A Computerized Digital Imaging Technique to Estimate Palm Oil Quality Based on Fruit Colour

Here, researchers developed an automated technique of estimating quality based on fruit colour captured in JPEG format.

This research has won the following awards:
GOLD— British Invention Show (BIS 2007).
GOLD— International Invention Innovation Industrial Design & Technology Exhibition (I-TEX 2007).

SILVER— UPM Invention, Research & Innovation Exhibition (PRPI 2006).

To see the Figures described in this write-up, please download the attached document.

TITLE: A Computerized Digital Imaging Technique to Estimate Palm Oil Quality Based on Fruit Colour

RESEARCHERS: Siva K Balasundram, Ahmad Husni Mohd Hanif* and Anuar Abdul Rahim

Typically, palm oil quality is determined in a destructive manner using wet chemical analysis. Such an analysis can be costly, time consuming and error prone. Deterioration of Bleachability Index (DOBI) is a key indicator of palm oil quality. An alternative evaluation procedure that combines consistency and reporting accuracy is desirable. One such method is the use of fruit color as an indicator of palm oil quality.

A recent study showed that fruit color was significantly correlated with palm oil content (Balasundram et al., 2006a). An extension of that work further showed that fruit color was also significantly correlated with DOBI (Balasundram et al., 2006b). Both these studies employed a digital imaging approach to quantify fruit color. Oil palm fruit images were captured using a high resolution digital camera. After imaging, the fruits were manually squeezed for oil and analyzed for total oil content and DOBI using standard laboratory procedures. The images were processed using ILWIS 3.2, an image analysis program, to generate percentage of color based on clustering and unsupervised classification. Each classified image was re-colored to its original color separation, which comprised black, red, orange and yellow components (Figure1). Correlation and stepwise regression techniques were used to determine the relationship between oil content/DOBI and fruit color components.

In this research, we developed an automated technique of quantifying fruit color components based on digital images captured in Joint Photographic Experts Group (JPEG) format. This technique is based on a simple computer program written in Visual Basics and interfaced with ILWIS 3.2 (Figure 2). Using the empirical relationship between oil content/DOBI and fruit color, an additional step of estimating % total oil content and/or DOBI values is also made possible. Essentially, the digital fruit image (in JPEG format) serves as the input data to churn out two levels of output sets, i.e., the percentage of color components (black, red, orange and yellow) and the predicted oil content and/or DOBI value.

This technique offers a non-destructive means of assessing palm oil quality and can enable oil yield and/or oil quality mapping to facilitate precision oil palm management. Currently, there is no practical method of mapping oil palm yields/quality due to logistical and instrumentation limitations.

For inquiries, please contact:

Siva K Balasundram
Department of Agriculture Technology
Faculty of Agriculture
Universiti Putra Malaysia
43400 UPM Serdang
Selangor, Malaysia
Tel: +603-8946 4186
Email: siva@agri.upm.edu.my