Research on weed identification method in rice fields based on UAV remote sensing.

Area of Science:

• Agricultural Science
• Remote Sensing
• Environmental Science

Background:

• Weed competition significantly impacts rice yield and quality, affecting global food security.
• Current herbicide application methods are often inefficient, leading to environmental pollution and crop contamination.
• Precise weed identification is crucial for targeted herbicide management in rice cultivation.

Purpose of the Study:

• To develop an effective weed identification index for rice fields using UAV multispectral imagery.
• To compare the performance of the new WDVI index against traditional vegetation indices for weed detection.
• To establish a robust method for weed identification and mapping in rice cultivation.

Main Methods:

• Construction of the Weed Detection Vegetation Index (WDVI) using Red, Green, and Near-Infrared reflectance bands from UAV multispectral images.
• Comparative analysis of WDVI against NDVI, LCI, NDRE, and OSAVI for weed identification accuracy.
• Application of small patch removal and clustering algorithms for weed vector result generation and accuracy verification using confusion matrix and Kappa coefficient.

Main Results:

• The WDVI vegetation index demonstrated superior performance in distinguishing weeds from rice, water cotton, and soil compared to traditional indices.
• The developed weed identification method achieved a high accuracy of 93.47% with a Kappa coefficient of 0.859.
• The study successfully generated weed identification vector results with high precision.

Conclusions:

• The WDVI index offers a novel and effective approach for precise weed identification in rice fields.
• This remote sensing-based method enables targeted weed management, reducing reliance on uniform herbicide application.
• The findings contribute to sustainable rice farming practices by minimizing environmental impact and improving crop quality.