Search

CN-122016668-A - Laser pretreatment post-waxy corn chlorophyll inversion method based on unmanned aerial vehicle multispectral remote sensing

CN122016668ACN 122016668 ACN122016668 ACN 122016668ACN-122016668-A

Abstract

The invention discloses a laser pretreatment waxy corn chlorophyll inversion method based on unmanned aerial vehicle multispectral remote sensing. The method comprises the steps of dividing waxy corn seeds into 6 groups and respectively adopting different laser parameters for pretreatment irradiation, acquiring multispectral images of all processing plots by using an unmanned aerial vehicle carrying a six-band multispectral sensor in the growth process of waxy corn, carrying out radiation correction and orthographic processing on the multispectral images, extracting texture features corresponding to six bands, synchronously adopting a chlorophyll meter to carry out SPAD measurement on all processing plots, establishing a regression inversion model between remote sensing texture features and ground SPAD, and substituting the texture features obtained by calculating the multispectral images of the plots to be detected into the model in the prediction stage to obtain inversion results of the waxy corn SPAD. Compared with the prior art, the method utilizes the multispectral texture characteristics to construct the inversion model, can realize the rapid and non-contact estimation of the chlorophyll value of waxy corn under different laser pretreatment conditions, and improves the monitoring efficiency and consistency.

Inventors

  • ZHU YIFENG
  • ZHAO YANG

Assignees

  • 长春理工大学

Dates

Publication Date
20260512
Application Date
20260105

Claims (10)

  1. 1. A method for inverting chlorophyll of waxy corn after laser pretreatment based on unmanned aerial vehicle multispectral remote sensing is characterized by comprising the following steps of S1, dividing waxy corn seeds into 6 treatment groups, respectively adopting different laser parameters to conduct pretreatment irradiation on the treatment group seeds and complete sowing and field management, S2, carrying out aerial survey on each treatment block by using an unmanned aerial vehicle carrying a six-band multispectral sensor in a target growth period of waxy corn growth to obtain multispectral images, S3, carrying out radiation correction, geometric correction and normal-emission splicing on the multispectral images to obtain reflectivity images for inversion calculation, S4, calculating gray level symbiotic matrix texture characteristics corresponding to six bands in corresponding areas of each treatment block based on the reflectivity images, at least comprising third band contrast characteristics b3_contast, fifth band contrast characteristics b5_contast and sixth band correlation characteristics b6_corration, S5, carrying out value measurement on sample directions of chlorophyll pieces of each treatment block by using a chlorophyll meter, obtaining a SPAD, carrying out regression on the texture values of chlorophyll pieces, and carrying out inversion model in the inversion step of SPAD 6, and outputting SPAD to obtain the inversion model.
  2. 2. The method according to claim 1, wherein the texture feature is calculated in step S4 based on a gray level co-occurrence matrix, the gray level co-occurrence matrix is constructed in four directions of 0 °, 45 °, 90 °, 135 °, respectively, and the texture feature is obtained by arithmetically averaging texture feature values in the four directions.
  3. 3. The method according to claim 1, wherein the texture feature in step S4 is calculated by sliding window having a size of Wherein An odd number of 9 to 31, preferably 11.
  4. 4. The method according to claim 1, wherein the regression inversion model in step S6 is a linear regression model, satisfying: . Wherein a, b, c, d is the model coefficient obtained by training sample fitting.
  5. 5. The method of claim 4, wherein the model coefficients satisfy a = 323.058, b= -0.05937, c = 0.00940, d= -225.189, resulting in:
  6. 6. The method according to claim 1, further comprising the step of screening and optimizing candidate texture features, prior to constructing the regression inversion model of step S6, based on the reflectance image calculation to obtain a set of candidate texture features comprising different wavebands and different texture indexes, pre-screening the candidate texture features according to their correlation with the ground SPAD, and further performing a multiple co-linearity test to reject redundant features, and then determining a subset of key texture features for modeling based on regression fit evaluation indexes and/or cross-validation errors, wherein the subset of key texture features comprises at least b3_comparison, b5_comparison, and b6_correlation.
  7. 7. The method of claim 1, wherein in step S4 the block area is determined by a block boundary vector, and the block boundary vector is buffered inward to form a sampling area, a predetermined number of random sampling points are generated in the sampling area, and spectral features and/or texture features of corresponding pixels or windows are extracted from the reflectivity image centered on the random sampling points, wherein the texture features at least include b3_comparison, b5_comparison, and b6_comparison for inversion in claim 1, and features of a plurality of random sampling points in the same block are statistically summarized to obtain block-level features, so as to avoid influence of mixed pixels at the block boundary on feature calculation.
  8. 8. The method of claim 1, wherein the radiation correction in step S3 is scaled with a white reflectance scale (whiteboard) to convert the digitally quantized values of the multispectral image to reflectance.
  9. 9. The waxy corn chlorophyll value inversion system for implementing the method of any one of claims 1-7 is characterized by comprising an unmanned aerial vehicle platform, a six-band multispectral sensor, a chlorophyll meter and a data processing terminal, wherein the data processing terminal is used for executing image correction, texture feature calculation, regression model training and SPAD inversion output.
  10. 10. A computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the method of any of claims 1 to 7.

Description

Laser pretreatment post-waxy corn chlorophyll inversion method based on unmanned aerial vehicle multispectral remote sensing Technical Field The invention relates to the field of agricultural remote sensing monitoring and crop physiological parameter inversion, in particular to a method for realizing inversion estimation of waxy corn chlorophyll content (represented by SPAD value) after laser pretreatment by utilizing six-band multispectral remote sensing images of an unmanned aerial vehicle and combining with gray level co-occurrence matrix texture characteristics. Background Chlorophyll content is an important physiological index reflecting the nitrogen nutrition status, photosynthetic capacity and growth potential of crops, and usually adopts the SPAD value measured by a chlorophyll meter as a characterization parameter. The existing chlorophyll monitoring method mainly comprises modes of manual sampling, instrument point measurement and the like, but the method has the problems of limited measuring points, time and labor consumption, insufficient space representativeness and the like, and is difficult to meet the requirements of large-scale, rapid and continuous field monitoring. With the development of unmanned aerial vehicle remote sensing technology, the utilization of unmanned aerial vehicle carried multispectral sensor to obtain crop canopy reflectivity information, and inversion of chlorophyll content through vegetation index or regression model has become an important means. However, the existing inversion method based on single spectrum or a small amount of vegetation indexes is easily influenced by factors such as illumination change, imaging conditions, soil background, canopy structure change and the like, especially in test scenes where seeds are pretreated by different laser parameters, different treatments may cause the differences of emergence, growth potential and canopy space structure, so that the stability of a model which only depends on spectrum intensity or single vegetation indexes is insufficient, and inversion precision fluctuation under the conditions of crossing plots and crossing treatments is larger. Therefore, a technical scheme which can give consideration to multispectral information and canopy space structural features and is suitable for rapid inversion of waxy corn chlorophyll parameters under different laser pretreatment conditions is needed, so that robustness and application reliability of chlorophyll inversion are improved. Disclosure of Invention The invention aims to provide a laser pretreatment waxy corn chlorophyll inversion method based on unmanned aerial vehicle multispectral remote sensing, which realizes quick and non-contact inversion estimation of waxy corn SPAD values under different laser pretreatment conditions by introducing multispectral texture features to participate in modeling, thereby reducing the manual measurement cost and improving the monitoring efficiency and stability of the land parcel scale. In order to achieve the above purpose, the invention adopts the following technical scheme: grouping waxy corn seeds and applying different laser parameters for pretreatment (comprising the combination of different irradiation durations of red light and blue light in a control group) before sowing; Unmanned aerial vehicle aerial survey is carried out in a target growth period of waxy corn growth, wherein the target growth period is a preset growth period for chlorophyll inversion, and can be selected from one or more of a jointing period, a large bell mouth period, a male-pulling flowering period or a grouting period, and aerial survey is preferably carried out in the male-pulling flowering period so as to utilize the period that canopy coverage is complete and chlorophyll characterization is more sensitive, and consistency and application value of inversion results are improved. During navigation, an unmanned aerial vehicle carrying a six-channel multispectral camera is adopted to acquire multispectral images of a test plot, the central wavelengths of the six channels of the camera are respectively 450 nm, 555 nm, 660 nm, 720 nm, 750 nm and 840 nm, and the radiation calibration of a white reflectivity calibration plate (white board) is completed before the navigation measurement, so that the influence of illumination change on the consistency of the image reflectivity is reduced. Splicing, radiation correction and orthographic processing are carried out on the acquired multispectral images to obtain reflectivity images for calculation; extracting features in the field through random sampling points, constructing a window by taking the sampling points as the center to calculate the texture features of the gray level co-occurrence matrix, adopting the directions of 0 degree, 45 degrees, 90 degrees and 135 degrees for texture calculation, taking an average value, and simultaneously stretching and quantizing window pixels, for example 11X 11 pixels, and