CN114881127A - Crop fine classification method based on high-resolution remote sensing satellite image - Google Patents
Crop fine classification method based on high-resolution remote sensing satellite image Download PDFInfo
- Publication number
- CN114881127A CN114881127A CN202210438771.3A CN202210438771A CN114881127A CN 114881127 A CN114881127 A CN 114881127A CN 202210438771 A CN202210438771 A CN 202210438771A CN 114881127 A CN114881127 A CN 114881127A
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- image
- algorithm
- resolution remote
- sensing satellite
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a crop fine classification method based on a high-resolution remote sensing satellite image, which comprises the following steps: s1, collecting high-resolution remote sensing satellite images and field sample data in a research area; s2, preprocessing the high-resolution remote sensing satellite image and the field sample data respectively; s3, processing the preprocessed high-resolution remote sensing satellite image by adopting a superpixel segmentation algorithm, acquiring a segmented remote sensing image vector diagram, and superposing the remote sensing image vector diagram on the preprocessed high-resolution remote sensing satellite image of the research area; s4, finely classifying crop species by adopting an object-oriented algorithm based on a result vector diagram of the superpixel segmentation algorithm; the method mainly aims at the complex crop planting area in the south, adopts high-resolution remote sensing images, combines unmanned aerial vehicle remote sensing, is based on a superpixel segmentation algorithm, and combines an object-oriented classification method, and can perform fine classification on crop species.
Description
Technical Field
The invention relates to the technical field of remote sensing information classification, in particular to a crop fine classification method based on a high-resolution remote sensing satellite image.
Background
China is a big agricultural country, and agricultural problems are always the focus of common attention of governments and people. Therefore, modern agricultural production in China gradually develops towards intensification and precision. With the change of agriculture, the demand for spatial information, especially for large-scale, dynamic, continuous and fast crop information, in the agricultural production process is becoming more important. Particularly for southern agriculture, the plot is small and complex, the variety is large, and the fast and accurate acquisition of the species crop information is relatively difficult. Therefore, the method can timely acquire the information of spatial distribution of crops, species information, crop growth vigor, crop yield, crop disasters and the like, and has important significance for realizing scientific management and crop yield increase, assisting governments in macroscopically mastering grain production and regulating agricultural product trade.
With the continuous development of aerospace industry in China, the quality, resolution and application of remote sensing satellite images in China are greatly improved. In particular, in recent years, the country opens a commercial remote sensing satellite with 0.5 m spatial resolution, so that the remote sensing satellite in China is unprecedentedly developed. More and more remote sensing satellites are lifted off one after another, and mass remote sensing data are widely applied to the ground.
Based on agricultural requirements, remote sensing has the advantages of large coverage area, high efficiency, timeliness, time saving and labor saving, can provide timely and accurate farmland information for agricultural departments, and is increasingly applied to agricultural production and management. Currently, remote sensing has become an important means for obtaining crop information in precision agriculture. More and more achievements are achieved in the fields of crop classification, crop growth analysis, crop phenological monitoring, crop yield estimation, agricultural disasters and the like. The fine classification of the species of the crops is a prerequisite for knowing the spatial distribution of agricultural crop planting, calculating the planting area of the crops and estimating the yield of the crops. Species type information of crops is rapidly and accurately obtained through remote sensing, and the method has important significance for perfecting a crop area monitoring method, carrying out remote sensing evaluation on crop production level and the like.
At present, the crop classification method based on remote sensing data is more, and the traditional remote sensing image classification method mainly comprises supervised classification and unsupervised classification. The supervised classification generally includes a minimum distance method, a parallel pipeline method, a maximum likelihood method, and the like, and the unsupervised classification includes a K-Means method and an ISODATA method. However, the methods classify the ground features based on the spectral information difference of the ground features in the remote sensing image, the algorithm ignores the texture information and the geometric information of the ground features, and the methods are difficult to distinguish between 'same-object different-spectrum' and 'same-spectrum foreign objects', so the classification accuracy is relatively low. With the development of deep learning, the remote sensing image classification is carried out by utilizing a deep learning algorithm, the effect is relatively high, but mass sample data is difficult to manufacture, especially a refined classification sample data set of species. Meanwhile, the deep learning algorithm has poor applicability, the classification results of the algorithm models have larger difference due to images in the same region at different time phases, and the algorithm is not suitable for rapidly and accurately extracting crop species information. Therefore, aiming at southern agriculture, how to rapidly and effectively perform fine classification of crop species based on high-resolution remote sensing satellite images becomes an urgent problem to be solved.
Disclosure of Invention
The invention aims to provide a crop fine classification method based on a high-resolution remote sensing satellite image, which mainly aims at the complex crop planting area in the south, adopts the high-resolution remote sensing image, combines unmanned aerial vehicle remote sensing, is based on a superpixel segmentation algorithm, is combined with an object-oriented classification method, can perform crop species fine classification, obtains a crop fine classification result, and can provide data support for agricultural planning, management, yield estimation, disaster control and the like of relevant departments.
In order to achieve the purpose, the invention adopts the following technical scheme:
a crop fine classification method based on a high-resolution remote sensing satellite image comprises the following steps:
s1, collecting high-resolution remote sensing satellite images and field sample data in a research area;
s2, preprocessing the high-resolution remote sensing satellite image and the field sample data respectively;
s3, processing the preprocessed high-resolution remote sensing satellite image by adopting a superpixel segmentation algorithm, acquiring a segmented remote sensing image vector diagram, and superposing the remote sensing image vector diagram on the preprocessed high-resolution remote sensing satellite image of the research area;
and S4, finely classifying the crop species by adopting an object-oriented algorithm based on a result vector diagram of the superpixel segmentation algorithm.
Preferably, in step S1, the data of the high-resolution remote sensing satellite image requires that the cloud amount is lower than 15%, and the spatial resolution is not lower than 1 meter; and the field sample data adopts an unmanned aerial vehicle to carry out average area image acquisition, and then is labeled in the field.
Preferably, the difference between the acquisition time of the solid sample data and the acquisition time of the high-resolution remote sensing satellite image in the step S1 is not more than 5 days.
Preferably, the preprocessing procedure of step S2 is:
s21, carrying out remote sensing pretreatment of radiometric calibration, geometric correction, atmospheric correction, image fusion and region-of-interest cutting on the collected high-resolution remote sensing satellite image to obtain a remote sensing image of a research area;
s22, acquiring specific crop species information of each area according to the high-definition images of the unmanned aerial vehicle acquired on the spot;
and S23, comparing the remote sensing images obtained in the step S21 according to the specific crop species information of each region, and manually drawing a sample vector data set to be used as a training sample of an object-oriented algorithm.
Preferably, the superpixel segmentation algorithm in step S3 adopts an SLIC algorithm, the SLIC algorithm converts the color image into a 5-dimensional feature vector in a CIELAB color space and XY coordinates, constructs a distance metric standard for the 5-dimensional feature vector, and performs local clustering on the image pixels.
Preferably, the specific process of step S4 is:
s41, adding geographic coordinate information to the result vector diagram based on the superpixel segmentation algorithm by using a GDAL toolkit, and enabling the result vector diagram to be completely fit with the remote sensing image of the research area;
s42, loading and viewing remote sensing images of the research area, vector results of the super-pixel segmentation algorithm and vector data of the field sampling by adopting eCoginization Developer software;
s43, performing multi-scale segmentation by adopting Yikang software and combining the vector diagram according to the vector diagram;
s44, loading sample vector data after segmentation, and adding a new algorithm in an easily-healthy Process Tree interface: converting the sample vector into a sample pattern type;
s45, after the sample vector is converted into a sample pattern spot, a new algorithm is continuously added: transforming the sample image objects to samples, and converting the sample image spots into an available sample data set;
s46, selecting an algorithm for model training, and adding a new algorithm in a Process Tree interface: the classifier selects the remote sensing image after segmentation, selects a sample data set, sets related parameters and selects a classification algorithm: decisionTree;
s47, adding a classification algorithm to the Process Tree interface as before: and (4) classifying, selecting the segmented image, and performing classification by using model classification selection application to obtain a classified image result picture.
After the technical scheme is adopted, the invention has the following beneficial effects: the method mainly aims at the complex crop planting area in the south, adopts a high-resolution remote sensing image, combines unmanned aerial vehicle remote sensing, is based on a superpixel segmentation algorithm, and combines an object-oriented classification method, so that crop species can be finely classified, a crop fine classification result is obtained, and data support can be provided for relevant departments for agricultural planning, management, yield estimation, disaster treatment and the like; the SLIC algorithm adopted by the invention can generate compact and approximately uniform superpixels, has higher comprehensive evaluation in the aspects of operation speed, object contour maintenance and superpixel shape, and meets the expected segmentation effect of people; compared with the northern large-area crop species classification, under the condition that the southern crop species are relatively complex, the actual overall classification accuracy of the method can reach more than 80%, and the fine classification accuracy of the method is higher.
Drawings
FIG. 1 is a remote sensing image of a study area after pretreatment according to the present invention;
FIG. 2 is a schematic diagram of the present invention in which a remote sensing image vector diagram is superimposed on an image of a study area;
FIG. 3 is a diagram illustrating specific segmentation parameters and threshold settings in step S43 according to the present invention;
FIG. 4 is a schematic diagram of adding a new algorithm in step S44 according to the present invention;
FIG. 5 is a schematic diagram of adding a new algorithm in step S45 according to the present invention;
FIG. 6 is a schematic diagram of adding a new algorithm in step S46 according to the present invention;
FIG. 7 is a schematic diagram of adding a new algorithm in step S47 according to the present invention;
FIG. 8 is a diagram showing the results of fine classification of crops according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
As shown in fig. 1 to 8, a crop fine classification method based on high-resolution remote sensing satellite images includes the following steps:
s1, collecting high-resolution remote sensing satellite images and field sample data in a research area;
in the step S1, the data of the high-resolution remote sensing satellite image requires that the cloud amount is lower than 15%, the spatial resolution is not lower than 1 meter, historical data is customized or purchased according to the requirement for acquisition of the high-resolution remote sensing satellite image, and the coverage area of the high-resolution remote sensing satellite image is mainly a complex crop planting area in a certain city in the south; the field sample data adopts an unmanned aerial vehicle to carry out average area image acquisition, and then is labeled in the field, so that the accuracy and precision of the algorithm model training stage are facilitated; the difference between the acquisition time of the solid sample data and the acquisition time of the high-resolution remote sensing satellite image in the step S1 is not more than 5 days;
s2, preprocessing the high-resolution remote sensing satellite image and the on-site sample data respectively;
the preprocessing procedure of step S2 is:
s21, carrying out remote sensing preprocessing of radiometric calibration, geometric correction, atmospheric correction, image fusion and region-of-interest cutting on the acquired high-resolution remote sensing satellite image to obtain a remote sensing image of a research area, as shown in figure 1;
s22, acquiring specific crop species information of each area according to the high-definition images of the unmanned aerial vehicle acquired on the spot;
s23, comparing the remote sensing images obtained in the step S21 according to the specific crop species information of each region, and manually drawing a sample vector data set to be used as a training sample of an object-oriented algorithm;
s3, processing the preprocessed high-resolution remote sensing satellite image by adopting a superpixel segmentation algorithm, acquiring a segmented remote sensing image vector diagram, and superposing the remote sensing image vector diagram on the preprocessed high-resolution remote sensing satellite image in the research area, as shown in FIG. 2;
the superpixel segmentation algorithm in the step S3 adopts an SLIC algorithm, the SLIC algorithm converts a color image into a 5-dimensional feature vector under a CIELAB color space and XY coordinates, a distance measurement standard is constructed for the 5-dimensional feature vector, and local clustering is performed on image pixels;
s4, finely classifying crop species by adopting an object-oriented algorithm based on a result vector diagram of the superpixel segmentation algorithm;
the specific process of step S4 is:
s41, adding geographic coordinate information to the result vector diagram based on the superpixel segmentation algorithm by using a GDAL toolkit, and enabling the result vector diagram to be completely fit with the remote sensing image of the research area;
s42, loading and viewing remote sensing images of the research area, vector results of the super-pixel segmentation algorithm and vector data of the field sampling by adopting eCoginization Developer software;
s43, performing multi-scale segmentation by using Yikang software according to the vector diagram and combining the vector diagram, wherein specific segmentation parameters and threshold values are shown in figure 3;
s44, loading sample vector data after segmentation, and adding a new algorithm in an easily-healthy Process Tree interface: an assign class by the synthetic layer, as shown in FIG. 4 specifically, converting the sample vector into a sample pattern type;
s45, after the sample vector is converted into a sample pattern spot, a new algorithm is continuously added: transforming the sample blobs into an available sample data set, as shown in fig. 5;
s46, selecting an algorithm to train the model, and adding a new algorithm in a Process Tree interface: the classifier selects the remote sensing image after segmentation, selects a sample data set, sets related parameters and selects a classification algorithm: the DecisionTree, the specific parameter setting is shown in fig. 6;
s47, adding a classification algorithm to the Process Tree interface as before: and (3) classifying, selecting the segmented image, classifying and selecting the model, wherein the specific content is shown in figure 7, and then performing classification to obtain a classified image result chart.
And (4) analyzing results:
through the steps, a crop species classification result graph in a research area can be obtained, as shown in fig. 8, through on-site comparison, the classification accuracy is relatively high, and the overall accuracy can reach over 80%. Compared with the northern large-area crop species classification, under the condition that southern crop species are relatively complex, the actual overall classification accuracy can reach more than 80%, and the fine classification accuracy of the method is high. If the crop fine classification method based on the high-resolution remote sensing satellite image is applied to the northern crop area, the classification accuracy is further improved mainly because the southern plot is small, the planting varieties are various, and the crop cycle time is short, so that the crop species classification accuracy of the embodiment is reduced.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (6)
1. A crop fine classification method based on a high-resolution remote sensing satellite image is characterized by comprising the following steps:
s1, collecting high-resolution remote sensing satellite images and field sample data in a research area;
s2, preprocessing the high-resolution remote sensing satellite image and the field sample data respectively;
s3, processing the preprocessed high-resolution remote sensing satellite image by adopting a superpixel segmentation algorithm, acquiring a segmented remote sensing image vector diagram, and superposing the remote sensing image vector diagram on the preprocessed high-resolution remote sensing satellite image of the research area;
and S4, finely classifying the crop species by adopting an object-oriented algorithm based on a result vector diagram of the superpixel segmentation algorithm.
2. The method for finely classifying crops based on the high-resolution remote sensing satellite image as claimed in claim 1, wherein: in the step S1, the data of the high-resolution remote sensing satellite image requires that the cloud amount is lower than 15%, and the spatial resolution is not lower than 1 meter; and the field sample data adopts an unmanned aerial vehicle to carry out average area image acquisition, and then is labeled in the field.
3. The method for finely classifying crops based on the high-resolution remote sensing satellite image as claimed in claim 2, wherein: and in the step S1, the difference between the acquisition time of the solid sample data and the acquisition time of the high-resolution remote sensing satellite image is not more than 5 days.
4. The method for finely classifying crops based on the high-resolution remote sensing satellite images as claimed in claim 1, wherein the preprocessing process of the step S2 is as follows:
s21, carrying out remote sensing pretreatment of radiometric calibration, geometric correction, atmospheric correction, image fusion and region-of-interest cutting on the collected high-resolution remote sensing satellite image to obtain a remote sensing image of a research area;
s22, acquiring specific crop species information of each area according to the high-definition images of the unmanned aerial vehicle acquired on the spot;
and S23, comparing the remote sensing images obtained in the step S21 according to the specific crop species information of each region, and manually drawing a sample vector data set to be used as a training sample of an object-oriented algorithm.
5. The method for finely classifying crops based on the high-resolution remote sensing satellite image as claimed in claim 1, wherein: the superpixel segmentation algorithm in the step S3 adopts an SLIC algorithm, the SLIC algorithm converts the color image into a 5-dimensional feature vector in a CIELAB color space and XY coordinates, then constructs a distance metric standard for the 5-dimensional feature vector, and performs local clustering on image pixels.
6. The method for finely classifying crops based on the high-resolution remote sensing satellite images as claimed in claim 4, wherein the specific process of step S4 is as follows:
s41, adding geographic coordinate information to the result vector diagram based on the superpixel segmentation algorithm by using a GDAL toolkit, and enabling the result vector diagram to be completely fit with the remote sensing image of the research area;
s42, loading and viewing remote sensing images of the research area, vector results of the super-pixel segmentation algorithm and vector data of the field sampling by adopting eCoginization Developer software;
s43, performing multi-scale segmentation by adopting Yikang software and combining the vector diagram according to the vector diagram;
s44, loading sample vector data after segmentation, and adding a new algorithm in an easily-healthy Process Tree interface: converting the sample vector into a sample pattern type;
s45, after the sample vector is converted into a sample pattern spot, a new algorithm is continuously added: transforming the sample image objects to samples, and converting the sample image spots into an available sample data set;
s46, selecting an algorithm for model training, and adding a new algorithm in a Process Tree interface: the classifier selects the remote sensing image after segmentation, selects a sample data set, sets related parameters and selects a classification algorithm: a precision Tree;
s47, adding a classification algorithm to the Process Tree interface as before: and (4) classifying, selecting the segmented image, and performing classification by using model classification selection application to obtain a classified image result picture.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210438771.3A CN114881127A (en) | 2022-04-21 | 2022-04-21 | Crop fine classification method based on high-resolution remote sensing satellite image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210438771.3A CN114881127A (en) | 2022-04-21 | 2022-04-21 | Crop fine classification method based on high-resolution remote sensing satellite image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114881127A true CN114881127A (en) | 2022-08-09 |
Family
ID=82671789
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210438771.3A Pending CN114881127A (en) | 2022-04-21 | 2022-04-21 | Crop fine classification method based on high-resolution remote sensing satellite image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114881127A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116030352A (en) * | 2023-03-29 | 2023-04-28 | 山东锋士信息技术有限公司 | Long-time-sequence land utilization classification method integrating multi-scale segmentation and super-pixel segmentation |
| CN116645618A (en) * | 2023-06-05 | 2023-08-25 | 广东省农业科学院设施农业研究所 | Agricultural data processing method, system and storage medium |
-
2022
- 2022-04-21 CN CN202210438771.3A patent/CN114881127A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116030352A (en) * | 2023-03-29 | 2023-04-28 | 山东锋士信息技术有限公司 | Long-time-sequence land utilization classification method integrating multi-scale segmentation and super-pixel segmentation |
| CN116645618A (en) * | 2023-06-05 | 2023-08-25 | 广东省农业科学院设施农业研究所 | Agricultural data processing method, system and storage medium |
| CN116645618B (en) * | 2023-06-05 | 2023-12-08 | 广东省农业科学院设施农业研究所 | Agricultural data processing method, system and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111104858B (en) | Large-scale crop phenology extraction method based on morphological model method | |
| CN110263717B (en) | Method for determining land utilization category of street view image | |
| CN111368736B (en) | Rice refined estimation method based on SAR and optical remote sensing data | |
| CN111738175A (en) | Agricultural drought monitoring system based on remote sensing image and convolutional neural network | |
| CN114881127A (en) | Crop fine classification method based on high-resolution remote sensing satellite image | |
| CN111339921A (en) | Insect disease detection unmanned aerial vehicle based on lightweight convolutional neural network and detection method | |
| CN110766690B (en) | Wheat ear detection and counting method based on deep learning point supervision thought | |
| CN112906666A (en) | Remote sensing identification method for agricultural planting structure | |
| CN102855485B (en) | The automatic testing method of one grow wheat heading | |
| CN114612794A (en) | Remote sensing identification method for land covering and planting structure in finely-divided agricultural area | |
| CN109671038B (en) | Relative radiation correction method based on pseudo-invariant feature point classification layering | |
| CN117456358A (en) | Method for detecting plant diseases and insect pests based on YOLOv5 neural network | |
| CN116011881A (en) | Black soil farmland productivity evaluation method and system based on pixel scale | |
| CN110909821B (en) | Method for carrying out high-space-time resolution vegetation index data fusion based on crop reference curve | |
| CN116897668A (en) | Electric-drive crop sowing and fertilizing control method and system | |
| CN116863341A (en) | Crop classification and identification method and system based on time sequence satellite remote sensing image | |
| CN115147730A (en) | Remote sensing classification method combining full convolution neural network and ensemble learning | |
| CN116664860A (en) | Object-oriented convolutional neural network supported high-resolution image pit water body extraction method | |
| CN111860038B (en) | Crop front end recognition device and method | |
| CN114494911A (en) | Crop classification method based on satellite images and unmanned aerial vehicle images | |
| CN114581429B (en) | Cervical cell detection method and imaging method based on regional relation | |
| CN111582246A (en) | Method and system for estimating grazing rate based on alpine meadow grassland grass yield | |
| CN114119536B (en) | Crop nitrogen content rapid monitoring and diagnosing method based on cloud platform technology | |
| Cui et al. | Paddy Fields Extraction and Spatiotemporal Evolution Analysis Based on Remote Sensing Cloud Platform-A case study of Hubei Province | |
| CN117036963B (en) | Detection and estimation method for typical grassland surface plant quantity |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |