UAV-Based Hyperspectral Imaging for Wheat Stripe Rust Monitoring

author: Callum
2025-02-19
texture features
Introduction
Wheat, one of China's three major grain crops (wheat, rice, and corn), is crucial for national food security. However, crop diseases (with over 20 common types) pose a severe threat to wheat production. Wheat stripe rust, caused by Puccinia striiformis, can spread rapidly, reducing yields by over 40% in epidemic years and even causing total crop failure. Effective control of wheat stripe rust is essential for food security, and accurate disease monitoring is critical for early intervention and loss mitigation.
Traditional disease monitoring relies on visual inspection, which has significant limitations such as small coverage and subjective judgment. Therefore, an efficient, non-destructive monitoring method is needed. Remote sensing has become a powerful tool for crop disease detection due to its ability to assess disease distribution at a lower cost. Among remote sensing techniques, hyperspectral imaging (HSI) offers more detailed spectral information than multispectral imaging (MSI). Previous studies have demonstrated HSI’s advantages over MSI in various applications, particularly in detecting physiological changes caused by diseases (e.g., pigment and water content variations).
HSI has been successfully used to monitor multiple crop diseases, including wheat stripe rust, powdery mildew, Fusarium head blight, peanut leaf spot, tomato spotted wilt virus, and bacterial leaf blight in rice. Vegetation indices (VI) have shown promising performance in disease detection, as they are sensitive to internal physiological changes in leaves. However, VI alone cannot capture surface texture changes. Texture features (TF), which reflect external leaf damage, have proven effective in disease monitoring. Studies have shown that combining spectral and texture features enhances disease characterization. However, most previous research has focused on leaf-scale detection, leaving the feasibility of field-scale monitoring unverified.
UAV-based hyperspectral imaging combines flexibility, low cost, and ease of operation, overcoming limitations of satellite-based remote sensing. By mounting hyperspectral sensors on UAVs, high-spatial and high-temporal resolution data can be rapidly collected and processed, supporting intelligent agricultural management. This study explores UAV-based hyperspectral monitoring of wheat stripe rust across different infection stages at multiple spatial scales.
Objectives
  1. Evaluate the performance of UAV-based VI, TF, and their combination in field-scale wheat stripe rust monitoring.
  2. Determine the optimal spatial resolution for UAV-based disease detection.
  3. Develop monitoring models for early, middle, and late infection stages using optimal spectral and spatial features.
Figure 1 Normal Wheat Leaf (Left) and Wheat Leaf with Stripe Rust (Right)
Methodology
1. Hyperspectral Data Acquisition & Preprocessing
A UAV equipped with a hyperspectral imaging sensor (Fig. 2) was used to capture wheat canopy images. The system includes:
  • A hexacopter UAV
  • A hyperspectral data acquisition system
  • A three-axis stabilization platform
  • A data processing system
Figure 2 UAV Hyperspectral Images
The UAV was flown at 30 m altitude with a speed of 4 m/s, achieving an 80% forward overlap and 60% side overlap. The acquired hyperspectral images had a spatial resolution of 1.2 cm and a spectral resolution of 4 nm. Flights were conducted under clear, low-wind conditions between 11:00 AM – 1:00 PM.
Monitoring was performed on six dates (April 25, May 4, May 11, May 18, May 24, and May 30), corresponding to 7, 16, 23, 30, 36, and 42 days post-inoculation (DPI).
Figure 3 Wheat at Different Infection Stages
A total of 144 samples were collected (57 healthy, 87 infected). Disease severity was assessed using Disease Index (DI), following the GB/T 15795-2011 national standard, which classifies infection levels into 9 grades (0%–100%).
2. Disease Detection Model
Partial Least Squares Regression (PLSR) was used to develop wheat stripe rust monitoring models. PLSR has been widely applied for estimating crop growth and biochemical parameters. It integrates Principal Component Analysis (PCA), Canonical Correlation Analysis (CCA), and Multiple Linear Regression (MLR), transforming high-dimensional hyperspectral data into a smaller set of predictive variables.
3. Hyperspectral Pixel Selection
Original images (1.2 cm resolution) were resampled using the nearest neighbor algorithm to generate images at 3, 5, 7, 10, 15, and 20 cm resolutions. Only wheat pixels were extracted using the NDVI thresholding method (threshold = 0.42), ensuring consistency across resolutions. This step removed background interference, allowing a pure spectral analysis of wheat canopies.
Figure 4 UAV Hyperspectral Images at Different Spatial Resolutions
Results & Conclusion
This study demonstrated UAV-based HSI’s effectiveness in field-scale wheat stripe rust monitoring. Key findings include:
  1. VI-based models achieved the highest accuracy in the middle infection stage, while TF-based models performed best in the late stage. However, TF-based models were unreliable for early detection (R² = 0.28).
  2. VI-TF combined models outperformed both VI and TF models across all infection stages, with the highest accuracy in the late stage (R² = 0.88). Notably, early detection accuracy also improved significantly, making it suitable for early-stage disease monitoring.
  3. Spatial resolution impact: VI-based models were less affected by resolution changes, while TF-based models were highly sensitive. The optimal resolution for UAV-based VI-TF disease monitoring was 10 cm.
  4.  
Figure 5 Identification Results
This research confirms that UAV-based HSI, integrating spectral and texture features, can enhance wheat stripe rust monitoring accuracy at different infection stages. The findings provide valuable insights for precision agriculture, supporting early detection and effective disease management.

UAV Hyperspectral Detection of Citrus Huanglongbing

Hyperspectral Analysis of Grain Year Identification

Related Article

default
Discover how UAV‑based hyperspectral imaging and deep learning can monitor SDI, TOC, and TEP – key membrane fouling indicators for seawater desalination during harmful algal blooms. Based on a recent Water Research study, this article explores how Optosky's ATH9010 enables spatial risk mapping and proactive intake management.
Applications | ATH9010 Enables Water Quality and Membrane Fouling Monitoring for Desalination During Algal Blooms
2026-09-11
default
Choosing between 785 nm and 1064 nm for Raman? This guide explains the physics of fluorescence, compares signal strength and fluorescence suppression, and provides a step‑by‑step decision tree. Real‑world drug detection and pesticide examples show why wavelength matters – and how dual‑wavelength coverage offers the ultimate solution.
785 nm or 1064 nm? Fluorescence Is the Dividing Line
2026-09-10
default
Inspired by the 2026 Science paper from Zhang Jun’s team on a video‑rate on‑chip hyperspectral microsystem, Optosky’s ATH series moves beyond lab‑only tools. We deliver tailored, real‑world hyperspectral systems that integrate compact hardware, edge AI, and application‑specific algorithms—turning raw spectral data into actionable decisions, right where you need them.
From Science to the Field: Custom Hyperspectral Solutions by Optosky
2026-09-08
default
See how 40 Optosky NY3300Pro multispectral units were deployed across a large‑scale demonstration farmland in Northwest China. With 9‑band imaging, solar power, 4G transmission, and automated data analytics, this solution enables precise growth monitoring, early pest detection, and water‑fertiliser optimisation – a true leap toward smart agriculture.
Case Study | Bulk NY3300Pro Installation Transforms Smart Farming
2026-08-28