Palm oil fresh fruit bunch ripeness grading identification using color features

Abstract

This research investigates the ripeness grading identification of the palm oil FFB using color
features that are color histogram, color moment and color correlogram. Palm is harvested
during the optimum stage of its ripeness since it improves the FFB oil quality and quantity.
Harvesting wrong bunches decreases the oil extraction rate of the palm. A preliminary
research on the palm oil FFB grading identification is conducted. Each ripeness stage has its
own unique color. A study on color features is investigated.A new dataset of images of FFB is
constructed. A comparative study between Support Vector Machine (SVM) and Naïve Bayes
classifiers has been performed using the values of color histogram, color moment and color
correlogram. The results of the experiments indicate that color moment with SVM produce a
higher palm oil FFB ripeness grading identification accuracy compared to color histogram
and color correlogram.

Keywords: color features, fresh fruit bunch (FFB), Naïve Bayes, ripeness stage, support
vector machine.