CN-121982567-A - Sugarcane lodging identification method, device and equipment based on remote sensing image mottle characteristics

CN121982567ACN 121982567 ACN121982567 ACN 121982567ACN-121982567-A

Abstract

The application discloses a sugarcane lodging recognition method, device and equipment based on a mottle characteristic of a remote sensing image, and relates to the field of remote sensing image processing, wherein the method comprises the steps of collecting multispectral remote sensing images in a research area; classifying pixels in the multispectral remote sensing image by adopting a pre-trained first machine learning model to determine a sugarcane planting area in the multispectral remote sensing image, calculating canopy coverage according to the types of the pixels in the sugarcane planting area, respectively carrying out section line analysis on the sugarcane planting area along the horizontal direction and the vertical direction to determine mottle characteristics, wherein the mottle characteristics comprise maximum background wavelength in the horizontal direction, maximum peak number in the vertical direction, peak period variance in the vertical direction and peak width variance in the vertical direction, and determining the lodging state of the sugarcane in the research area by adopting a pre-trained second machine learning model according to the canopy coverage and the mottle characteristics. The application improves the recognition efficiency and precision of the sugarcane lodging state.

Inventors

LIN YITONG
CHEN YANLI
YE JUNFEI
HUANG LU

Assignees

广西壮族自治区气象科学研究所

Dates

Publication Date: 20260505
Application Date: 20260122

Claims (10)

1. The sugarcane lodging identification method based on the mottle characteristics of the remote sensing image is characterized by comprising the following steps of: collecting multispectral remote sensing images in a research area; Classifying pixels in the multispectral remote sensing image by adopting a pre-trained first machine learning model, and determining a sugarcane planting area in the multispectral remote sensing image; Calculating canopy coverage according to the categories of each pixel in the sugarcane planting area, wherein the categories of the pixels comprise sugarcane and background; performing section line analysis on the sugarcane planting area along the horizontal direction and the vertical direction respectively to determine mottle characteristics, wherein the mottle characteristics comprise the maximum background wavelength in the horizontal direction, the maximum wave crest number in the vertical direction, the wave crest period variance in the vertical direction and the wave crest width variance in the vertical direction; and determining the lodging state of the sugarcane in the research area by adopting a pre-trained second machine learning model according to the canopy coverage and the mottled characteristics.
2. The method of claim 1, further comprising, prior to classifying the pixels in the multispectral remote sensing image using the pre-trained first machine learning model: and sequentially carrying out geometric correction, sugarcane field region cutting and image rotation treatment on the multispectral remote sensing image.
3. The method for identifying lodging of sugarcane based on mottle characteristics of a remote sensing image according to claim 1, wherein the bands of the multispectral remote sensing image comprise a near infrared band, a red light band, a green light band and a blue light band, classifying pixels in the multispectral remote sensing image by using a pre-trained first machine learning model, and determining a sugarcane planting area in the multispectral remote sensing image comprises: Calculating an ultra-green vegetation index according to the reflectivity of a red light wave band, the reflectivity of a green light wave band and the reflectivity of a blue light wave band in the multispectral remote sensing image; Calculating a normalized vegetation index according to the reflectivity of the near infrared band and the reflectivity of the red band in the multispectral remote sensing image; And determining the category of each pixel in the multispectral remote sensing image by adopting a first machine learning model trained in advance according to the supergreen vegetation index and the normalized vegetation index, wherein the first machine learning model is obtained by training a first training sample set in advance, and the first training sample set comprises the supergreen vegetation indexes, the normalized vegetation indexes and the category of each pixel of a plurality of multispectral remote sensing image samples.
4. The method for identifying sugarcane lodging based on remote sensing image mottle characteristics according to claim 1, wherein the first machine learning model and the second machine learning model are both support vector machines.
5. The method for identifying sugarcane lodging based on remote sensing image mottle characteristics according to claim 1, wherein the canopy coverage is calculated using the following formula: ; Wherein, the For the coverage of the canopy layer, For the number of sugarcane picture elements, Is the number of background pixels.
6. The method for identifying lodging of sugarcane based on mottle characteristics of remote sensing images according to claim 1, wherein performing section line analysis on the sugarcane planting area along horizontal and vertical directions respectively, determining mottle characteristics comprises: extracting section lines of the sugarcane planting area along the horizontal direction and the vertical direction respectively, and determining a plurality of section lines in the horizontal direction and a plurality of section lines in the vertical direction; Determining the maximum background wavelength in the horizontal direction according to the continuous longest length of the background pixels on each section line in the horizontal direction; determining the wave peak, the wave peak period and the wave peak width of each section line in the vertical direction according to the wave form of each section line in the vertical direction; Determining the maximum wave crest number in the vertical direction according to the maximum wave crest number of each section line in the vertical direction; calculating the variance of the period of adjacent wave peaks in each section line in the vertical direction to obtain the variance of the period of the wave peaks in the vertical direction; And calculating the variance of the peak width in each section line in the vertical direction to obtain the variance of the peak width in the vertical direction.
7. The method for identifying lodging of sugarcane based on mottle characteristics of remote sensing images according to claim 6, wherein the vertical direction peak period variance is determined using the following formula: ; Wherein, the The variance of the peak period in the vertical direction is given, n is the number of peaks in the vertical direction, For the ith period of a single vertical section line, For the average period of a single vertical section line, Representing the calculation of the average of the periodic variance of all section lines in the vertical direction.
8. The method for identifying lodging of sugarcane based on mottle characteristics of remote sensing images according to claim 6, wherein the vertical direction peak width variance is determined using the following formula: ; Wherein, the The variance of the width of the wave peak in the vertical direction is given, n is the number of wave peaks in the vertical direction, The jth peak width for a single vertical section line, The average peak width for a single vertical section line, Representing the calculation of the average of the peak width variances for all cross-hatching in the vertical direction.
9. A sugarcane lodging recognition device based on remote sensing image mottle characteristics, characterized in that the device performs the sugarcane lodging recognition method based on remote sensing image mottle characteristics according to any one of claims 1-8, the device comprising: The image acquisition module is used for acquiring multispectral remote sensing images in the research area; The sugarcane area extraction module is used for classifying pixels in the multispectral remote sensing image by adopting a pre-trained first machine learning model and determining a sugarcane planting area in the multispectral remote sensing image; The canopy coverage calculating module is used for calculating canopy coverage according to the categories of the pixels in the sugarcane planting area, wherein the categories of the pixels comprise sugarcane and background; The mottle characteristic extraction module is used for respectively carrying out section line analysis on the sugarcane planting area along the horizontal direction and the vertical direction to determine mottle characteristics, wherein the mottle characteristics comprise the maximum background wavelength in the horizontal direction, the maximum wave crest number in the vertical direction, the wave crest period variance in the vertical direction and the wave crest width variance in the vertical direction; And the lodging state identification module is used for determining the lodging state of the sugarcane in the research area by adopting a second machine learning model trained in advance according to the canopy coverage and the mottled characteristics.
10. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the remote sensing image mottled feature-based sugarcane lodging identification method of any one of claims 1-8.

Description

Sugarcane lodging identification method, device and equipment based on remote sensing image mottle characteristics Technical Field The application relates to the field of remote sensing image processing, in particular to a sugarcane lodging identification method, device and equipment based on the mottled characteristic of a remote sensing image. Background Sugarcane lodging is one of the key factors affecting the yield and quality of sugarcane, and the rapid and accurate identification of lodging areas is critical for post-disaster assessment and production management. The traditional sugarcane lodging monitoring mainly relies on manual field investigation, and the method is low in efficiency, high in labor intensity, limited by manual subjective experience, difficult to realize synchronous and objective screening of a large-scale planting area and extremely easy to miss the optimal agronomic intervention time. With the rapid development of remote sensing technology and unmanned aerial vehicle platforms, the utilization of unmanned aerial vehicles for crop growth monitoring has become an important technical means of modern agriculture. The unmanned aerial vehicle has the remarkable advantages of flexible operation, high imaging resolution, low data acquisition cost and the like, and provides an ideal data source for rapidly and automatically acquiring the growth state of crops in the field. At present, lodging recognition research based on unmanned aerial vehicle images is mostly dependent on spectral features and conventional texture features, but in practical application, too many feature parameters mean huge calculated amount and high requirements on hardware computing power, which restricts deployment and application efficiency of an algorithm in a large-scale and business scene to a certain extent. Disclosure of Invention The application aims to provide a sugarcane lodging identification method, device and equipment based on the mottle characteristics of remote sensing images, which can automatically and accurately identify the lodging state of sugarcane. In order to achieve the above object, the present application provides the following solutions: In a first aspect, the application provides a sugarcane lodging identification method based on a mottle characteristic of a remote sensing image, which comprises the following steps: collecting multispectral remote sensing images in a research area; Classifying pixels in the multispectral remote sensing image by adopting a pre-trained first machine learning model, and determining a sugarcane planting area in the multispectral remote sensing image; Calculating canopy coverage according to the categories of each pixel in the sugarcane planting area, wherein the categories of the pixels comprise sugarcane and background; performing section line analysis on the sugarcane planting area along the horizontal direction and the vertical direction respectively to determine mottle characteristics, wherein the mottle characteristics comprise the maximum background wavelength in the horizontal direction, the maximum wave crest number in the vertical direction, the wave crest period variance in the vertical direction and the wave crest width variance in the vertical direction; and determining the lodging state of the sugarcane in the research area by adopting a pre-trained second machine learning model according to the canopy coverage and the mottled characteristics. In a second aspect, the application provides a sugarcane lodging recognition device based on a mottle characteristic of a remote sensing image, which comprises: The image acquisition module is used for acquiring multispectral remote sensing images in the research area; The sugarcane area extraction module is used for classifying pixels in the multispectral remote sensing image by adopting a pre-trained first machine learning model and determining a sugarcane planting area in the multispectral remote sensing image; The canopy coverage calculating module is used for calculating canopy coverage according to the categories of the pixels in the sugarcane planting area, wherein the categories of the pixels comprise sugarcane and background; The mottle characteristic extraction module is used for respectively carrying out section line analysis on the sugarcane planting area along the horizontal direction and the vertical direction to determine mottle characteristics, wherein the mottle characteristics comprise the maximum background wavelength in the horizontal direction, the maximum wave crest number in the vertical direction, the wave crest period variance in the vertical direction and the wave crest width variance in the vertical direction; And the lodging state identification module is used for determining the lodging state of the sugarcane in the research area by adopting a second machine learning model trained in advance according to the canopy coverage and the mottled characteristics. In a third aspect, the