Semi-Automated Field Plot Segmentation From UAS Imagery for Experimental Agriculture

Area of Science:

• Agricultural Science
• Remote Sensing
• Image Processing

Background:

• Unmanned Aerial System (UAS) use in agriculture offers a cost-effective alternative for crop monitoring.
• Accurate segmentation of crop plots is crucial for assessing crop varieties and treatments.
• Existing methods often require extensive manual parameterization, limiting their efficiency.

Purpose of the Study:

• To develop an automated image processing method for precise crop plot segmentation from UAS imagery.
• To improve the accuracy and efficiency of agricultural monitoring through reliable plot segmentation.
• To reduce the need for manual parameterization in crop plot segmentation.

Main Methods:

• The study employed a novel image processing technique combining edge detection and Hough line detection.
• This method establishes plot boundaries and calculates pixel/point-based metrics for each segment.
• Limited manual parameterization was used to adapt the segmentation process.

Main Results:

• The developed method achieved consistent segmentation accuracy exceeding 89% across various crop types and conditions.
• Performance is comparable to highly contrasted scenarios like rice paddies.
• This represents a significant advancement over previous segmentation methods for dry land crops.

Conclusions:

• The new image processing method provides a reliable and accurate solution for crop plot segmentation using UAS imagery.
• The approach significantly reduces manual input requirements, enhancing cost-effectiveness and timeliness in agricultural monitoring.
• This method offers a substantial improvement for assessing diverse agricultural landscapes.