Assessment of cluster yield components by image analysis.

Area of Science:

• Agricultural Science
• Computer Vision
• Horticulture

Background:

• Grape yield estimation relies on berry weight, berry number, and cluster weight.
• Current methods for yield prediction are destructive, labor-intensive, and time-consuming.

Purpose of the Study:

• Develop a novel, rapid, and cost-effective image analysis methodology for determining grapevine cluster yield components.
• To provide an alternative to traditional, time-consuming methods.

Main Methods:

• Utilized image analysis techniques, including Canny and logarithmic image processing, coupled with the Hough Transform for berry detection.
• Photographed clusters from seven Vitis vinifera L. red varieties under laboratory conditions.
• Analyzed single images and multiple images (four per cluster) from different orientations.

Main Results:

• The Canny algorithm with four images per cluster yielded the best results, achieving R(2) values between 69-95% for berry detection and 65-97% for cluster weight estimation.
• The image analysis model demonstrated an 84% capability in predicting berry weight.
• The developed methodology proved effective in assessing cluster yield components.

Conclusions:

• The presented image analysis methodology offers a low-cost and time-saving solution for assessing grapevine cluster yield components.
• This approach provides valuable information compared to conventional manual methods.
• The study highlights the potential of image analysis in viticulture for efficient yield estimation.