CN-121982543-A - Cotton planting area early identification method and system based on annual remote sensing data

CN121982543ACN 121982543 ACN121982543 ACN 121982543ACN-121982543-A

Abstract

The invention discloses a cotton planting area early identification method and system based on annual remote sensing data, wherein the method comprises the steps of constructing an annual remote sensing time sequence data set; dividing a research area into a previous year cotton planting area and a previous year non-cotton planting area according to the time sequence characteristics of the cotton growth period extracted from the remote sensing time sequence data of the previous year, obtaining current year reconstruction time sequence data according to the growth period matching relation between the remote sensing time sequence data of the current year and the complete growth period remote sensing time sequence data of the previous year, determining a current year cotton planting area within the range of the previous year cotton planting area according to the current year reconstruction time sequence data and the time sequence characteristics of the previous year cotton growth period, identifying the current year cotton planting area within the range of the previous year non-cotton planting area according to the time sequence characteristics of the cotton extracted from the current year cotton planting area, and determining and outputting a current year complete cotton planting area distribution result according to the current year cotton planting area.

Inventors

ZHENG JIANGHUA
FAN HONG

Assignees

新疆大学

Dates

Publication Date: 20260505
Application Date: 20260130

Claims (8)

1. A method and a system for early identification of a cotton planting area based on annual remote sensing data are characterized by comprising the following steps: Constructing a trans-year remote sensing time sequence data set according to remote sensing time sequence data of the cotton covered by the previous year of the research area in the complete growth period and remote sensing time sequence data of the cotton covered by the current year in the early growth period; Dividing the research area into a previous year cotton planting area and a previous year non-cotton planting area according to the cotton growth period time sequence characteristics extracted from the remote sensing time sequence data of the previous year; According to the growth cycle matching relation between the remote sensing time sequence data in the early current year and the remote sensing time sequence data in the complete growth period of the previous year, performing time sequence reconstruction on the remote sensing time sequence data in the early current year to obtain reconstruction time sequence data in the current year; Determining a current year cotton planting area in the previous year cotton planting area according to the similarity analysis result of the current year reconstruction time sequence data and the cotton growth period time sequence characteristics in the previous year cotton planting area; identifying a current year cotton planting area within the previous year non-cotton planting area according to the extracted early cotton timing characteristics in the current year cotton planting area; And determining and outputting the complete cotton planting area distribution result in the current year according to the current-year cotton planting area in the previous-year cotton planting area and the current-year cotton planting area in the previous-year non-cotton planting area.
2. The method for early identification of cotton planting areas based on annual remote sensing data according to claim 1, wherein prior to constructing the annual remote sensing time series data set, the previous annual remote sensing time series data and the current annual early remote sensing time series data are subjected to unified preprocessing, wherein the unified preprocessing comprises at least one of radiation correction, atmospheric correction and geometric correction so as to eliminate the influence of different sensors, imaging conditions and geometric position differences.
3. The method of claim 1, wherein dividing the study area into a previous year cotton planting area and a previous year non-cotton planting area based on the cotton growth period time sequence features extracted from the previous year remote sensing time sequence data comprises: calculating normalized vegetation index time sequence data according to the previous year remote sensing time sequence data; constructing statistical model parameters of a cotton time sequence curve template according to time sequence feature vectors of the normalized vegetation index time sequence data at preset cotton sample point positions; and determining the previous year cotton planting area and the previous year non-cotton planting area according to the mahalanobis distance between the normalized vegetation index time sequence feature vector of each pixel in the cultivated land and the cotton time sequence curve template.
4. The method for early recognition of cotton planting areas based on annual remote sensing data according to claim 1, wherein the performing of time series reconstruction on the annual remote sensing time series data according to a growth cycle matching relationship between the annual remote sensing time series data and the whole growth period remote sensing time series data of the previous year, and the obtaining of the annual reconstructed time series data comprises: According to the alignment analysis of the change trend of the normalized vegetation index time sequence curve corresponding to the current year early remote sensing time sequence data and the normalized vegetation index time sequence curve of the previous year complete growth period in the growth initial stage, establishing a corresponding relation between the current year early time node and the previous year growth period time node; And performing time scale adjustment or interpolation processing on the current year early remote sensing time sequence data according to the corresponding relation, so that the reconstructed current year time sequence data is consistent with the growth period time sequence characteristics of the cotton of the previous year in the dimension of the breeding stage.
5. The method of claim 4, wherein determining the current year cotton planting area within the previous year cotton planting area comprises: Calculating the mahalanobis distance between the characteristic vector of the current year reconstruction time sequence data and the cotton growth period time sequence characteristic template; And judging the cotton planting area in the current year and the non-cotton area in the current year in the range of the cotton planting area in the previous year according to the comparison result of the mahalanobis distance and the preset threshold value.
6. The method of claim 5, wherein identifying the current year cotton planting area within the previous year non-cotton planting area based on the extracted early timing characteristics of cotton in the current year cotton planting area comprises: Sampling and extracting sample points from the current-year cotton planting area identified in the range of the previous-year cotton planting area, and constructing a current-year cotton early-time sequence feature template; Calculating the similarity between the current-year early-stage timing sequence characteristic of each pixel in the range of the previous-year non-cotton planting area and the current-year cotton early-stage timing sequence characteristic template; and identifying a cotton planting area newly increased in the current year within the range of the non-cotton planting area in the previous year, namely the cotton planting area in the current year, according to the comparison result of the similarity and the preset threshold value.
7. The method for early identifying a cotton planting area based on annual remote sensing data according to claim 6, wherein the identifying method adopts a dual-stage identifying structure, wherein the first stage identifies the cotton of the current year in the cotton planting area of the previous year based on annual time sequence feature similarity, the second stage additionally identifies the cotton of the current year in the cotton planting area of the previous year based on annual time sequence feature, and a union of two stage identifying results forms the cotton planting area of the current year.
8. An early cotton planting area identification system based on annual remote sensing data is used for implementing the identification method according to any one of claims 1-7, and is characterized by comprising a data acquisition module, a time sequence data construction module, a cotton partitioning module, a growth period matching module, an identification module and a result output module; The data acquisition module is used for researching remote sensing time sequence data of the cotton covered in the previous year in the whole growth period of the area and remote sensing time sequence data of the cotton covered in the early growth period of the current year; The time sequence data construction module is used for constructing a cross-year remote sensing time sequence data set according to remote sensing time sequence data of the previous year of coverage cotton in the whole growth period of the research area and remote sensing time sequence data of the current year of coverage cotton in the early growth stage; The cotton partitioning module is used for partitioning the research area into a previous year cotton planting area and a previous year non-cotton planting area according to the cotton growth period time sequence characteristics extracted from the remote sensing time sequence data of the previous year; The growth period matching module is used for carrying out time sequence reconstruction on the remote sensing time sequence data in the current year according to a growth period matching relation between the remote sensing time sequence data in the current year and the complete growth period remote sensing time sequence data in the previous year, so as to obtain reconstruction time sequence data in the current year; The identification module is used for determining a current year cotton planting area in a previous year cotton planting area according to the similarity analysis result of the current year reconstruction time sequence data and the cotton growth period time sequence characteristics in the previous year cotton planting area range; and the result output module is used for determining and outputting the current-year complete cotton planting area distribution result according to the current-year cotton planting area in the previous-year cotton planting area range and the current-year cotton planting area in the previous-year non-cotton planting area range.

Description

Cotton planting area early identification method and system based on annual remote sensing data Technical Field The invention belongs to the technical field of agricultural remote sensing and crop identification, and particularly relates to a method and a system for early identification of a cotton planting area based on annual remote sensing data. Background The cotton is used as an important cash crop, and the accurate acquisition of a planting area of the cotton has important significance for agricultural production management, resource allocation and policy decision. Existing cotton remote sensing identification technology is mostly focused on the middle and later stages of cotton growth, especially the harvest stage. In this stage, the cotton canopy structure and spectral characteristics are more remarkable, and the identification by using the remote sensing image is relatively easy, so that the cotton canopy structure and spectral characteristics are widely used in practical application. However, the spatial distribution condition of cotton in the early growth stage is difficult to reflect in time based on the identification mode of the harvest stage or the later growth stage, and the actual requirement of agricultural management such as precise irrigation, fertilization regulation and control, pest control and the like on early acquisition of cotton planting information cannot be met. In the early stage of cotton growth, the vegetation coverage degree and spectral characteristics of the cotton are not stable, and the difference between the vegetation coverage degree and other crops or bare land areas is not obvious, so that the problems of low recognition precision, high misjudgment rate and the like in the existing method generally exist in early recognition. With the dynamic adjustment of cotton planting structures, cotton planting areas have certain changes among different years, and crop rotation or planting structure adjustment of partial plots can occur. The existing partial early identification method often relies on remote sensing images with limited time of the current year to analyze, and cannot fully utilize the spatial distribution information of cotton planting areas in the previous year, and lacks a distinguishing identification strategy for cotton and non-cotton areas in the historical year, so that cotton planting change conditions in different spatial areas in the current year are difficult to effectively distinguish, and the integrity of identification results is affected. In addition, the sowing and growing processes of cotton are affected by factors such as climate conditions, regional management modes and the like, and the growth period among different years has a certain difference. If the remote sensing images of different years are directly compared and analyzed, errors are easily introduced due to asynchronous growth periods, and the technical difficulty of early identification of cotton in trans-annual application is further increased. Therefore, in the prior art, a technical scheme capable of comprehensively utilizing the remote sensing time sequence information of the cotton in the growth early stage of the previous year, combining the remote sensing time sequence data in the early year and carrying out recognition analysis aiming at different historical space partition differentiation is not available, so that the cotton planting area is accurately recognized while the different growth stages of the cotton are effectively processed, and further research and improvement are needed. Disclosure of Invention In order to solve the technical problems, the invention provides a method and a system for early recognition of cotton planting areas based on annual remote sensing data, which are used for solving the problems that the distribution of cotton planting areas is difficult to stably and accurately acquire at the early stage of cotton growth in the existing cotton remote sensing recognition technology and the recognition result is unstable due to the differences of sowing time and fertility progress of cotton of different years. In order to achieve the above purpose, the invention provides a cotton planting area early identification method based on annual remote sensing data, which comprises the following steps: Constructing a trans-year remote sensing time sequence data set according to remote sensing time sequence data of the cotton covered by the previous year of the research area in the complete growth period and remote sensing time sequence data of the cotton covered by the current year in the early growth period; Dividing the research area into a previous year cotton planting area and a previous year non-cotton planting area according to the cotton growth period time sequence characteristics extracted from the remote sensing time sequence data of the previous year; According to the growth cycle matching relation between the remote sensing time sequence data in the early