Robust Grape Cluster Detection in a Vineyard by Combining the AdaBoost Framework and Multiple Color Components.

Area of Science:

• Agricultural Robotics
• Computer Vision

Background:

• Automated fruit detection and precision picking in unstructured environments remain challenging for harvesting robots.
• Accurate identification of grape clusters in vineyards is crucial for efficient harvesting.

Purpose of the Study:

• To develop an automated approach for detecting ripe grape clusters using a simple vision sensor.
• To combine the AdaBoost framework with multiple color components for enhanced grape detection.

Main Methods:

• Acquired a dataset of grape cluster images from vineyard scenes using a color digital camera.
• Extracted effective color components and constructed linear classification models using the threshold method.
• Developed a strong classifier using the AdaBoost framework and applied region thresholding and morphological filtering for noise elimination.

Main Results:

• The strong classifier achieved a classification accuracy of 96.56% on 900 testing samples, outperforming other linear models.
• The approach demonstrated an average detection rate of 93.74% on 200 images captured under varying illumination conditions.
• The method showed resilience to complex backgrounds, including weather conditions, leaves, and changing illumination.

Conclusions:

• The proposed AdaBoost-based approach effectively detects grape clusters in vineyards.
• This method offers a robust solution for automated grape harvesting, adaptable to environmental variations.