CN-121982568-A - Intelligent automatic drawing system and method

CN121982568ACN 121982568 ACN121982568 ACN 121982568ACN-121982568-A

Abstract

The invention discloses an intelligent automatic drawing system and method, which relate to the technical field of agricultural remote sensing monitoring, wherein the method comprises the steps of obtaining multi-temporal remote sensing images of a farmland area, dividing the farmland into space units after pretreatment and spatial alignment, extracting time sequence spectrums and vegetation index features of the space units to form a visual feature vector sequence, further utilizing an unsupervised cluster to analyze the sequence, automatically identifying time sequence evolution track categories representing different growth states, finally, automatically labeling the space units judged to be abnormal on a map by the system and generating early warning, and outputting farmland abnormal thematic map.

Inventors

DONG XIAOCUI
ZHANG XINYUE
QIN CHUAN
Zou Lishu
SONG WAI
XU KANG
ZHAN QIANG
Deng Yongwang
MA XIN
Ma Bixuan

Assignees

长春市规划编制研究中心(长春市城乡规划设计研究院)

Dates

Publication Date: 20260505
Application Date: 20260123

Claims (10)

1. An intelligent automatic drawing method is characterized by comprising the following steps: Step S1, acquiring a remote sensing image of a farmland target area in a continuous time period, and performing preprocessing operation and space alignment processing on the remote sensing image to obtain a standardized remote sensing image; S2, dividing the farmland target area into a plurality of space units, extracting visual features of the space units in time sequence images in the continuous time period, and forming a visual feature vector sequence of the space units by time sequence aggregation; s3, analyzing a time sequence change mode of the space unit based on the visual feature vector sequence, and judging a time sequence evolution track category to which the space unit belongs; and S4, identifying the space unit judged to be the abnormal category as a farmland abnormal area based on the time sequence evolution track category and a preset threshold value, and performing visual labeling and early warning to generate a farmland abnormal map.
2. An intelligent automated drafting method according to claim 1, wherein step S1 comprises: s1-1, determining a farmland target area, acquiring remote sensing images of the farmland target area in a continuous time period, and forming a remote sensing image set , wherein, , ,..., ,..., Respectively representing remote sensing images of the 1 st, 2 nd, & gt, k th, & gt, n th time node; S1-2, sequentially carrying out radiation correction and atmosphere correction on the remote sensing images in the remote sensing image set A, and carrying out noise filtering and image cutting to obtain preprocessed remote sensing images; step S1-3, obtaining a standardized remote sensing image set through geometric registration and spatial resampling of the preprocessed remote sensing image , wherein, , ,..., ,..., Respectively, represent the 1 st, 2 nd, k 'th, n' th time node standardized telemetry image.
3. An intelligent automated drafting method according to claim 2, wherein step S2 comprises: S2-1, dividing the farmland target area into m space units to form a space unit set , wherein, , ,..., ,..., Respectively, 1 st, 2 nd, i, m-th space unit, and i-th space unit Assigning unique spatial identification ; Step S2-2, regarding the kth standardized remote sensing image in the standardized remote sensing image set B Calculate the ith space cell The average value and standard deviation of the reflectivity of all pixels in a plurality of spectrum bands are calculated, and the ith space unit is calculated The normalized vegetation index and the average value of the enhanced vegetation index of all pixels in the matrix are arranged according to a preset sequence to form a d-dimensional visual feature vector Wherein the dimension d of the visual feature vector is the total number of selected spectral features and vegetation index features; step S2-3 for the ith spatial element The visual feature vectors aggregated at all time nodes in chronological order are: , ,..., Constitute for the ith space unit Is a visual feature vector sequence of (1) , wherein, , ,..., Respectively representing the ith space unit At 1 st, 2 nd..the visual feature vector of the nth time node, and establishing a spatial identification With a sequence of visual feature vectors Is a relationship of association of the above.
4. An intelligent automated drafting method according to claim 3, wherein step S3 comprises: Step S3-1 from the ith space element Is a visual feature vector sequence of (1) In (1) extracting a sequence of visual feature vectors The method comprises the steps of forming a time sequence observation sequence of each characteristic dimension by using a least square method to fit a linear trend to the time sequence observation sequence of any characteristic dimension, recording the obtained slope as a change trend parameter of the characteristic dimension, calculating the standard deviation of the time sequence observation sequence and recording the standard deviation as a change fluctuation parameter, calculating the difference between the maximum value and the minimum value of the time sequence observation sequence and recording the difference as a change amplitude parameter, and recording the visual characteristic vector sequence The variation trend parameter, variation fluctuation parameter and variation amplitude parameter of all feature dimensions are collected to form the ith space unit Is a variable feature parameter set of (1) ; Step S3-2, collecting the variation characteristic parameters of all m space units As a sample, performing unsupervised clustering through a clustering algorithm, outputting t clusters, and marking each cluster as a time sequence evolution track category; step S3-3, according to the clustering result, the ith space unit Assigning a category identification , , wherein, , ,..., ,..., The method comprises the steps of respectively representing a1 st time-sequence evolution track category, a2 nd time-sequence evolution track category, a j th time-sequence evolution track category and a t time-sequence evolution track category, and identifying the time-sequence evolution track category to which the time-sequence evolution track category belongs, and establishing a relation between a space unit and the time-sequence evolution track category.
5. An intelligent automated drafting method according to claim 4, wherein step S4 comprises: S4-1, judging according to a preset threshold value based on the change characteristic parameters of each category in the time-sequence evolution track categories, and determining a time-sequence evolution track abnormality category; s4-2, extracting the space identification of the farmland abnormal area Based on spatial identification In a coordinate system of a standardized remote sensing image, adopting a preset abnormal mark sign to carry out visual marking on the position of the abnormal farmland area; S4-3, generating early warning information containing abnormal region positions, abnormal types and abnormal characteristic parameters according to the time sequence evolution track abnormal category; and S4-4, summarizing the spatial position information, the abnormality category information and the early warning grade information of all farmland abnormal areas, and constructing a farmland abnormal map by taking a geographic space coordinate system of a standardized remote sensing image as a reference.
6. An intelligent automatic drawing system for executing the intelligent automatic drawing method of any one of claims 1-5, which is characterized by comprising a data preprocessing module, a time sequence feature construction module, a track classification module and an abnormal map generation module; The data preprocessing module is used for acquiring remote sensing images of a farmland target area in a continuous time period, and performing preprocessing operation and space alignment processing on the remote sensing images to obtain standardized multi-temporal remote sensing images; The time sequence feature construction module is used for dividing the farmland target area into a plurality of space units, extracting the spectrum and vegetation index features of each space unit in the multi-temporal remote sensing image, and forming a visual feature vector sequence of each space unit through time sequence aggregation; The track classification module is used for judging the time sequence evolution track category of each space unit by analyzing the time sequence change mode of each space unit through unsupervised clustering based on the visual feature vector sequence; the abnormal map generation module is used for identifying the space unit which is judged to be the abnormal time sequence evolution track type as a farmland abnormal area, and carrying out visual marking and early warning based on the geographic space position of the space unit to generate a farmland abnormal map.
7. An intelligent automated drafting system according to claim 6, wherein the data preprocessing module comprises: the data acquisition unit is used for determining a farmland target area and acquiring a plurality of remote sensing images of the farmland target area in a continuous time period; the correction unit is used for sequentially carrying out radiation correction and atmosphere correction on the acquired remote sensing image, and carrying out noise filtering and image cutting; And the registration unit is used for carrying out geometric registration and space resampling on the corrected and cut remote sensing image to form a standardized remote sensing image sequence with consistent space-time.
8. An intelligent automated drafting system according to claim 7, wherein the timing characteristics building module comprises: the space dividing unit is used for dividing the farmland target area into a plurality of space units and distributing unique geographic identifiers for each space unit; the characteristic calculation unit is used for calculating the statistical value of each spectral band of the pixels in each space unit and the statistical value of at least one vegetation index according to the remote sensing images in the standardized image sequence, and generating the characteristic vector of the space unit in the time phase; and the sequence aggregation unit is used for aggregating the characteristic vectors of each space unit in all time phases in time sequence to form a visual characteristic vector sequence of the space unit and establishing the association of the visual characteristic vector sequence with the geographic identification.
9. An intelligent automated cartography system according to claim 8, wherein the trajectory classification module comprises: the parameter extraction unit is used for extracting a meta-feature parameter set capable of representing a time sequence change mode from the visual feature vector sequence of each space unit, wherein the meta-feature parameter comprises a change trend, change fluctuation and change amplitude; The clustering analysis unit is used for analyzing the meta-feature parameter sets of all the space units by using a clustering algorithm and classifying the space units with similar time sequence change modes into the same time sequence evolution track category; The category allocation unit is used for allocating a time sequence evolution track category identifier for each space unit and establishing a mapping relation between the space units and the category.
10. An intelligent automated cartography system according to claim 9, wherein the anomaly atlas creation module comprises: The anomaly identification unit is used for determining an anomaly category according to a preset threshold value based on the change characteristic parameters of each category in the time sequence evolution track category and screening out a space unit belonging to the anomaly category; the space labeling unit is used for determining the position of the abnormal unit according to the geographic identification of the abnormal unit under the unified space reference system corresponding to the standardized remote sensing image, and performing visual labeling by using a preset abnormal labeling symbol; The early warning generation unit is used for generating early warning information comprising the position of the abnormal region, the abnormal type, the abnormal characteristic parameter and the early warning level according to the abnormal type and the variation characteristic parameter of the abnormal region of the farmland; and the map construction unit is used for integrating the spatial positions, the categories and the early warning information of all abnormal areas to generate a comprehensive farmland abnormal thematic map.

Description

Intelligent automatic drawing system and method Technical Field The invention relates to the technical field of agricultural remote sensing monitoring, in particular to an intelligent automatic drawing system and method. Background The remote sensing technology is used for farmland monitoring, and becomes an important means for modern accurate agricultural management. At present, a common farmland remote sensing monitoring method in practice mainly depends on analysis of single-phase remote sensing images. By calculating the vegetation index and setting the threshold value, it is possible to preliminarily identify an area exhibiting abnormality at a specific point in time, for example, a field with poor growth. Although the method is simple to implement, the method has obvious limitation that the method can only reflect the representation of crops at a certain moment and can not capture and quantify the dynamic change process of the crops in the whole growth period. Therefore, it is difficult to effectively distinguish between temporary environmental interference and persistent growth stress, and early warning of abnormal patterns with typical timing characteristics such as growth retardation, premature senility, etc. is also impossible. In order to obtain more comprehensive information, professionals also perform contrast analysis on images of multiple phases. However, this approach typically relies on manual visual interpretation and expert experience, requiring a one-by-one comparison and interpretation of the multi-phase images. The process is time-consuming, labor-consuming, low in efficiency, difficult to meet the requirements of large-range and high-frequency business monitoring, and the analysis result is greatly influenced by personal subjective factors, lacks consistent objective standards and is not strong in repeatability and comparability. Therefore, how to automatically and intelligently analyze time sequence remote sensing data of farmlands in the whole growth period, objectively and efficiently identify abnormal areas with different evolution rules and generate visual maps is a technical problem to be solved for improving the capability of agricultural remote sensing monitoring business. Disclosure of Invention The invention aims to provide an intelligent automatic drawing system and method for solving the problems in the prior art. In order to achieve the above purpose, the invention provides the following technical proposal that an intelligent automatic drawing system and method, The method comprises the following steps: Step S1, acquiring a remote sensing image of a farmland target area in a continuous time period, and performing preprocessing operation and space alignment processing on the remote sensing image to obtain a standardized remote sensing image; through data standardization processing, a unified data base is established for subsequent time sequence analysis; S2, dividing the farmland target area into a plurality of space units, extracting visual features of the space units in time sequence images in the continuous time period, and forming a visual feature vector sequence of the space units by time sequence aggregation; constructing a characteristic sequence for describing the time sequence change rule of each space unit; s3, analyzing a time sequence change mode of the space unit based on the visual feature vector sequence, and judging a time sequence evolution track category to which the space unit belongs; performing pattern classification on the space units according to the time sequence change characteristics; s4, identifying the space unit judged to be the abnormal category as a farmland abnormal area based on the time sequence evolution track category and a preset threshold value, and performing visual labeling and early warning to generate a farmland abnormal map; and converting the classification result into an anomaly monitoring map with spatial location information. Further, step S1 includes: s1-1, determining a farmland target area, acquiring remote sensing images of the farmland target area in a continuous time period, and forming a remote sensing image set , wherein,,,...,,...,Respectively representing remote sensing images of the 1 st, 2 nd, & gt, k th, & gt, n th time node; The remote sensing image data acquisition covering a plurality of time nodes is completed; S1-2, sequentially carrying out radiation correction and atmosphere correction on the remote sensing images in the remote sensing image set A, and carrying out noise filtering and image cutting to obtain preprocessed remote sensing images; The influence of the sensor, the atmosphere and the noise on the image data is reduced; step S1-3, obtaining a standardized remote sensing image set through geometric registration and spatial resampling of the preprocessed remote sensing image , wherein,,,...,,...,Respectively, 1 st, 2 nd, k 'th, n' th time node standardized remote sensing image; the spatial position