WO2022057319A1 - 基于云平台的耦合主被动遥感影像的大蒜作物识别方法 - Google Patents

基于云平台的耦合主被动遥感影像的大蒜作物识别方法 Download PDF

Info

Publication number
WO2022057319A1
WO2022057319A1 PCT/CN2021/097839 CN2021097839W WO2022057319A1 WO 2022057319 A1 WO2022057319 A1 WO 2022057319A1 CN 2021097839 W CN2021097839 W CN 2021097839W WO 2022057319 A1 WO2022057319 A1 WO 2022057319A1
Authority
WO
WIPO (PCT)
Prior art keywords
garlic
ndvi
image
pixel
crop
Prior art date
Application number
PCT/CN2021/097839
Other languages
English (en)
French (fr)
Inventor
田海峰
秦耀辰
沈威
周伯燕
王永久
Original Assignee
河南大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 河南大学 filed Critical 河南大学
Priority to US17/531,826 priority Critical patent/US12094198B2/en
Publication of WO2022057319A1 publication Critical patent/WO2022057319A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G06V20/13—Satellite images
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
    • G01S13/90—Radar or analogous systems specially adapted for specific applications for mapping or imaging using synthetic aperture techniques, e.g. synthetic aperture radar [SAR] techniques
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
    • G01S13/90—Radar or analogous systems specially adapted for specific applications for mapping or imaging using synthetic aperture techniques, e.g. synthetic aperture radar [SAR] techniques
    • G01S13/9094—Theoretical aspects
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G01S7/418—Theoretical aspects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/24323—Tree-organised classifiers

Definitions

  • the invention relates to the technical field of remote sensing target recognition, in particular to a garlic crop recognition method based on a cloud platform coupling active and passive remote sensing images.
  • Garlic crops and winter wheat crops are both winter crops and have similar growth periods and spectral characteristics. It is difficult to accurately distinguish garlic crops from winter wheat crops solely on optical satellite images. Compared with the sown area of winter wheat crops, the sown area of garlic crops is less, and the garlic plots are mostly staggered with the winter wheat plots, which further restricts the accuracy of remote sensing identification of garlic crops. How to accurately distinguish garlic crops from winter wheat crops through satellite images is one of the difficulties in the field of crop remote sensing identification.
  • the processing process of remote sensing big data is complex, and the amount of data calculation is huge, far beyond the data processing capability of personal computers.
  • the Google Earth Engine cloud computing platform integrates global public remote sensing data and provides personalized data processing services. Users only need to program and run the relevant data processing algorithms in the Google Earth Engine language, and Google Earth Engine will automatically call in the background. Google's tens of thousands of servers quickly complete data processing tasks in parallel computing, and feed back the results to users. Google Earth Engine solves the difficult problem of remote sensing big data processing, and provides an opportunity and platform for the automatic and business operation of crop remote sensing identification.
  • the present invention proposes a garlic crop identification method based on a cloud platform coupled with active and passive remote sensing images, which solves the technical problem that garlic crops cannot be accurately identified in the existing remote sensing identification technology.
  • a garlic crop identification method based on a cloud platform coupled with active and passive remote sensing images the steps are as follows:
  • the Sentinel-2 time-series image and Landsat-8 time-series satellite image of the main garlic crop producing area of the target year are retrieved, and combined with the phenological information of the garlic crop, the optical synthetic image dataset is obtained;
  • S3. Use handheld GPS to obtain and record the geographic coordinate information of garlic crops and winter wheat crops in the main garlic crop producing areas;
  • step S4 based on the optical synthetic image data set obtained in step S2 and the geographic coordinate information of the garlic crop obtained in step S3, build a garlic crop optical image recognition decision tree model;
  • step S7 obtain a radar synthetic image dataset according to the image features of garlic crops and winter wheat crops in step S6;
  • step S9 classify the radar synthetic image data set obtained in step S7 according to the garlic crop radar image recognition decision tree model obtained in step S8, and obtain a garlic crop radar distribution map;
  • the method for obtaining the optical synthetic image data set is as follows: in the period when the time series MODIS-NDVI of garlic crops is higher than that of other forest and grass vegetation, extract the NDVI of the Sentinel-2 time series image at the pixel i position. The maximum value is taken as the pixel value of pixel i, and all pixel positions of Sentinel-2 time series images are traversed in turn to obtain a composite image with the maximum NDVI value, which is recorded as NDVI max ; in the time series of garlic crops, MODIS-NDVI is smaller than that of other forest and grass vegetation.
  • a composite image with the minimum NDVI value is obtained, which is recorded as NDVI min ; in the period when the time series MODIS-NDVI of garlic crops is smaller than that of other forest and grass vegetation, Sentinel-2 and The median value of the NDVI of the Landsat-8 time series satellite image is used as the pixel value of the pixel i, and all pixel positions of the Sentinel-2 and Landsat-8 time series satellite images are traversed in turn to obtain the NDVI median value composite image, which is recorded as NDVI med ;
  • the composite image of NDVI maximum value NDVI max , the composite image of NDVI minimum value NDVI min and the composite image of NDVI median value NDVI med are combined into an optical composite image dataset.
  • the construction method of the garlic crop optical image recognition decision tree model is: obtaining the pixel median threshold ⁇ according to the distribution of the NDVI median synthetic image NDVI med on the geographic coordinate position of the garlic crop; synthesizing the image NDVI max according to the NDVI maximum value in the The distribution on the geographic coordinate position of garlic crops obtains the first threshold ⁇ of the maximum pixel value and the second threshold value ⁇ of the maximum pixel value; according to the distribution of the NDVI min synthetic image NDVI min on the geographic coordinate position of the garlic crop, the minimum pixel value is obtained.
  • the method for obtaining the optical distribution map of the garlic crop is: constructing the first constraint condition of the garlic crop according to the optical image recognition decision tree model of the garlic crop: According to the first constraint condition of garlic crop, the pixels in the synthetic image of NDVI maximum value NDVI max , the synthetic image of NDVI median value NDVI med and the synthetic image of NDVI minimum value NDVI min are respectively screened, and the optical distribution map of garlic crop is obtained.
  • NDVI max,i represents the pixel value of the ith pixel in the NDVI max composite image NDVI max
  • NDVI med,i represents the pixel value of the ith pixel in the NDVI median composite image NDVI med
  • NDVI min,i represents The pixel value of the ith pixel in the NDVI min composite image NDVI min .
  • the method for obtaining the radar synthetic image dataset is as follows: synthesizing the median image of the Sentinel-1 synthetic aperture radar satellite image during the winter wheat overwintering period, denoted as the SVV1 med synthetic image; synthesizing the Sentinel-1 synthetic image during the winter wheat tillering to booting period.
  • the median image of the aperture radar satellite image is recorded as the SVV2 med synthetic image; the SVV1 med synthetic image and the SVV2 med synthetic image are combined into a radar synthetic image dataset.
  • the construction method of the garlic crop radar image recognition decision tree model is as follows: obtaining a difference threshold ⁇ and a threshold ⁇ according to the distribution of the SVV1 med synthetic image and the SVV2 med synthetic image on the geographic coordinate position of the garlic crop; Threshold ⁇ obtains a decision tree model for garlic crop radar image recognition.
  • the method for obtaining the garlic crop radar distribution map is: constructing the second constraint condition of the garlic crop according to the garlic crop radar image recognition decision tree model, The pixels in the SVV1 med synthetic image and the SVV2 med synthetic image are screened according to the second constraint of garlic crops, and the radar distribution map of garlic crops is obtained, where SVV1 med,i represents the i-th pixel in the SVV1 med synthetic image. Pixel value, SVV2 med, i represents the pixel value of the i-th pixel in the SVV2 med composite image.
  • the method for obtaining the final remote sensing identification result of the garlic crop is as follows: when the pixel i is a garlic crop in the garlic crop radar distribution map and the garlic crop optical distribution map, it is judged that the pixel i is a garlic crop, otherwise, the pixel i is not a garlic crop. Garlic crop; traverse all the pixels in the garlic crop radar distribution map and the garlic crop optical distribution map in turn to complete the final remote sensing identification of the garlic crop.
  • the garlic crop remote sensing automatic identification model created by the present invention makes full use of the advantages that time-series optical images are sensitive to the phenological characteristics of garlic crops, and Sentinel-1 synthetic aperture radar images are sensitive to the plant structure characteristics of garlic crops and winter wheat crops. , to achieve accurate identification of garlic crop distribution by remote sensing;
  • the present invention can be transplanted to the Google Earth Engine cloud computing platform, avoids the difficult problem of remote sensing big data processing, can realize automatic and rapid identification of garlic crops on a large scale in geographic space, and provides theories and technologies for remote sensing identification of garlic crop distribution Base.
  • Fig. 1 is the flow chart of the present invention
  • Fig. 2 is the garlic crop optical image recognition decision tree model of the present invention
  • Fig. 3 is the garlic crop radar image recognition decision tree model of the present invention.
  • Fig. 4 is the garlic distribution remote sensing identification result of the present invention.
  • the embodiment of the present invention provides a method for identifying garlic crops based on a cloud platform coupled with active and passive remote sensing images, and the specific steps are as follows:
  • the main garlic producing area in North China with the target year from October 1, 2019 to June 30, 2020 (geographical coordinates: 34° north latitude to 34.6° north latitude, east longitude 113.5° to 118.5° east longitude) MODIS-NDVI time series images, according to MODIS-NDVI time series images to obtain the phenological information of garlic crops and other forest and grass vegetation, in order to discover the uniqueness of the phenology of garlic crops relative to the phenology of forest and grass vegetation (garlic crops
  • the uniqueness of garlic is as follows: from December to March of the following year, the garlic crop is in the growth period or wintering period, and the forest and grass vegetation is mostly in the deciduous and withered period (the forest and grass vegetation in the garlic planting area in northern China is mostly very green vegetation.
  • the NDVI of the garlic crop is higher than that of the forest and grass vegetation on the satellite image; during the garlic sowing and harvesting periods, the garlic crop is in the early or late growth stage, and the forest and grass vegetation is mostly in the vigorous growth period. It is shown that the NDVI of garlic crops is lower than that of forest and grass vegetation); since the phenological information of winter wheat is similar to that of garlic crops, the NDVI (normalized difference vegetation index) of winter wheat crops in this example is similar to that of garlic crops.
  • the Sentinel-2 time-series image and Landsat-8 time-series satellite image of the main garlic crop producing area of the target year are retrieved, and combined with the phenological information of the garlic crop, the optical synthetic image dataset is obtained;
  • a time-series optical image synthesis scheme is designed to enhance the image information of garlic.
  • the method for obtaining the optical synthetic image dataset is as follows: during the period when the time series MODIS-NDVI of garlic crops is higher than that of other forest and grass vegetation (December 1, 2019 to March 20, 2020), Extract the maximum value of NDVI of the Sentinel-2 time series image at the position of pixel i as the pixel value of pixel i, and traverse all the pixel positions of the Sentinel-2 time series image in turn to obtain a composite image of the maximum NDVI value, which is recorded as NDVI max ; During the period when the time series MODIS-NDVI of garlic crops is smaller than that of other forest and grass vegetation (October 1, 2019 to October 30, 2019 and May 20, 2020 to June 20, 2020 ), extract the minimum value of NDVI of the Sentinel-2 time series image at the position of pixel i as the pixel value of pixel i, traverse all the pixel positions of the Sentinel-2 time series image
  • the NDVI median synthetic image is obtained, which is recorded as NDVI med ; the NDVI maximum synthetic image NDVI max , the NDVI minimum synthetic image NDVI min and the NDVI median synthetic image NDVI med are combined into an optical synthetic image dataset.
  • the composite images are all done on the Google Earth Engine cloud computing platform.
  • the experimenter arrives at the actual distribution location of garlic and other ground objects (including winter wheat plots, woodlands, settlements, etc.), and uses handheld GPS to obtain and record the geographic coordinate information of garlic crops and winter wheat crops in the main garlic crop production area. Sample collection work.
  • step S4 Based on the optical synthetic image data set obtained in step S2 and the geographic coordinate information of the garlic crop obtained in step S3, construct a decision tree model for optical image recognition of garlic crops; count the geographic coordinates of the synthetic image NDVI max of NDVI max in the garlic crop
  • the distribution interval of the 2600 pixel values at the location is ⁇ 1, but they are concentrated in two sub-intervals, namely ⁇ and ⁇ 1, the garlic growth potential difference in the interval ⁇ or there is a mixed pixel phenomenon , the garlic grows well in the interval ⁇ 1, according to which the first threshold ⁇ of pixel maximum and the second threshold ⁇ of pixel maximum are obtained; statistical NDVI median synthetic image NDVI med on the geographic coordinate position of garlic crop
  • the distribution interval of 2600 pixel values is from 0 to ⁇ , according to which the median threshold value ⁇ of the pixel is obtained; the distribution interval of the 2600 pixel values of the synthetic image NDVI min at the geographic coordinate position of the garlic crop is less than ⁇ , according to which the
  • the decision tree model of garlic crop optical image recognition is obtained.
  • the method for obtaining the optical distribution map of the garlic crop is: constructing the first constraint condition of the garlic crop according to the garlic crop optical image recognition decision tree model, According to the first constraint condition of garlic crop, the pixels in the synthetic image of NDVI maximum value NDVI max , the synthetic image of NDVI median value NDVI med and the synthetic image of NDVI minimum value NDVI min are respectively screened, and the optical distribution map of garlic crop is obtained.
  • NDVI max,i represents the pixel value of the ith pixel in the NDVI max composite image NDVI max
  • NDVI med,i represents the pixel value of the ith pixel in the NDVI median composite image NDVI med
  • NDVI min,i represents The pixel value of the ith pixel in the NDVI min composite image NDVI min .
  • the garlic crop optical distribution map is implemented on the Google Earth Engine cloud computing platform.
  • the Sentinel-1 time-series synthetic aperture radar satellite image of the main garlic crop producing area in the target year (October 1, 2019 to June 30, 2020) is retrieved, and combined with step S3
  • the obtained geographic coordinate information of the garlic crop and the winter wheat crop obtains the image features of the garlic crop and the winter wheat crop;
  • the method for obtaining the image features of the garlic crop and the winter wheat crop is as follows: count the mean value of the pixel values of the garlic crop sample and the winter wheat crop sample on the Sentinel-1 synthetic aperture radar satellite image in each period, and arrange them in time order, that is, to obtain Image characteristics of garlic crops and winter wheat crops in time-series Sentinel-1 synthetic aperture radar satellite imagery.
  • the differences in image characteristics of garlic and winter wheat crops are as follows: from winter wheat overwintering period to winter wheat tillering and booting period, the pixel value of winter wheat in the time-series synthetic aperture radar satellite image (VV polarized image) shows a downward trend; garlic shows a downward trend. for the opposite trend.
  • step S7 obtain a radar synthetic image dataset according to the image features of garlic crops and winter wheat crops in step S6;
  • the method for obtaining the radar synthetic image dataset is as follows: the median image of the Sentinel-1 synthetic aperture radar satellite image is synthesized during the winter wheat overwintering period (January 1, 2020 to January 30, 2020), which is recorded as SVV1 med Synthetic image; the median image of Sentinel-1 synthetic aperture radar satellite image is synthesized during the tillering and booting period of winter wheat (April 1, 2020 to April 30, 2020), recorded as SVV2 med synthetic image; SVV1 med The synthetic images and SVV2 med synthetic images are combined into a radar synthetic image dataset.
  • VV represents the image imaged with the VV polarization in the Sentinel-1 image.
  • step S8 according to the radar synthetic image data set obtained in step S7 and the geographic coordinate information of garlic crops obtained in step S3, construct a garlic crop radar image recognition decision tree model;
  • the distribution over the geographic coordinate positions obtains a difference threshold ⁇ and a threshold ⁇ .
  • the SVV1 med synthetic image and the SVV2 med synthetic image are subtracted to obtain the difference image, and the pixel value of the difference image at the geographic coordinate position of the garlic crop is counted, and its distribution interval is less than ⁇ ; the statistical SVV2 med synthetic image is in The pixel value on the geographic coordinate position of garlic crop, and its distribution interval is greater than ⁇ .
  • the difference threshold ⁇ and threshold ⁇ the decision tree model of garlic crop radar image recognition is obtained.
  • the tree model builds the second constraint for garlic crops, Screen the pixels in the SVV1 med synthetic image and the SVV2 med synthetic image respectively according to the second constraint of garlic crops, and obtain the garlic crop radar distribution map, where SVV1 med,i represents the i-th pixel in the SVV1 med synthetic image.
  • the pixel value of SVV2 med, i represents the pixel value of the i-th pixel in the SVV2 med composite image.
  • the garlic crop radar distribution map is implemented on the Google Earth Engine cloud computing platform.
  • the garlic crop radar distribution map in step S9 and the garlic crop optical distribution map in step S5 are coupled, that is, the intersection of the two distribution maps is selected to obtain the garlic crop remote sensing identification result.
  • pixel i is a garlic crop in both the garlic crop radar distribution map and the garlic crop optical distribution map, it is judged that the pixel i is a garlic crop, otherwise, the pixel i is not a garlic crop; traverse the garlic crop radar distribution map and the garlic crop in turn. All the pixels in the optical distribution map complete the remote sensing identification of garlic crops.
  • FIG. 4 The identification result of this embodiment is shown in FIG. 4 . It can be seen from Figure 4 that the main garlic concentrated planting areas in the example area have been completely identified, such as Shandong Jinxiang County, Henan Qi County, Jiangsu Pizhou and other garlic concentrated planting areas. It can be seen from the partial enlarged picture that the texture information such as the boundary of the garlic planting plot is complete, and other features such as roads can be effectively distinguished, which shows the reliability and accuracy of the invention for garlic distribution identification.

Landscapes

  • Engineering & Computer Science (AREA)
  • Remote Sensing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Electromagnetism (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Astronomy & Astrophysics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Multimedia (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Image Processing (AREA)

Abstract

本发明提出了一种基于云平台的耦合主被动遥感影像的大蒜作物识别方法，其步骤为：首先，基于大蒜物候特征获得光学卫星遥感影像，并结合大蒜的地理坐标信息构建大蒜光学影像识别决策树模型，进而得到大蒜光学分布图；其次，基于合成孔径雷达卫星获得大蒜、冬小麦的雷达影像特征，并结合大蒜的地理坐标信息构建大蒜雷达影像识别决策树模型，进而得到大蒜雷达分布图；最后，对大蒜光学分布图、大蒜雷达分布图进行耦合，即选取两种分布图的交集，完成大蒜遥感识别制图。本发明综合利用了光学卫星影像和合成孔径雷达影像的优势，解决了大蒜与冬小麦不易区分的技术难题，实现了基于云平台的地理空间大区域尺度上大蒜分布遥感精准自动识别。

Description

基于云平台的耦合主被动遥感影像的大蒜作物识别方法 技术领域
本发明涉及遥感目标识别技术领域，特别是指一种基于云平台的耦合主被动遥感影像的大蒜作物识别方法。
背景技术
大蒜作物和冬小麦作物同属于越冬作物，具有相似的生长期和光谱特征。单一依靠光学卫星影像难以准确区分大蒜作物与冬小麦作物。相对冬小麦作物的播种面积，大蒜作物的播种面积较少，且大蒜地块多与冬小麦地块交错分布，进一步制约了大蒜作物遥感识别的精度。如何通过卫星影像实现大蒜作物与冬小麦作物的准确区分，是作物遥感识别领域的难点之一。
遥感大数据的处理过程复杂，数据计算量巨大，远远超出个人计算机的数据处理能力。Google Earth Engine云计算平台集成了全球公开的遥感数据，并且提供个性化的数据处理服务，用户只需将相关数据处理算法以Google Earth Engine语言的方式编程运行，Google Earth Engine在后台就会自动调用Google的数以万计的服务器以并行运算的方式快速完成数据处理任务，并将结果反馈给用户。Google Earth Engine解决了遥感大数据处理的难题，为作物遥感识别的自动化、业务化运行提供了机遇和平台。
发明内容
针对上述背景技术中存在的不足，本发明提出了一种基于云平台的耦合主被动遥感影像的大蒜作物识别方法，解决了现有遥感识别技术中无法准确识别大蒜作物的技术问题。
本发明的技术方案是这样实现的：
一种基于云平台的耦合主被动遥感影像的大蒜作物识别方法，其步骤如下：
S1、在Google Earth Engine云计算平台上通过调取目标年份大蒜作物主产区的MODIS-NDVI时序影像，根据MODIS-NDVI时序影像获取大蒜作物和其他林草植被的物候信息；
S2、在Google Earth Engine云计算平台上通过调取目标年份大蒜作物主产区的Sentinel-2时序影像及Landsat-8时序卫星影像，并结合大蒜作物的物候信息获得光学合成影像数据集；
S3、使用手持GPS获取、记录大蒜作物主产区的大蒜作物和冬小麦作物的地理坐标信息；
S4、基于步骤S2中得到的光学合成影像数据集和步骤S3中得到的大蒜作物的地理坐标信息，构建大蒜作物光学影像识别决策树模型；
S5、根据步骤S4中得到的大蒜作物光学影像识别决策树模型对步骤S2中得到的光学合成影像数据集进行分类，得到大蒜作物光学分布图；
S6、在Google Earth Engine云计算平台上通过调取目标年份大蒜作物主产区的Sentinel-1时序合成孔径雷达卫星影像，结合步骤S3中得到的大蒜作物、冬小麦作物的地理坐标信息获取大蒜作物和冬小麦作物的雷达图像特征；
S7、根据步骤S6中大蒜作物和冬小麦作物的图像特征获得雷达合成影像数据集；
S8、根据步骤S7中得到的雷达合成影像数据集和步骤S3中得到的大蒜作物的地理坐标信息，构建大蒜作物雷达影像识别决策树模型；
S9、根据步骤S8中得到的大蒜作物雷达影像识别决策树模型对步骤S7中得到的雷达合成影像数据集进行分类，得到大蒜作物雷达分布图；
S10、在Google Earth Engine云计算平台上将步骤S9中的大蒜作物雷达分布图和步骤S5中的大蒜作物光学分布图进行耦合，得到大蒜作物遥感识别结果。
所述光学合成影像数据集的获得方法为：在大蒜作物的时序MODIS-NDVI高于其他林草植被的时序MODIS-NDVI的时期内，提取像元i位置上的Sentinel-2时序影像的NDVI的最大值作为像元i的像元值，依次遍历Sentinel-2时序影像的所有像元位置，得到NDVI最大值合成影像，记为NDVI max；在大蒜作物的时序MODIS-NDVI小于其他林草植被的时序MODIS-NDVI的时期内，提取像元i位置上的Sentinel-2和Landsat-8时序影像的NDVI的最小值作为像元i的像元值，依次遍历Sentinel-2和Landsat-8时序影像的所有像元位置，得到NDVI最小值合成影像，记为NDVI min；在大蒜作物的时序MODIS-NDVI小于其他林草植被的时序MODIS-NDVI的时期内，提取像元i位置上的Sentinel-2和Landsat-8时序卫星影像的NDVI的中值作为像元i的像元值，依次遍历Sentinel-2和Landsat-8时序卫星影像的所有像元位置，得到NDVI中值合成影像，记为NDVI med；将NDVI最大值合成影像NDVI max、NDVI最小值合成影像NDVI min和NDVI中值合成影像NDVI med组合为光学合成影像数据集。
所述大蒜作物光学影像识别决策树模型的构建方法为：根据NDVI中值合成影像NDVI med在大蒜作物的地理坐标位置上的分布获得像元中值阈值α；根据NDVI最大值合成影像NDVI max在大蒜作物的地理坐标位置上的分布获得像元最大值第一阈值γ和像元最大值第二阈值β；根据NDVI最小值合成影像NDVI min在大蒜作物的地理坐标位置上的分布获得像元最小值阈值δ；根据像元中值阈值α、像元最大值第一阈值γ、像元最大值第二阈值β和像元最小值阈值δ得到大蒜作物光学影像识别决策树模型。
所述大蒜作物光学分布图的获得方法为：根据大蒜作物光学影像识别决策树模型构建大 蒜作物的第一约束条件：
Figure PCTCN2021097839-appb-000001
根据大蒜作物的第一约束条件分别对NDVI最大值合成影像NDVI max、NDVI中值合成影像NDVI med和NDVI最小值合成影像NDVI min中的像元进行筛选，得到大蒜作物光学分布图，其中，NDVI max,i表示NDVI最大值合成影像NDVI max中第i个像元的像元值，NDVI med,i表示NDVI中值合成影像NDVI med中第i个像元的像元值，NDVI min,i表示NDVI最小值合成影像NDVI min中第i个像元的像元值。
所述雷达合成影像数据集的获得方法为：在冬小麦越冬期内合成Sentinel-1合成孔径雷达卫星影像的中值影像，记为SVV1 med合成影像；在冬小麦分蘖至孕穗期内合成Sentinel-1合成孔径雷达卫星影像的中值影像，记为SVV2 med合成影像；将SVV1 med合成影像和SVV2 med合成影像组合为雷达合成影像数据集。
所述大蒜作物雷达影像识别决策树模型的构建方法为：根据SVV1 med合成影像和SVV2 med合成影像在大蒜作物的地理坐标位置上的分布获得差值阈值ε和阈值ζ；根据差值阈值ε和阈值ζ得到大蒜作物雷达影像识别决策树模型。
所述大蒜作物雷达分布图的获得方法为：根据大蒜作物雷达影像识别决策树模型构建大蒜作物的第二约束条件，
Figure PCTCN2021097839-appb-000002
根据大蒜作物的第二约束条件对SVV1 med合成影像和SVV2 med合成影像中的像元进行筛选，得到大蒜作物雷达分布图，其中，SVV1 med,i表示SVV1 med合成影像中第i个像元的像元值，SVV2 med,i表示SVV2 med合成影像中第i个像元的像元值。
所述大蒜作物最终遥感识别结果的获得方法为：当像元i在大蒜作物雷达分布图和大蒜作物光学分布图中均为大蒜作物时，判断像元i为大蒜作物，否则，像元i不是大蒜作物；依次遍历大蒜作物雷达分布图和大蒜作物光学分布图中的所有像元，完成大蒜作物的最终遥感识别。
本技术方案能产生的有益效果：
(1)本发明创建的大蒜作物遥感自动识别模型，充分利用了时序光学影像对大蒜作物物候特征响应敏感，以及Sentinel-1合成孔径雷达影像对大蒜作物与冬小麦作物的植株结构特征响应敏感的优势，实现了大蒜作物分布遥感精准识别；
(2)本发明可移植到Google Earth Engine云计算平台上，避免了遥感大数据处理的难题， 可实现地理空间大尺度上的大蒜作物自动快速识别，为大蒜作物分布遥感识别提供了理论和技术基础。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本发明的流程图；
图2为本发明的大蒜作物光学影像识别决策树模型；
图3为本发明的大蒜作物雷达影像识别决策树模型；
图4为本发明的大蒜分布遥感识别结果。
具体实施方式
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有付出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
如图1所示，本发明实施例提供了一种基于云平台的耦合主被动遥感影像的大蒜作物识别方法，具体步骤如下：
S1、在Google Earth Engine云计算平台上通过调取目标年份为2019年10月1日至2020年6月30日的华北地区大蒜主产区(地理坐标为：北纬34°至北纬34.6°，东经113.5°至东经118.5°)的MODIS-NDVI时序影像，根据MODIS-NDVI时序影像获取大蒜作物和其他林草植被的物候信息，以发现大蒜作物的物候相对于林草植被物候的独特性(大蒜作物的独特性具体表现为：在12月至次年3月内，大蒜作物处于生长期或越冬期，林草植被多处于落叶枯萎期(中国北方大蒜种植区内的林草植被多为非常绿植被)，在卫星影像上表现为大蒜作物的NDVI高于林草植被的NDVI；在大蒜播种期及收获期内，大蒜作物处于生长初期或末期，林草植被多处于生长旺盛期，在卫星影像上表现为大蒜作物的NDVI低于林草植被的NDVI)；由于冬小麦的物候信息与大蒜作物的物候信息相似，本实施例中冬小麦作物的NDVI(normalized difference vegetation index)与大蒜作物的NDVI相似。
S2、在Google Earth Engine云计算平台上通过调取目标年份大蒜作物主产区的Sentinel-2时序影像及Landsat-8时序卫星影像，并结合大蒜作物的物候信息获得光学合成影像数据集；
根据大蒜物候的独特性，设计时序光学影像合成方案，以增强大蒜的影像信息。所述光 学合成影像数据集的获得方法为：在大蒜作物的时序MODIS-NDVI高于其他林草植被的时序MODIS-NDVI的时期内(2019年12月1日至2020年3月20日)，提取像元i位置上的Sentinel-2时序影像的NDVI的最大值作为像元i的像元值，依次遍历Sentinel-2时序影像的所有像元位置，得到NDVI最大值合成影像，记为NDVI max；在大蒜作物的时序MODIS-NDVI小于其他林草植被的时序MODIS-NDVI的时期内(2019年10月1日至2019年10月30日和2020年5月20日至2020年6月20日)，提取像元i位置上的Sentinel-2时序影像的NDVI的最小值作为像元i的像元值，依次遍历Sentinel-2时序影像的所有像元位置，得到NDVI最小值合成影像，记为NDVI min；在大蒜作物的时序MODIS-NDVI小于其他林草植被的时序MODIS-NDVI的时期内(2019年10月1日至2019年10月30日和2020年5月20日至2020年6月20日)，提取像元i位置上的Sentinel-2和Landsat-8时序卫星影像的NDVI的中值作为像元i的像元值，依次遍历Sentinel-2和Landsat-8时序卫星影像的所有像元位置，得到NDVI中值合成影像，记为NDVI med；将NDVI最大值合成影像NDVI max、NDVI最小值合成影像NDVI min和NDVI中值合成影像NDVI med组合为光学合成影像数据集。合成影像均在Google Earth Engine云计算平台上完成的。
S3、实验人员到达大蒜及其他地物(包括冬小麦地块、林地、居民点等)的实际分布地点，使用手持GPS获取、记录大蒜作物主产区的大蒜作物和冬小麦作物的地理坐标信息，完成样本采集工作。
S4、基于步骤S2中得到的光学合成影像数据集和步骤S3中得到的大蒜作物的地理坐标信息，构建大蒜作物光学影像识别决策树模型；统计NDVI最大值合成影像NDVI max在大蒜作物的地理坐标位置上的2600个像元值的分布区间为γ～1，但集中分布在两个子区间内，即γ～β和β～1，在区间γ～β内的大蒜长势差或存在混合像元现象，在区间β～1内的大蒜长势良好，据此获得像元最大值第一阈值γ和像元最大值第二阈值β；统计NDVI中值合成影像NDVI med在大蒜作物的地理坐标位置上的2600个像元值的分布区间为0～α，据此获得像元中值阈值α；统计NDVI最小值合成影像NDVI min在大蒜作物的地理坐标位置上的2600个像元值的分布区间为小于δ，据此获得像元最小值阈值δ。根据像元最大值第一阈值γ、像元最大值第二阈值β、像元中值阈值α和像元最小值阈值δ得到大蒜作物光学影像识别决策树模型。本实施例中，α＝0.51、β＝0.48、γ＝0.33、δ＝0.15。
S5、根据步骤S4中得到的大蒜作物光学影像识别决策树模型对步骤S2中得到的光学合成影像数据集进行分类，得到大蒜作物光学分布图；
如图2所述，所述大蒜作物光学分布图的获得方法为：根据大蒜作物光学影像识别决策 树模型构建大蒜作物的第一约束条件，
Figure PCTCN2021097839-appb-000003
根据大蒜作物的第一约束条件分别对NDVI最大值合成影像NDVI max、NDVI中值合成影像NDVI med和NDVI最小值合成影像NDVI min中的像元进行筛选，得到大蒜作物光学分布图，其中，NDVI max,i表示NDVI最大值合成影像NDVI max中第i个像元的像元值，NDVI med,i表示NDVI中值合成影像NDVI med中第i个像元的像元值，NDVI min,i表示NDVI最小值合成影像NDVI min中第i个像元的像元值。大蒜作物光学分布图是在Google Earth Engine云计算平台上实现的。
S6、在Google Earth Engine云计算平台上通过调取目标年份(2019年10月1日至2020年6月30日)大蒜作物主产区的Sentinel-1时序合成孔径雷达卫星影像，结合步骤S3中得到的大蒜作物、冬小麦作物的地理坐标信息获取大蒜作物和冬小麦作物的图像特征；
所述大蒜作物和冬小麦作物的图像特征的获得方法为：统计大蒜作物样本和冬小麦作物样本在每期Sentinel-1合成孔径雷达卫星影像上的像元值的均值，并按照时间顺序排列，即得到大蒜作物和冬小麦作物在时序Sentinel-1合成孔径雷达卫星影像中的图像特征。大蒜作物和冬小麦作物图像特征的差异性表现为：从冬小麦越冬期至冬小麦分蘖孕穗期内，冬小麦在时序合成孔径雷达卫星影像(VV极化影像)中的像元值呈现下降趋势；大蒜则表现为与之相反的趋势。
S7、根据步骤S6中大蒜作物和冬小麦作物的图像特征获得雷达合成影像数据集；
根据大蒜和冬小麦的图像特征(从冬小麦越冬期至冬小麦分蘖孕穗期内，冬小麦在时序合成孔径雷达卫星影像(VV极化影像)中的像元值呈现下降趋势；大蒜则呈现总体上升趋势)，设计时序Sentinel-1影像的合成方案，实现冗余数据的剔除和大蒜与冬小麦影像特征差异的增强。所述雷达合成影像数据集的获得方法为：在冬小麦越冬期(2020年1月1日至2020年1月30日)内合成Sentinel-1合成孔径雷达卫星影像的中值影像，记为SVV1 med合成影像；在冬小麦的分蘖、孕穗期内(2020年4月1日至2020年4月30日)合成Sentinel-1合成孔径雷达卫星影像的中值影像，记为SVV2 med合成影像；将SVV1 med合成影像和SVV2 med合成影像组合为雷达合成影像数据集。VV表示Sentinel-1影像中以VV极化方式成像的影像。
S8、根据步骤S7中得到的雷达合成影像数据集和步骤S3中得到的大蒜作物的地理坐标信息，构建大蒜作物雷达影像识别决策树模型；根据SVV1 med合成影像和SVV2 med合成影像在大蒜作物的地理坐标位置上的分布获得差值阈值ε和阈值ζ。
将SVV1 med合成影像与SVV2 med合成影像作减法运算，得到其差值影像，统计差值影像 在大蒜作物的地理坐标位置上的像元值，其分布区间为小于ε；统计SVV2 med合成影像在大蒜作物的地理坐标位置上的像元值，其分布区间为大于ζ。根据差值阈值ε和阈值ζ得到大蒜作物雷达影像识别决策树模型。本实施例中，ε＝3dB、ζ＝-16dB。
S9、根据步骤S8中得到的大蒜作物雷达影像识别决策树模型对步骤S7中得到的雷达合成影像数据集进行分类，得到大蒜作物雷达分布图；如图3所示，根据大蒜作物雷达影像识别决策树模型构建大蒜作物的第二约束条件，
Figure PCTCN2021097839-appb-000004
根据大蒜作物的第二约束条件分别对SVV1 med合成影像和SVV2 med合成影像中的像元进行筛选，得到大蒜作物雷达分布图，其中，SVV1 med,i表示SVV1 med合成影像中第i个像元的像元值，SVV2 med,i表示SVV2 med合成影像中第i个像元的像元值。大蒜作物雷达分布图是在Google Earth Engine云计算平台上实现的。
S10、在Google Earth Engine云计算平台上将步骤S9中的大蒜作物雷达分布图和步骤S5中的大蒜作物光学分布图进行耦合，即选取两种分布图的交集，得到大蒜作物遥感识别结果。当像元i在大蒜作物雷达分布图和大蒜作物光学分布图中均为大蒜作物时，判断像元i为大蒜作物，否则，像元i不是大蒜作物；依次遍历大蒜作物雷达分布图和大蒜作物光学分布图中的所有像元，完成大蒜作物的遥感识别。
本实施例的识别结果如图4所示。通过图4可以看出实施例区域内的主要大蒜集中种植区得到了完整识别，如山东金乡县、河南杞县、江苏邳州等大蒜集中种植区。从局部放大图中可以看出大蒜种植地块的边界等纹理信息完整，道路等其他地物可以被有效区分，说明了本发明对大蒜分布识别的可靠性、准确性。
以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内，所作的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。

Claims (8)

  1. 一种基于云平台的耦合主被动遥感影像的大蒜作物识别方法，其特征在于，其步骤如下：
    S1、在Google Earth Engine云计算平台上通过调取目标年份大蒜作物主产区的MODIS-NDVI时序影像，根据MODIS-NDVI时序影像获取大蒜作物和其他林草植被的物候信息；
    S2、在Google Earth Engine云计算平台上通过调取目标年份大蒜作物主产区的Sentinel-2时序影像及Landsat-8时序卫星影像，并结合大蒜作物的物候信息获得光学合成影像数据集；
    S3、使用手持GPS获取、记录大蒜作物主产区的大蒜作物和冬小麦作物的地理坐标信息；
    S4、基于步骤S2中得到的光学合成影像数据集和步骤S3中得到的大蒜作物的地理坐标信息，构建大蒜作物光学影像识别决策树模型；
    S5、根据步骤S4中得到的大蒜作物光学影像识别决策树模型对步骤S2中得到的光学合成影像数据集进行分类，得到大蒜作物光学分布图；
    S6、在Google Earth Engine云计算平台上通过调取目标年份大蒜作物主产区的Sentinel-1时序合成孔径雷达卫星影像，结合步骤S3中得到的大蒜作物、冬小麦作物的地理坐标信息获取大蒜作物和冬小麦作物的雷达图像特征；
    S7、根据步骤S6中大蒜作物和冬小麦作物的图像特征获得雷达合成影像数据集；
    S8、根据步骤S7中得到的雷达合成影像数据集和步骤S3中得到的大蒜作物的地理坐标信息，构建大蒜作物雷达影像识别决策树模型；
    S9、根据步骤S8中得到的大蒜作物雷达影像识别决策树模型对步骤S7中得到的雷达合成影像数据集进行分类，得到大蒜作物雷达分布图；
    S10、在Google Earth Engine云计算平台上将步骤S9中的大蒜作物雷达分布图和步骤S5中的大蒜作物光学分布图进行耦合，得到大蒜作物遥感识别结果。
  2. 根据权利要求1所述的基于云平台的耦合主被动遥感影像的大蒜作物识别方法，其特征在于，所述光学合成影像数据集的获得方法为：在大蒜作物的时序MODIS-NDVI高于其他林草植被的时序MODIS-NDVI的时期内，提取像元i位置上的Sentinel-2时序影像的NDVI的最大值作为像元i的像元值，依次遍历Sentinel-2时序影像的所有像元位置，得到NDVI最大值合成影像，记为NDVI max；在大蒜作物的时序MODIS-NDVI小于其他林草植被的时序MODIS-NDVI的时期内，提取像元i位置上的Sentinel-2和Landsat-8时序影像的NDVI的最小值作为像元i的像元值，依次遍历Sentinel-2和Landsat-8时序影像的所有像元位置，得到NDVI最小值合成影像，记为NDVI min；在大蒜作物的时序MODIS-NDVI小于其他林草植被的时序MODIS-NDVI的时期内，提取像元i位置上的Sentinel-2和Landsat-8时序卫星影像的 NDVI的中值作为像元i的像元值，依次遍历Sentinel-2和Landsat-8时序卫星影像的所有像元位置，得到NDVI中值合成影像，记为NDVI med；将NDVI最大值合成影像NDVI max、NDVI最小值合成影像NDVI min和NDVI中值合成影像NDVI med组合为光学合成影像数据集。
  3. 根据权利要求2所述的基于云平台的耦合主被动遥感影像的大蒜作物识别方法，其特征在于，所述大蒜作物光学影像识别决策树模型的构建方法为：根据NDVI中值合成影像NDVI med在大蒜作物的地理坐标位置上的分布获得像元中值阈值α；根据NDVI最大值合成影像NDVI max在大蒜作物的地理坐标位置上的分布获得像元最大值第一阈值γ和像元最大值第二阈值β；根据NDVI最小值合成影像NDVI min在大蒜作物的地理坐标位置上的分布获得像元最小值阈值δ；根据像元中值阈值α、像元最大值第一阈值γ、像元最大值第二阈值β和像元最小值阈值δ得到大蒜作物光学影像识别决策树模型。
  4. 根据权利要求3所述的基于云平台的耦合主被动遥感影像的大蒜作物识别方法，其特征在于，所述大蒜作物光学分布图的获得方法为：根据大蒜作物光学影像识别决策树模型构建大蒜作物的第一约束条件：
    Figure PCTCN2021097839-appb-100001
    或
    Figure PCTCN2021097839-appb-100002
    根据大蒜作物的第一约束条件分别对NDVI最大值合成影像NDVI max、NDVI中值合成影像NDVI med和NDVI最小值合成影像NDVI min中的像元进行筛选，得到大蒜作物光学分布图，其中，NDVI max,i表示NDVI最大值合成影像NDVI max中第i个像元的像元值，NDVI med,i表示NDVI中值合成影像NDVI med中第i个像元的像元值，NDVI min,i表示NDVI最小值合成影像NDVI min中第i个像元的像元值。
  5. 根据权利要求1或4所述的基于云平台的耦合主被动遥感影像的大蒜作物识别方法，其特征在于，所述雷达合成影像数据集的获得方法为：在冬小麦越冬期内合成Sentinel-1合成孔径雷达卫星影像的中值影像，记为SVV1 med合成影像；在冬小麦分蘖至孕穗期内合成Sentinel-1合成孔径雷达卫星影像的中值影像，记为SVV2 med合成影像；将SVV1 med合成影像和SVV2 med合成影像组合为雷达合成影像数据集。
  6. 根据权利要求5所述的基于云平台的耦合主被动遥感影像的大蒜作物识别方法，其特征在于，所述大蒜作物雷达影像识别决策树模型的构建方法为：根据SVV1 med合成影像和SVV2 med合成影像在大蒜作物的地理坐标位置上的分布获得差值阈值ε和阈值ζ；根据差值阈值ε和阈值ζ得到大蒜作物雷达影像识别决策树模型。
  7. 根据权利要求6所述的基于云平台的耦合主被动遥感影像的大蒜作物识别方法，其特 征在于，所述大蒜作物雷达分布图的获得方法为：根据大蒜作物雷达影像识别决策树模型构建大蒜作物的第二约束条件，
    Figure PCTCN2021097839-appb-100003
    根据大蒜作物的第二约束条件对SVV1 med合成影像和SVV2 med合成影像中的像元进行筛选，得到大蒜作物雷达分布图，其中，SVV1 med,i表示SVV1 med合成影像中第i个像元的像元值，SVV2 med,i表示SVV2 med合成影像中第i个像元的像元值。
  8. 根据权利要求7所述的基于云平台的耦合主被动遥感影像的大蒜作物识别方法，其特征在于，所述大蒜作物最终遥感识别结果的获得方法为：当像元i在大蒜作物雷达分布图和大蒜作物光学分布图中均为大蒜作物时，判断像元i为大蒜作物，否则，像元i不是大蒜作物；依次遍历大蒜作物雷达分布图和大蒜作物光学分布图中的所有像元，完成大蒜作物的最终遥感识别。
PCT/CN2021/097839 2020-09-21 2021-06-02 基于云平台的耦合主被动遥感影像的大蒜作物识别方法 WO2022057319A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/531,826 US12094198B2 (en) 2020-09-21 2021-11-22 Cloud platform-based garlic crop recognition method by coupling active and passive remote sensing images

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010995102.7 2020-09-21
CN202010995102.7A CN112101256B (zh) 2020-09-21 2020-09-21 基于云平台的耦合主被动遥感影像的大蒜作物识别方法

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/531,826 Continuation US12094198B2 (en) 2020-09-21 2021-11-22 Cloud platform-based garlic crop recognition method by coupling active and passive remote sensing images

Publications (1)

Publication Number Publication Date
WO2022057319A1 true WO2022057319A1 (zh) 2022-03-24

Family

ID=73756348

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/097839 WO2022057319A1 (zh) 2020-09-21 2021-06-02 基于云平台的耦合主被动遥感影像的大蒜作物识别方法

Country Status (3)

Country Link
CN (1) CN112101256B (zh)
WO (1) WO2022057319A1 (zh)
ZA (1) ZA202109334B (zh)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114782840A (zh) * 2022-04-20 2022-07-22 南京农业大学 一种基于无人机rgb影像的小麦物候期实时分类方法
CN115372970A (zh) * 2022-08-19 2022-11-22 陕西省土地工程建设集团有限责任公司 一种山区丘陵区作物sar遥感提取方法
CN115641504A (zh) * 2022-10-26 2023-01-24 南京农业大学 一种基于作物物候特征与决策树模型的田块边界自动化遥感提取方法
CN116664959A (zh) * 2023-07-28 2023-08-29 航天宏图信息技术股份有限公司 基于多源卫星遥感的无样本农作物分类方法及装置
CN117423003A (zh) * 2023-12-19 2024-01-19 山东科技大学 一种冬小麦越冬期苗情分级遥感监测方法
CN117437534A (zh) * 2023-08-28 2024-01-23 北京爱科农科技有限公司 一种土壤地力确定方法及装置
CN117456368A (zh) * 2023-12-22 2024-01-26 安徽大学 一种果蔬识别采摘方法、系统及装置
CN117576580A (zh) * 2023-11-17 2024-02-20 广东省地质测绘院 基于遥感卫星的虾稻田信息提取方法、电子设备和介质
CN117953278A (zh) * 2024-01-10 2024-04-30 河海大学 一种基于遥感云平台和机器学习算法的灌区作物分类方法及系统
CN118363007A (zh) * 2024-06-19 2024-07-19 自然资源部第二海洋研究所 一种基于潮滩遥感淹没频率校正的潮滩地形时序监测方法及系统

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US12094198B2 (en) 2020-09-21 2024-09-17 Henan University Cloud platform-based garlic crop recognition method by coupling active and passive remote sensing images
CN112101256B (zh) * 2020-09-21 2021-07-09 河南大学 基于云平台的耦合主被动遥感影像的大蒜作物识别方法
CN114792491A (zh) * 2021-01-26 2022-07-26 深圳市柔宇科技股份有限公司 可折叠显示模组及可折叠显示设备
CN113505635B (zh) * 2021-05-24 2024-05-31 中国农业大学 基于光学和雷达的冬小麦与大蒜混种区识别方法及装置
CN113221806B (zh) * 2021-05-25 2022-02-01 河南大学 基于云平台融合多源卫星影像和茶树物候期的茶园自动识别方法
CN114419463B (zh) * 2022-01-26 2022-09-30 河南大学 一种基于云平台的全球太阳能光伏板遥感自动识别方法
CN114821362B (zh) * 2022-07-01 2022-09-23 江苏省水利科学研究院 一种基于多源数据的水稻种植面积提取方法
CN115457401B (zh) * 2022-08-30 2023-11-03 中海云科(北京)科技有限公司 一种针对不同淡水资源类型的sar遥感精细识别方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106897707A (zh) * 2017-03-02 2017-06-27 苏州中科天启遥感科技有限公司 基于多源中分的特征影像时间序列合成方法及装置
WO2019008570A1 (en) * 2017-07-02 2019-01-10 Manna Irrigation Ltd. METHODS AND SYSTEMS FOR GUIDING IRRIGATION
CN110751727A (zh) * 2019-10-30 2020-02-04 南京林业大学 一种基于Landsat长时间序列的合成图像构建方法
CN111652193A (zh) * 2020-07-08 2020-09-11 中南林业科技大学 基于多源影像的湿地分类方法
CN111666815A (zh) * 2020-05-06 2020-09-15 武汉大学 一种基于Sentinel-2遥感影像的大蒜种植信息自动化提取方法
CN112101256A (zh) * 2020-09-21 2020-12-18 河南大学 基于云平台的耦合主被动遥感影像的大蒜作物识别方法

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106529451B (zh) * 2016-10-28 2019-06-18 山东省农业可持续发展研究所 一种冬小麦—夏玉米种植模式遥感识别方法
CN108399400A (zh) * 2018-03-23 2018-08-14 中国农业科学院农业资源与农业区划研究所 一种基于高分辨率遥感数据的早期作物识别方法和系统
CN109919083B (zh) * 2019-03-06 2020-11-17 武汉大学 一种基于Sentinel-2影像数据的早期自动化冬小麦制图方法
CN110298322B (zh) * 2019-07-02 2021-05-14 北京师范大学 一种基于遥感数据的耕地提取方法及系统
CN111104896B (zh) * 2019-12-18 2021-08-27 河南大学 一种基于Sentinel-1微波卫星影像的陆表水自动识别方法
CN111598019B (zh) * 2020-05-19 2023-05-26 华中农业大学 基于多源遥感数据的作物类型与种植模式识别方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106897707A (zh) * 2017-03-02 2017-06-27 苏州中科天启遥感科技有限公司 基于多源中分的特征影像时间序列合成方法及装置
WO2019008570A1 (en) * 2017-07-02 2019-01-10 Manna Irrigation Ltd. METHODS AND SYSTEMS FOR GUIDING IRRIGATION
CN110751727A (zh) * 2019-10-30 2020-02-04 南京林业大学 一种基于Landsat长时间序列的合成图像构建方法
CN111666815A (zh) * 2020-05-06 2020-09-15 武汉大学 一种基于Sentinel-2遥感影像的大蒜种植信息自动化提取方法
CN111652193A (zh) * 2020-07-08 2020-09-11 中南林业科技大学 基于多源影像的湿地分类方法
CN112101256A (zh) * 2020-09-21 2020-12-18 河南大学 基于云平台的耦合主被动遥感影像的大蒜作物识别方法

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114782840A (zh) * 2022-04-20 2022-07-22 南京农业大学 一种基于无人机rgb影像的小麦物候期实时分类方法
CN115372970A (zh) * 2022-08-19 2022-11-22 陕西省土地工程建设集团有限责任公司 一种山区丘陵区作物sar遥感提取方法
CN115641504A (zh) * 2022-10-26 2023-01-24 南京农业大学 一种基于作物物候特征与决策树模型的田块边界自动化遥感提取方法
CN116664959A (zh) * 2023-07-28 2023-08-29 航天宏图信息技术股份有限公司 基于多源卫星遥感的无样本农作物分类方法及装置
CN116664959B (zh) * 2023-07-28 2023-11-03 航天宏图信息技术股份有限公司 基于多源卫星遥感的无样本农作物分类方法及装置
CN117437534A (zh) * 2023-08-28 2024-01-23 北京爱科农科技有限公司 一种土壤地力确定方法及装置
CN117576580A (zh) * 2023-11-17 2024-02-20 广东省地质测绘院 基于遥感卫星的虾稻田信息提取方法、电子设备和介质
CN117423003A (zh) * 2023-12-19 2024-01-19 山东科技大学 一种冬小麦越冬期苗情分级遥感监测方法
CN117423003B (zh) * 2023-12-19 2024-03-19 山东科技大学 一种冬小麦越冬期苗情分级遥感监测方法
CN117456368A (zh) * 2023-12-22 2024-01-26 安徽大学 一种果蔬识别采摘方法、系统及装置
CN117456368B (zh) * 2023-12-22 2024-03-08 安徽大学 一种果蔬识别采摘方法、系统及装置
CN117953278A (zh) * 2024-01-10 2024-04-30 河海大学 一种基于遥感云平台和机器学习算法的灌区作物分类方法及系统
CN118363007A (zh) * 2024-06-19 2024-07-19 自然资源部第二海洋研究所 一种基于潮滩遥感淹没频率校正的潮滩地形时序监测方法及系统

Also Published As

Publication number Publication date
ZA202109334B (en) 2022-02-23
CN112101256A (zh) 2020-12-18
CN112101256B (zh) 2021-07-09

Similar Documents

Publication Publication Date Title
WO2022057319A1 (zh) 基于云平台的耦合主被动遥感影像的大蒜作物识别方法
CA3088641C (en) Crop type classification in images
De Wit et al. Efficiency and accuracy of per-field classification for operational crop mapping
Ismail et al. Satellite data classification accuracy assessment based from reference dataset
Alganci et al. Parcel-level identification of crop types using different classification algorithms and multi-resolution imagery in Southeastern Turkey
CN109117811A (zh) 一种基于低空遥感测量技术估算城市植被覆盖率的系统及方法
CN110598532B (zh) 一种树木病虫害监控系统及方法
JP5886054B2 (ja) 画像解析システム、及び方法
JP6273473B2 (ja) 森林情報管理装置
CN107527014A (zh) 县级作物种植面积遥感统计抽样调查方案设计方法
Sharifi Estimation of biophysical parameters in wheat crops in Golestan province using ultra-high resolution images
Aplin et al. Advances in object-based image classification
CN109508633A (zh) 一种基于光学遥感数据的甘蔗分布识别方法
Qian et al. Mapping regional cropping patterns by using GF-1 WFV sensor data
CN109918531A (zh) 一种毒品原植物的搜查方法、装置和计算机可读存储介质
Liu et al. An efficient approach based on UAV orthographic imagery to map paddy with support of field-level canopy height from point cloud data
Liu et al. UAV-based crops classification with joint features from orthoimage and DSM data
Diao Innovative pheno-network model in estimating crop phenological stages with satellite time series
Thomas et al. Impacts of abrupt terrain changes and grass cover on vertical accuracy of UAS-SfM derived elevation models
US12094198B2 (en) Cloud platform-based garlic crop recognition method by coupling active and passive remote sensing images
Mustafa et al. Object based technique for delineating and mapping 15 tree species using VHR WorldView-2 imagery
Ozdarici-Ok et al. Object-based classification of multi-temporal images for agricultural crop mapping in Karacabey Plain, Turkey
Davidse Semi-automatic detection of field boundaries from high-resolution satellite imagery
Sun et al. How can integrated Space–Air–Ground observation contribute in aboveground biomass of shrub plants estimation in shrub-encroached Grasslands?
Arikan Parcel based crop mapping through multi-temporal masking classification of Landsat 7 images in Karacabey, Turkey

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21868155

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21868155

Country of ref document: EP

Kind code of ref document: A1