claimed in any one of claims 1 to 6.
10. A computer program product comprising a computer program which when executed by a processor implements the method for dynamic drafting of a staple food crop based on time-series remote sensing images as claimed in any one of claims 1 to 6.

Description

Dynamic drawing method for staple grain crops based on time sequence remote sensing image Technical Field The invention relates to the technical field of remote sensing, in particular to a main grain crop dynamic drawing method based on time sequence remote sensing images. Background Monitoring the spatial distribution of crops is a key basis for regulating agricultural production and adjusting planting structures. The high-precision main grain crop distribution map is timely and accurately drawn, and is important for guaranteeing grain safety and sustainable development of the environment. With the continuous progress of satellite remote sensing and ground monitoring technologies, the precision of crop monitoring is continuously improved, and the remote sensing technology plays an increasingly important role in crop drawing and evaluation. In the prior art, the crop mapping based on remote sensing is often a crop classification chart, namely judging which crop is mainly planted in a large-scale area, that is, the crop classification chart usually focuses on the overall crop type distribution, and is difficult to show the situation of local planting change or symbiosis of various crops in detail, while the crop proportion chart can reflect the crop proportion in a local small-scale area, so that it is important to comprehensively know the space-time difference inside and among uniform crop pixels, further directly quantify the planting areas of different crops, and more comprehensively reflect the agricultural production structure. There is no method in the prior art for generating a local crop fraction map. Disclosure of Invention The invention provides a main grain crop dynamic drawing method based on a time sequence remote sensing image, which is used for solving the defect that a method for generating a local crop proportion graph does not exist in the prior art and realizing the generation of the local crop proportion graph. The invention provides a main grain crop dynamic drawing method based on time sequence remote sensing images, which comprises the following steps: Acquiring a time sequence remote sensing image group of a target area, wherein the time sequence remote sensing image group comprises a plurality of remote sensing images arranged in time sequence; Inputting the time sequence remote sensing image group into a trained first prediction model, and obtaining a primary drawing result of the main grain crop output by the first prediction model, wherein the primary drawing result of the main grain crop comprises a space image of the target area and a predicted value of the occupation area proportion of the main grain crop in the area corresponding to each pixel point in the space image; Determining a first main grain crop proportion threshold corresponding to the target area based on the geographic position of the target area, and obtaining a main grain crop target drawing result of the target area based on the first main grain crop proportion threshold corresponding to the target area and the main grain crop preliminary drawing result, wherein the first main grain crop proportion threshold corresponding to the target area reflects a credible value of the main grain crop proportion of the geographic position of the target area; The first prediction model is obtained based on training of multiple groups of first training data, each group of first training data comprises a sample time sequence remote sensing image group and a primary drawing result label of a main grain crop corresponding to the sample time sequence remote sensing image group, and the primary drawing result label of the main grain crop comprises labeling values of the occupation area proportion of the main grain crop in the area corresponding to each sample pixel point. According to the method for dynamically drawing the staple food crops based on the time sequence remote sensing images, which is provided by the invention, before the time sequence remote sensing image group is input into a trained first prediction model, the method comprises the following steps: and acquiring historical farmland data of the target area, and adding a mask into the remote sensing image based on the historical farmland data, wherein the mask is used for removing a non-farmland part in the remote sensing image. According to the method for dynamically drawing the main grain crop based on the time sequence remote sensing image, which is provided by the invention, the method comprises the following steps before obtaining the main grain crop target drawing result of the target area based on the first main grain crop proportion threshold value corresponding to the target area and the main grain crop preliminary drawing result: Determining a second main grain crop proportion threshold corresponding to the target area based on a true value of the historical main grain crop occupation area proportion of the target area, wherein the second main grain crop pro