A non-destructive oil palm ripeness recognition system using relative entropy

• A relative entropy based image processing approach to detect oil palm ripeness.
• It can replace human’s visual inspections for oil palm grading activity.
• The results are very accurate with very fast process.
• An Android application based on our algorithm was also developed.
• The system is simple and can be useful for oil palm farmers and entrepreneurs.

This paper introduces a relative entropy based image processing approach for the non-destructive prediction of the maturity of oil palm fresh fruit bunches (FFB) which enables the determination of the correct time for harvesting. The results of an experimental study of applying the Kullback–Leibler distance to the problem of oil palm classification are presented. It is shown that the proposed algorithm has an excellent accuracy and it can be computed very fast. The overall proposed system is simple and useful for oil palm farmers and entrepreneurs.