Water stress detection method for tomatoes in seedling stage based on micro-CT and polarization-hyperspectral imaging multi-feature fusion

Patent metadata

Field	Value
Publication number	US-11436824-B2
Application number	US-201716646351-A
Country	US
Kind code	B2
Filing date	Dec 19, 2017
Priority date	Dec 5, 2017
Publication date	Sep 6, 2022
Grant date	Sep 6, 2022

Abstract

A water stress detection method for tomatoes in a seedling stage based on micro-CT and polarization-hyperspectral imaging multi-feature fusion, comprising: using micro-CT to scan microscopic morphological features such as water stress stomata, spongy body, palisade tissue, cilia, vascular bundle, root volume, main root, and root hair density of tomatoes; using a polarization-hyperspectral imaging system to obtain macroscopic morphological features such as crown width, plant height, and leaf inclination of water stress plants, as well as leaf vein distribution, average gray, and leaf margin shaded area under a water-sensitive wavelength of 1450 nm, and macroscopic morphological features such as polarization states, stock vectors, and Mueller matrix variables of 1450 nm feature images at 0°, 45°, 90°, 135°, and 180° feature polarization angles. By fusion of internal and external structures, above-ground, underground, and macroscopic and microscopic morphological features of water stress tomatoes, and mutual fusion of water stress feature wavelength images and polarization state features, advantages are complementary, comprehensive and precise extraction and precise quantitative analysis of water stress features of the tomatoes are implemented, and a basis for scientific management of water and fertilizer integration of facilities is provided.

First claim

The invention claimed is:

1. A water stress detection method for tomatoes in a seedling stage based on micro computed tomography (micro-CT) and polarization-hyperspectral imaging multi-feature fusion, the method comprising the following steps:
step 1: utilizing pearlite as a substrate, employing soilless cultivation to plant tomatoes by using a nutrient solution, and managing the tomatoes to ensure a supply of nutrient elements and water to the tomatoes;
step 2: after one week of planting, culturing samples of the planted tomatoes under different levels of water stress for three days, while keeping the nutrient elements unchanged;
step 3: performing continuous tracking and detection of water stress for the tomato samples under water stress, and performing imaging, the imaging comprising performing micro-CT detection to acquire micro-CT feature parameters of the tomato samples and performing polarization-hyperspectral imaging to acquire polarization-hyperspectral feature parameters of the tomato samples, the micro-CT feature parameters comprising a pore size, a pore density, a thickness of cavernous body, a palisade tissue, a cilia density, a cross-sectional structure of vascular bundles of plant leaves and stems, a volume of a root system, and density and distribution parameters of main root and root hair, and the polarization-hyperspectral feature parameters comprising a plant crown width, a plant height, leaf inclination angle images, a distribution of leaf vein, an average grayscale, a shadow area of leaf margin at 1,450 nanometers (nm) hyperspectral water-sensitive wavelength, a polarization state, a Stock vector, and Muller matrix variables of a crown layer of the tomato samples under water stress in 1,450 nm feature images at 0°, 45°, 90°, 135°, and 180° characteristic polarization angles;
step 4: measuring a water content in the tomato samples with a dry-wet weight method, using scanning electron microscope (SEM) and micro-imaging techniques to obtain measured values of density of the pore density, the cilia density, the thickness of cavernous body, the palisade tissue, and a distribution density and diameter of vascular bundle of the tomato samples, the dry-wet weight method comprising weighing dry and wet weight the tomato samples to determine a true value of the water content in the tomato samples;
step 5: carrying out a normalization of the micro-CT feature parameters and the polarization-hyperspectral feature parameters acquired in step 3 to unify a range of the micro-CT feature parameters and the polarization-hyperspectral feature parameters to 0 to 1 and obtain normalized feature parameters;
step 6: carrying out feature dimension reduction and optimization of the normalized feature parameters obtained in step 5 by means of principal component analysis in combination with a piecewise and stepwise regression method based on a principle of correlation and independence at a significance level of α=0.005, keeping a variable if F>4.14 when the variable is taken into the analysis, weeding out a variable if F<2.91 in the analysis during discrimination, while maintaining R 2 >0.9, and carrying out feature optimization based on optimization principles of maximum correlation, minimum multi-collinearity, and minimum relative detection error to obtain optimal feature parameters, from the micro-CT feature parameters and the polarization-hyperspectral feature parameters, for diagnosis of water stress of the tomato samples, where F is a significance coefficient of a linear relationship in the piecewise and stepwise regression method and R 2 is a correlation coefficient the piecewise and stepwise regression method;
step 7: utilizing a support vector machine regression (SVR) method to carry out feature layer fusion on the optimal feature parameters, and establishing an accurate and quantitative water stress detection model with multi-feature fusion based on the optimal feature parameters;
step 8: acquiring an updated set of micro-CT feature parameters of the tomato samples from the micro-CT detection and an updated set of polarization-hyperspectral feature parameters of the tomato samples from the polarization-hyperspectral imaging according to step 3, and utilizing the accurate and quantitative water stress detection model with multi-feature fusion established in step 7, along with the updated set of micro-CT feature parameters and the updated set of polarization-hyperspectral feature parameters to carry out detection of water stress in an environment having the tomatoes in the seedling stage.

2. The method according to claim 1 , wherein the performing of the micro-CT detection to acquire micro-CT feature parameters of the tomato samples comprises:
(1) placing the tomato samples, comprising five tomato samples, under different levels of water stress on a rotating sample bracket in a sample chamber of a micro-CT scanning and imaging system sequentially, starting the micro-CT scanning and imaging system via a control computer and performing scanning sequentially to obtain respective CT profiles of the tomato samples;
(2) using initial program load (IPL) software to select boundaries and contours in the CT images of the tomato samples;
(3) selecting, within each CT image of the CT images of the tomato samples, different tomography sections for image analysis, adjusting a high threshold and a low threshold according to different grayscale levels of a target in the respective CT image, selecting a threshold range for the target, and binarizing the respective CT image;
(4) using the IPL software in combination with image analysis to obtain the pore size, the pore density, the thickness of cavernous body, the palisade tissue, the cilia density, and the cross-sectional structure of vascular bundle;
(5) removing, from each CT image of the CT images of the tomato samples, the substrate on a basis of the selected boundaries and high threshold and low threshold, generating a three-dimensional image of the root system, and carrying out IPL language to obtain the volume of the root system and the density and distribution parameters of the main root and root hair.

3. The method according to claim 1 , wherein the performing of the polarization-hyperspectral imaging to acquire polarization-hyperspectral feature parameters of the tomato samples comprises:
(1) placing each tomato sample of the tomato samples on a double coordinate sample table of a polarization-hyperspectral imaging system, setting a wavelength range of a visible light-near infrared light source system to 300 nm to 2,200 nm and setting a light intensity range to 0 lux to 6,500 lux;
(2) using two hyperspectral imaging systems with pre-polarization filters, and setting sampling polarization angles of the polarization filters to 0°, 45°, 90°, 135°, and 180°, respectively; using hyperspectral pre-filters with 1,450 nm transmission wavelength, and performing push-broom scanning and imaging in a horizontal plane direction and a vertical plane direction to obtain front-view and top-view, respectively, polarization-hyperspectral feature images of the respective tomato sample;
(3) extracting hyperspectral feature images of the respective tomato sample under water stress in front-view and top-view fields, and extracting the plant crown width, the plant height, the leaf inclination angle images of the respective tomato sample by means of coordinate matching and front-view/top-view feature image fusion;
(4) extracting a hyperspectral feature image of the crown layer at the wavelength of 1,450 nm, and extracting the distribution of leaf vein, the average greyscale, and the shadow area of leaf margin at 1,450 nm hyperspectral water-sensitive wavelength, based on the hyperspectral pre-filters with a 1,450 nm transmission wavelength;
(5) extracting the polarization state, the Stock vector, and the Muller matrix variables of th

Assignees

• Univ Jiangsu

Inventors

• Zhang Xiaodong
• Mao Hanping
• Gao Hongyan
• Zuo Zhiyu
• Zhang Yixue

Classifications

• G01N33/0098 Primary

  Plants or trees (wood G01N33/46) · CPC title

• G06V10/806

  of extracted features · CPC title

• G06V10/7715

  Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06V20/188 Primary

  Vegetation · CPC title