A Crop Classification Method and Device Based on Multispectral Images

Abstract

The present application provides a method and device for classifying crops based on multispectral images, relating to the field of remote sensing image recognition, including: acquiring image data of a study area, preprocessing the image data to obtain first data; calculating an index for the first data according to a preset formula to obtain second data; acquiring third data, analyzing the land use type based on the second data and the third data, and classifying according to a preset threshold to obtain a first classification result; synthesizing the first classification result to obtain a first mask layer, and synthesizing the third data according to the second data to obtain a second mask layer; analyzing the first classification result, the first mask layer and the second mask layer according to the overlay analysis method to obtain a second classification result. The classification accuracy of crops is improved, the classification misclassification probability is reduced, the classification edge is smoothed, and noise points are removed.

Description

Technical Field

[0001] The present invention relates to the field of remote sensing image recognition, and in particular to a method and device for classifying crops based on multispectral images. Background Art

[0002] Among human activities, agricultural activities are closely related to the natural environment, and food production occupies an important position in the entire agricultural field, which is related to the national economy and people's livelihood and the development of the national economy. Northeast China is the most potential commodity grain base and an important main grain production area in China, providing a large amount of high-quality rations, feed grains and industrial grains for the whole country. In recent years, the focus of China's food production has gradually shifted to the northeast direction, which further reflects the importance of Northeast China in the pattern of China's food production. Therefore, studying the pattern of food production and the spatio-temporal distribution of crops is conducive to optimizing the food production structure, realizing the sustainable development of agriculture, and is of great significance to food security. Liaoning is located in a high-latitude region, lacking heat, and the crops can only be harvested once a year, so the congenital conditions for the growth and development of crops are not superior. However, this region is more vast than other regions, with a relatively small population, and the soil texture is black soil, so the yield and quality of food crops are high. Soybeans, rice, corn, and peanuts are the main food crops in the Liaoning region, supplying a large amount of food to the whole country every year.

[0003] In the existing technologies for crop recognition, most classify crops based on image data of a single period, which will lose a large amount of phenological information of crops, so the classification accuracy is not high; or do not use multi-source data for comprehensive extraction of crops, and a single data source will increase the probability of misclassification due to the lack of necessary bands and time phases; moreover, most use supervised classification for classification. After inputting the feature bands and sample points, the threshold cannot be modified manually, and the whole process is automatically completed by the computer; and the post-processing steps of the classification results are not perfect, often resulting in problems such as uneven edges of ground object classification and the existence of noise points. Summary of the Invention

[0004] The present invention provides a method and device for classifying crops based on multispectral images, aiming to solve the problems of low classification accuracy, high probability of misclassification, inability to manually modify the classification threshold, uneven classification edges, and the existence of noise points in the above-mentioned existing technologies.

[0005] To achieve the above object, the present application adopts the following technical solutions, including:

[0006] Obtain the image data of the study area, preprocess the image data to obtain the first data, and the image data includes Sentinel-2 multispectral image data and Landsat multispectral image data;

[0007] Perform an exponential calculation on the first data according to a preset formula to obtain second data;

[0008] Obtain the land use type map of the study area. The land use type map is the third data. Analyze the land use type based on the second data and the third data, and classify it according to a preset threshold to obtain the first classification result;

[0009] Synthesize the first classification result to obtain a crop mask layer. The crop mask layer is the first mask layer. Synthesize the third data according to the second data to obtain an impervious mask layer. The impervious mask layer is the second mask layer;

[0010] Analyze the first classification result, the first mask layer, and the second mask layer according to the overlay analysis method to obtain the second classification result.

[0011] Preferably, the obtaining of the image data of the study area and the preprocessing of the image data to obtain the first data include:

[0012] Perform atmospheric correction, coordinate system conversion, and band synthesis processing on the Sentinel-2 multispectral image data to obtain the first image data;

[0013] When the phase of the Sentinel-2 multispectral image data is missing, perform radiometric calibration, atmospheric correction, coordinate system conversion, and band synthesis processing on the Landsat multispectral image data to obtain the second image data, and summarize the first image data and the second image data to obtain the first data.

[0014] Preferably, the performing of an exponential calculation on the first data according to a preset formula to obtain second data includes:

[0015] Calculate the first data according to the formula EVI = 2.5 * (NIR - RED) / (NIR + 6 * RED - 7.5 * BLUE + 1) to obtain the first index, calculate the first data according to the formula LSWI = (NIR - SWIR1) / (NIR + SWIR1) to obtain the second index, calculate the first data according to the formula NDBI = (SWIR1 - NIR) / (SWIR1 + NIR) (RED < NIR < SWIR1) to obtain the third index, where EVI is the enhanced vegetation index, LSWI is the land surface water index, NDBI is the normalized difference built-up index, BLUE is the blue band, RED is the red band, NIR is the near-infrared band, and SWIR1 is the shortwave infrared band 1;

[0016] Summarize the first index, the second index, and the third index to obtain the second data.

[0017] Preferably, the acquiring of the third data, analyzing the land use type according to the second data and the third data, and classifying the land use type according to a preset threshold value to obtain a first classification result includes:

[0018] Determine whether the land type of the study area is a water body based on the third data and the second data; if not, classify the land type using a decision tree method based on the first index and the second index to obtain the first classification result.

[0019] Preferably, analyzing the first classification result, the first mask layer and the second mask layer according to the superposition analysis method to obtain the second classification result includes:

[0020] Determine whether the pixel point of the image data of the study area falls into the first mask layer area and does not fall into the second mask layer area;

[0021] If yes, fill the pixel point according to the first classification result to obtain a first result; if no, the pixel point is a non-crop area and a second result is obtained; the first result and the second result are aggregated to obtain the second classification result.

[0022] A crop classification device based on multispectral images, comprising:

[0023] Image data preprocessing module: used to obtain image data of the study area, preprocess the image data to obtain first data, the image data includes Sentinel-2 multispectral image data and Landsat multispectral image data;

[0024] Index calculation module: used to perform index calculation on the first data according to a preset formula to obtain second data;

[0025] The first crop classification module is used to obtain a land use type map of the study area, which is the third data. The land use type is analyzed according to the second data and the third data, and classified according to a preset threshold value to obtain a first classification result.

[0026] A mask layer synthesis module: used for synthesizing the first classification result to obtain a crop mask layer, the crop mask layer is a first mask layer, synthesizing the third data according to the second data to obtain a water-tight mask layer, the water-tight mask layer is a second mask layer;

[0027] The crop second classification module is used to analyze the first classification result, the first mask layer and the second mask layer according to the overlay analysis method to obtain a second classification result.

[0028] Preferably, the image data preprocessing module includes:

[0029] The first image processing module: used to perform atmospheric correction, coordinate system conversion, and band synthesis processing on the Sentinel-2 multispectral image data to obtain the first image data;

[0030] The second image processing module: used to perform radiometric calibration, atmospheric correction, coordinate system conversion, and band synthesis processing on the Landsat multispectral image data when the phase of the Sentinel-2 multispectral image data is missing, to obtain the second image data, and summarize the first image data and the second image data to obtain the first data.

[0031] Preferably, the index calculation module includes:

[0032] The sub-index calculation module: used to calculate the first index from the first data according to the formula EVI = 2.5 * (NIR - RED) / (NIR + 6 * RED - 7.5 * BLUE + 1), calculate the second index from the first data according to the formula LSWI = (NIR - SWIR1) / (NIR + SWIR1), and calculate the third index from the first data according to the formula NDBI = (SWIR1 - NIR) / (SWIR1 + NIR) (RED < NIR < SWIR1), where EVI is the Enhanced Vegetation Index, LSWI is the Land Surface Water Index, NDBI is the Normalized Difference Built-up Index, BLUE is the blue light band, RED is the red light band, NIR is the near-infrared band, and SWIR1 is the shortwave infrared band 1;

[0033] The sub-index summarization module: used to summarize the first index, the second index, and the third index to obtain the second data.

[0034] A crop classification device based on multispectral images includes a memory and a processor, and the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement a crop classification method based on multispectral images as described in any one of the above.

[0035] A computer-readable storage medium storing a computer program, and the computer program, when executed by a computer, implements a crop classification method based on multispectral images as described in any one of the above.

[0036] The present invention has the following beneficial effects:

[0037] This technical solution comprehensively considers the physiological periods of major food crops in data acquisition, uses time-series data as the characteristics of various food crops for crop classification, solves the problem in the prior art that crop classification based on image data in a single period results in the loss of a large amount of phenological information of crops, and improves the classification accuracy; in this technical solution, Landsat data and 2017 land cover type data are supplemented as auxiliary data. Landsat is used to supplement missing temporal data, and the 2017 land cover type data will participate in mask establishment as prior knowledge. Multi-source data is used for comprehensive extraction of crops, avoiding the problem that the misclassification probability increases due to the lack of necessary bands and temporal phases in a single data source, and improving the accuracy and precision of crop classification based on data; this technical solution uses a decision tree for hierarchical classification of major food crops, and the threshold conditions for crop classification can be manually modified. Since the threshold is manually set, manual intervention can be quickly achieved to modify the classification model, and it can be manually modified according to the actual situation of the field scene, avoiding inaccurate classification results caused by regional differences caused by unified computer settings, and improving the overall accuracy of crop classification; this technical solution uses median filtering to filter the mask layer, crop layer, and classification result file, eliminating holes and noise, making the land cover classification edge smooth, and improving the accuracy of the overall classification result; after obtaining the first classification result, a composite mask is further used to process the first classification result to obtain the final classification result. This process removes some non-crop areas and some impurities that affect the classification of major crops, improving the accuracy of major crops in the final classification result, reducing interference, and improving precision. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a flowchart of a method for classifying crops based on multi-spectral images implemented in an embodiment of the present invention.

[0039] Figure 2 It is a flowchart of a method for preprocessing image data and calculating an index to obtain second data implemented in an embodiment of the present invention.

[0040] Figure 3 It is a schematic structural diagram of an apparatus for classifying crops based on multi-spectral images implemented in an embodiment of the present invention.

[0041] Figure 4 It is a schematic structural diagram of a preprocessing module 10 for image data in an apparatus for classifying crops based on multi-spectral images implemented in an embodiment of the present invention.

[0042] Figure 5 It is a schematic structural diagram of an index calculation module 20 in an apparatus for classifying crops based on multi-spectral images implemented in an embodiment of the present invention.

[0043] Figure 6 Schematic diagram of an electronic device for implementing a crop classification device based on multispectral images according to an embodiment of the present invention Detailed implementation manners

[0044] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0045] The terms "first", "second", etc. in the claims and the description of the present application are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such terms can be interchanged under appropriate circumstances. This is only a way of distinguishing when describing objects with the same attributes in the embodiments of the present application. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion, so that a process, method, system, product or device including a series of units does not have to be limited to those units, but may include other units not clearly listed or inherent to these processes, methods, products or devices.

[0046] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used in the description of this application herein are only for the purpose of describing specific embodiments, and are not intended to limit this application.

[0047] Embodiment 1

[0048] As Figure 1 shown, a crop classification method based on multispectral images includes the following steps:

[0049] S11. Obtain image data of the study area, preprocess the image data to obtain first data, where the image data includes Sentinel-2 multispectral image data and Landsat multispectral image data;

[0050] S12. Calculate an index for the first data according to a preset formula to obtain second data;

[0051] S13. Obtain a land use type map of the study area, where the land use type map is third data, analyze the land use type according to the second data and the third data, and classify according to a preset threshold to obtain a first classification result;

[0052] S14. Synthesize the first classification result to obtain a crop mask layer, where the crop mask layer is the first mask layer. Synthesize the third data according to the second data to obtain an impervious mask layer, where the impervious mask layer is the second mask layer;

[0053] S15. Analyze the first classification result, the first mask layer, and the second mask layer according to the overlay analysis method to obtain a second classification result.

[0054] In this embodiment, first, image data of different periods in the study area is obtained. In this embodiment, the study area is the Liaoning Province area, and Sentinel-2 multispectral image data is mainly used. When there is a lack of time phase, Landsat multispectral image data is used for supplementation, and operations such as atmospheric correction, coordinate system conversion, and band synthesis are performed on the image data. The processed image data is summarized to obtain the first data. Then, according to the calculation formulas of the Enhanced Vegetation Index (EVI), Land Surface Water Index (LSWI), and Normalized Difference Built-up Index (NDBI), the relevant indices of the image data in the first data are calculated, and the calculation results are summarized to obtain the second data. Then, according to the 10-meter resolution land cover type map made by Tsinghua University using Sentinel data, the land use type map of the study area is obtained, that is, the third data, where the corresponding code for water bodies is 60, and the code for impervious surfaces is 80. Then, the decision tree hierarchical classification method is used to classify the crop types in the study area. The land use type data and the calculated remote sensing index data, that is, the second data and the third data, are input, and the following analysis and judgment are carried out for crop classification: (1) Judge whether the land use type is a water body. If it is not a water body, enter crop classification; (2) Judge whether there is LSWI < 0 in April or May, and EVI > 0.5 in July and EVI > 0.2 in September. If so, it is rice; (3) For the remaining area, judge whether the EVI in September minus the EVI in July is greater than 0.1. If so, judge it as peanut; (4) For the remaining area, judge whether there is EVI > 0.62 in July. If so, judge it as corn; (5) For the remaining area, judge whether there is 0.15 < EVI < 0.34 in July. If so, judge it as soybean, and the remaining part is classified as non-crop.Summarize the classification results of the above five categories to obtain the first classification result. Then, create a crop mask and an impervious surface mask, namely the first mask layer and the second mask layer. For the crop mask, use the preliminary classification result, i.e., the first classification result, to synthesize the crop mask, and use median filtering with a 5*5 operator to filter this mask. For the impervious surface mask, use the NDBI image in the second data. When the NDBI image values of a certain pixel point in July and October both satisfy NDBI > -0.25, consider this pixel point as an impervious surface. Take the union of the calculation result of this step and the area with a median value of 80 in the land use type map in the third data to synthesize the impervious surface mask, and use median filtering with a 5*5 operator for filtering. Then, perform the last step of crop type classification. At this step, there are three pieces of data, namely: the classification result data (rice, peanut, corn, soybean), i.e., the first classification data, the crop mask layer, i.e., the first mask layer, and the impervious mask layer, i.e., the second mask layer. Perform overlay analysis on the above data to determine whether a certain pixel point in the image data of the study area meets the following conditions: falling within the area of the crop mask layer and not falling within the area of the impervious mask layer. If it meets the conditions, fill in the classification result data at this pixel point; if it does not meet the conditions, consider this pixel point as a non-crop point. Perform median filtering on the above overlay analysis result with a 3*3 operator to obtain the classification result of the four main food crops, i.e., the second classification result, and output it. The beneficial effects of this embodiment are as follows: In the data acquisition of this technical solution, the physiological periods of the main food crops are comprehensively considered, and time-series data is used as the characteristics of various food crops for crop classification, solving the problem in the prior art that crop classification based on image data of a single period will result in the loss of a large amount of phenological information of crops, improving the classification accuracy. This technical solution uses a decision tree for hierarchical classification of the main food crops, and the threshold conditions for crop classification can be manually modified. Since the threshold is manually set, it can quickly achieve manual intervention and modify the classification model, and can be manually modified according to the actual situation of the field scene, avoiding inaccurate classification results caused by regional differences caused by computer unified settings, and improving the overall accuracy of crop classification. This technical solution uses median filtering to filter the mask layer, crop layer, and classification result file, eliminating holes and noise, making the classification edges of ground objects smooth, and improving the overall classification result accuracy. After obtaining the first classification result, synthesize the mask and process the first classification result again to obtain the final classification result. This process removes some non-crop areas and some impurities that affect the classification of the main crops, improving the accuracy of the main crops in the final classification result, reducing interference, and improving the accuracy.

[0055] Example 2

[0056] As Figure 2As shown, a method for preprocessing image data and calculating an index to obtain second data includes the following steps:

[0057] S21. Perform atmospheric correction, coordinate system conversion, and band synthesis processing on the Sentinel-2 multispectral image data to obtain first image data;

[0058] S22. When the phase of the Sentinel-2 multispectral image data is missing, perform radiometric calibration, atmospheric correction, coordinate system conversion, and band synthesis processing on the Landsat multispectral image data to obtain second image data, and summarize the first image data and the second image data to obtain the first data;

[0059] S23. Calculate the first index by using the formula EVI = 2.5 * (NIR - RED) / (NIR + 6 * RED - 7.5 * BLUE + 1) for the first data, calculate the second index by using the formula LSWI = (NIR - SWIR1) / (NIR + SWIR1) for the first data, and calculate the third index by using the formula NDBI = (SWIR1 - NIR) / (SWIR1 + NIR) (RED < NIR < SWIR1) for the first data, where EVI is the enhanced vegetation index, LSWI is the land surface water index, NDBI is the normalized difference built-up index, BLUE is the blue light band, RED is the red light band, NIR is the near-infrared band, and SWIR1 is the shortwave infrared band 1;

[0060] S24. Summarize the first index, the second index, and the third index to obtain the second data.

[0061] In this embodiment, the selection of data refers to the selection of image data in the study area. Here, one scene of multispectral data for each of April, May, July, September, and October is selected. Sentinel-2 multispectral image data is preferred. If there is a lack of time phase, Landsat multispectral image data is used for supplementation. For Sentinel-2 multispectral image data, the tool Sen2Cor is needed for atmospheric correction, and the tool SNAP is used for coordinate system conversion and band synthesis. After the above operations, the processed Sentinel-2 multispectral image data, that is, the first image data, is obtained. Then, when there is a lack of time phase, Landsat multispectral image data is used for supplementation. For Landsat multispectral data, radiometric calibration is required, and the flaash atmospheric correction method is used for correction. Then, coordinate system conversion and band synthesis are carried out. After the above operations, the processed Landsat multispectral image data, that is, the second image data, is obtained. The first image data and the second image data are summarized to obtain the first data. Then, according to the first data, the corresponding indices of the study area are calculated. Here, 3 indices need to be calculated, namely the Enhanced Vegetation Index (EVI), the Land Surface Water Index (LSWI), and the Normalized Difference Built-up Index (NDBI). The 3 indices have their respective functions. Among them, EVI is used to judge the growth of crops, LSWI is used to extract flood signals, and NDBI is used to obtain non-permeable surfaces. The calculation methods of the three indices are as follows: EVI = 2.5 * (NIR - RED) / (NIR + 6 * RED - 7.5 * BLUE + 1); LSWI = (NIR - SWIR1) / (NIR + SWIR1); NDBI = (SWIR1 - NIR) / (SWIR1 + NIR) (when RED < NIR < SWIR1 is satisfied, otherwise the value is 0), where EVI is the Enhanced Vegetation Index, LSWI is the Land Surface Water Index, NDBI is the Normalized Difference Built-up Index, BLUE is the blue light band, RED is the red light band, NIR is the near-infrared band, and SWIR1 is the short-wave infrared band 1. After calculating the numerical values of the three corresponding indices respectively, they are summarized to obtain the comprehensive index data of the study area, that is, the second data.The beneficial effects of this embodiment are as follows: In this technical solution, Landsat data and 2017 land cover type data are supplemented as auxiliary data. Landsat is used to supplement the missing temporal data, and the 2017 land cover type data will participate in the establishment of the mask as prior knowledge. Multi-source data are used for comprehensive extraction of crops, avoiding the problem of increased misclassification probability caused by a single data source lacking necessary bands and temporal phases, improving the accuracy and precision of crop classification based on data. Moreover, the Enhanced Vegetation Index (EVI), Land Surface Water Index (LSWI), and Normalized Difference Built-up Index (NDBI) of the study area are calculated respectively, and the crops in the study area are classified based on these indices, which can classify according to the characteristics of different crops at different times, making the classification results more accurate.

[0062] Embodiment 3

[0063] As Figure 3 shown, a crop classification device based on multi-spectral images includes:

[0064] An image data preprocessing module 10: used to obtain the image data of the study area, preprocess the image data to obtain the first data, and the image data includes Sentinel-2 multi-spectral image data and Landsat multi-spectral image data;

[0065] An index calculation module 20: used to calculate the index of the first data according to a preset formula to obtain the second data;

[0066] A first crop classification module 30: used to obtain the land use type map of the study area, the land use type map is the third data, analyze the land use type according to the second data and the third data, and classify according to a preset threshold to obtain the first classification result;

[0067] A mask layer synthesis module 40: used to synthesize the first classification result to obtain a crop mask layer, the crop mask layer is the first mask layer, synthesize the third data according to the second data to obtain an impervious mask layer, and the impervious mask layer is the second mask layer;

[0068] A second crop classification module 50: used to analyze the first classification result, the first mask layer and the second mask layer according to the overlay analysis method to obtain the second classification result.

[0069] One implementation of the above device is as follows: In the image data preprocessing module 10, the image data of the study area is acquired, and the image data is preprocessed to obtain the first data. The image data includes Sentinel-2 multispectral image data and Landsat multispectral image data. In the index calculation module 20, the first data is calculated according to a preset formula to obtain the second data. In the first crop classification module 30, the third data is acquired, the land use type is analyzed according to the second data and the third data, and classification is performed according to a preset threshold to obtain the first classification result. In the mask layer synthesis module 40, the first classification result is synthesized to obtain the first mask layer, and the third data is synthesized according to the second data to obtain the second mask layer. In the second crop classification module 50, the first classification result, the first mask layer, and the second mask layer are analyzed according to the overlay analysis method to obtain the second classification result.

[0070] Example 4

[0071] As Figure 4 shown, the image data preprocessing module 10 in a crop classification device based on multispectral images includes:

[0072] The first image processing module 11: is used to perform atmospheric correction, coordinate system conversion, and band synthesis processing on the Sentinel-2 multispectral image data to obtain the first image data;

[0073] The second image processing module 12: is used to perform radiometric calibration, atmospheric correction, coordinate system conversion, and band synthesis processing on the Landsat multispectral image data when the phase of the Sentinel-2 multispectral image data is missing, to obtain the second image data, and summarize the first image data and the second image data to obtain the first data.

[0074] One implementation of the above device is as follows: In the first image processing module 11, atmospheric correction, coordinate system conversion, and band synthesis processing are performed on the Sentinel-2 multispectral image data to obtain the first image data. In the second image processing module 12, when the phase of the Sentinel-2 multispectral image data is missing, radiometric calibration, atmospheric correction, coordinate system conversion, and band synthesis processing are performed on the Landsat multispectral image data to obtain the second image data, and the first image data and the second image data are summarized to obtain the first data.

[0075] Example 5

[0076] As Figure 5 shown, the index calculation module 20 in a crop classification device based on multispectral images includes:

[0077] Sub - index calculation module 21: It is used to calculate the first index from the first data according to the formula EVI = 2.5*(NIR - RED) / (NIR + 6*RED - 7.5*BLUE + 1), calculate the second index from the first data according to the formula LSWI = (NIR - SWIR1) / (NIR + SWIR1), and calculate the third index from the first data according to the formula NDBI = (SWIR1 - NIR) / (SWIR1 + NIR) (RED < NIR < SWIR1), where EVI is the Enhanced Vegetation Index, LSWI is the Land Surface Water Index, NDBI is the Normalized Difference Built - up Index, BLUE is the blue light band, RED is the red light band, NIR is the near - infrared band, and SWIR1 is the short - wave infrared band 1;

[0078] Sub - index summarization module 22: It is used to summarize the first index, the second index, and the third index to obtain the second data.

[0079] One implementation of the above - mentioned device is that in the sub - index calculation module 21, the first index is calculated from the first data according to the formula EVI = 2.5*(NIR - RED) / (NIR + 6*RED - 7.5*BLUE + 1), the second index is calculated from the first data according to the formula LSWI = (NIR - SWIR1) / (NIR + SWIR1), and the third index is calculated from the first data according to the formula NDBI = (SWIR1 - NIR) / (SWIR1 + NIR) (RED < NIR < SWIR1), where EVI is the Enhanced Vegetation Index, LSWI is the Land Surface Water Index, NDBI is the Normalized Difference Built - up Index, BLUE is the blue light band, RED is the red light band, NIR is the near - infrared band, and SWIR1 is the short - wave infrared band 1. In the sub - index summarization module 22, the first index, the second index, and the third index are summarized to obtain the second data.

[0080] Example 6

[0081] As Figure 6 shown, an electronic device includes a memory 601 and a processor 602. The memory 601 is used to store one or more computer instructions. Among them, the one or more computer instructions are executed by the processor 602 to implement any one of the above - mentioned methods.

[0082] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working process of the above - described electronic device can refer to the corresponding process in the foregoing method embodiments and will not be elaborated herein.

[0083] A computer-readable storage medium storing a computer program, the computer program causing a computer to implement any one of the methods described above when executed.

[0084] Exemplarily, the computer program may be divided into one or more modules / units, and one or more modules / units are stored in the memory 601 and executed by the processor 602, and the input / output interface transmission of data is completed by the input interface 605 and the output interface 606 to complete the present invention. One or more modules / units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the computer device.

[0085] The computer device may be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The computer device may include, but is not limited to, the memory 601 and the processor 602. Those skilled in the art can understand that this embodiment is merely an example of the computer device and does not constitute a limitation on the computer device. It may include more or fewer components, or combine certain components, or different components. For example, the computer device may further include an input device 607, a network access device, a bus, etc.

[0086] The processor 602 may be a central processing unit (CPU), or may also be other general-purpose processors 602, digital signal processors 602 (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor 602 may be a microprocessor 602 or the processor 602 may also be any conventional processor 602, etc.

[0087] The memory 601 may be an internal storage unit of the computer device, such as the hard disk or memory of the computer device. The memory 601 may also be an external storage device of the computer device, such as a plug-in hard disk equipped on the computer device, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the memory 601 may also include both the internal storage unit and the external storage device of the computer device. The memory 601 is used to store computer programs and other programs and data required by the computer device. The memory 601 may also be used to temporarily store in the outputter 608, and the foregoing storage media include various media such as USB flash drives, mobile hard disks, read-only memory ROM603, random access memory RAM604, diskettes or optical discs that can store program codes.

[0088] The above are only specific embodiments of the present invention, but the technical features of the present invention are not limited thereto. Any changes or modifications made by any person skilled in the art within the scope of the present invention are covered by the patent scope of the present invention.

Claims

1. A crop classification method based on multispectral images, characterized in that, Including: Obtain the image data of the study area, preprocess the image data to obtain the first data, where the image data includes Sentinel-2 multispectral image data and Landsat multispectral image data; Perform exponential calculation on the first data according to a preset formula to obtain the second data; Obtain the land use type map of the study area, where the land use type map is the third data, analyze the land use type according to the second data and the third data, and classify it according to a preset threshold to obtain the first classification result; Synthesize the first classification result to obtain a crop mask layer, where the crop mask layer is the first mask layer, and synthesize the third data according to the second data to obtain an impervious mask layer, where the impervious mask layer is the second mask layer; Analyze the first classification result, the first mask layer and the second mask layer according to the overlay analysis method to obtain the second classification result.

2. The crop classification method based on multi-spectral images according to claim 1, characterized in that, The obtaining of the image data of the study area, preprocessing the image data to obtain the first data, includes: Perform atmospheric correction, coordinate system conversion and band synthesis processing on the Sentinel-2 multispectral image data to obtain the first image data; When the phase of the Sentinel-2 multispectral image data is missing, perform radiometric calibration, atmospheric correction, coordinate system conversion and band synthesis processing on the Landsat multispectral image data to obtain the second image data, and summarize the first image data and the second image data to obtain the first data.

3. The crop classification method based on multispectral images according to claim 2, wherein, The performing of exponential calculation on the first data according to a preset formula to obtain the second data, includes: Calculate the first data according to the formula EVI = 2.5 * (NIR - RED) / (NIR + 6 * RED - 7.5 * BLUE + 1) to obtain the first index, calculate the first data according to the formula LSWI = (NIR - SWIR1) / (NIR + SWIR1) to obtain the second index, calculate the first data according to the formula NDBI = (SWIR1 - NIR) / (SWIR1 + NIR) (RED < NIR < SWIR1) to obtain the third index, where EVI is the enhanced vegetation index, LSWI is the land surface water index, NDBI is the normalized difference built-up index, BLUE is the blue band, RED is the red band, NIR is the near-infrared band, and SWIR1 is the shortwave infrared band 1; Summarize the first index, the second index, and the third index to obtain the second data.

4. A crop classification method based on multi-spectral images according to claim 3, characterized in that The obtaining of the third data, analyzing the land use type according to the second data and the third data, and classifying it according to a preset threshold to obtain the first classification result, includes: Judge whether the land type of the study area is water according to the third data and the second data. If not, classify the land type by the decision tree method according to the first index and the second index to obtain the first classification result.

5. A method for classifying crops based on multi-spectral images according to claim 4, characterized in that, Analyzing the first classification result, the first mask layer, and the second mask layer according to the superposition analysis method to obtain a second classification result, including: Determining whether the pixel points of the image data in the study area fall within the area of the first mask layer and do not fall within the area of the second mask layer; If yes, filling the pixel points according to the first classification result to obtain a first result; if no, the pixel points are non-crop areas to obtain a second result, and summarizing the first result and the second result to obtain the second classification result.

6. A crop classification device based on multispectral images, which is used to implement a crop classification method based on multispectral images as described in claim 1, characterized in that, Including: Image data preprocessing module: used to obtain the image data in the study area, preprocess the image data to obtain first data, and the image data includes Sentinel-2 multispectral image data and Landsat multispectral image data; Index calculation module: used to calculate an index for the first data according to a preset formula to obtain second data; Crop first classification module: used to obtain the land use type map of the study area, the land use type map is third data, analyze the land use type according to the second data and the third data, and classify according to a preset threshold to obtain a first classification result; Mask layer synthesis module: used to synthesize the first classification result to obtain a crop mask layer, the crop mask layer is the first mask layer, synthesize the third data according to the second data to obtain an impervious mask layer, and the impervious mask layer is the second mask layer; Crop second classification module: used to analyze the first classification result, the first mask layer, and the second mask layer according to the superposition analysis method to obtain a second classification result.

7. The crop classification device based on multi-spectral images according to claim 6, characterized in that The image data preprocessing module includes: Image first processing module: used to perform atmospheric correction, coordinate system conversion, and band synthesis processing on the Sentinel-2 multispectral image data to obtain first image data; Image second processing module: used to perform radiometric calibration, atmospheric correction, coordinate system conversion, and band synthesis processing on the Landsat multispectral image data when the phase of the Sentinel-2 multispectral image data is missing to obtain second image data, and summarize the first image data and the second image data to obtain the first data.

8. The crop classification device based on multispectral images according to claim 7, wherein, The index calculation module includes: Sub - index calculation module: It is used to calculate the first index from the first data according to the formula EVI = 2.5*(NIR - RED) / (NIR + 6*RED - 7.5*BLUE + 1), calculate the second index from the first data according to the formula LSWI = (NIR - SWIR1) / (NIR + SWIR1), and calculate the third index from the first data according to the formula NDBI = (SWIR1 - NIR) / (SWIR1 + NIR) (RED < NIR < SWIR1), where EVI is the Enhanced Vegetation Index, LSWI is the Land Surface Water Index, NDBI is the Normalized Difference Built - up Index, BLUE is the blue light band, RED is the red light band, NIR is the near - infrared band, and SWIR1 is the short - wave infrared band 1; Sub - index summarization module: It is used to summarize the first index, the second index, and the third index to obtain the second data.

9. A crop classification device based on multispectral images, characterized in that, It includes a memory and a processor, and the memory is used to store one or more computer instructions. Among them, the one or more computer instructions are executed by the processor to implement a crop classification method based on multi - spectral images as described in any one of claims 1 to 5.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a computer, it implements a crop classification method based on multi - spectral images as described in any one of claims 1 to 5.

Images