(For USM Staff/Student Only)

EngLib USM > Ω School of Electrical & Electronic Engineering >

Oil palm fruit ripeness detection kit for harvesting decision

Oil palm fruit ripeness detection kit for harvesting decision / Heng Peng Zhi
Klasifikasi kematangan buah tandan segar (FFB) kelapa sawit semasa penuaian adalah penting bagi memastikan FFB dituai pada peringkat optimum untuk pengeluaran minyak sawit yang maksimum. Penuaian FFB pada peringkat kematangan yang salah telah menyebabkan hampir separuh daripada jumlah kilang-kilang minyak sawit di Malaysia mempunyai kadar perahan minyak (OER) kurang daripada 20%. Projek ini membentangkan kit pengesanan kematangan buah kelapa sawit yang mengesan kematangan FFB sebelum penuaian dan memberi keputusan penuaian. Projek ini mencatat kemajuan dalam lima peringkat, iaitu pemerolehan imej-imej sample FFB kelapa sawit, pembangunan algoritma pemprosesan imej, pembangunan kaedah pengekstrakan ciri imej, pembangunan rangkaian neural tiruan (ANN) pengelas kematangan dan akhirnya pelaksanaan dalam perkakasan. Kit ini menggunakan papan kawalan NI sbRIO-9632XT, satu kamera protokol internet, dan sebuah LCD modul paparan. Sistem ini berfungsi seperti berikut: (1) memperoleh imej FFB sawit; (2) melaksanakan segmentasi imej; (3) mengekstrak nilai warna dari imej; (4) mengklasifikasi kematangan FFB menggunakan rangkaian neural “multilayer perceptron” (MLP) dan (5) memaparkan tahap kematangan dan keputusan penuaian pada panel LCD. Penyelidikan untuk kaedah pengekstrakan ciri imej dan model rangkaian neural MLP yang paling sesuai telah dijalankan. Ketepatan klasifikasi kematangan yang dicapai ialah 80.88% dan ketepatan untuk keputusan penuaian ialah 86.76%. Kit pengesanan ini mewujudkan satu piawaian yang objektif dalam klasifikasi kematangan FFB sawit dan dapat memberi keputusan penuaian yang boleh dipercayai kepada para penuai. _______________________________________________________________________________________________________ Ripeness classification of oil palm fresh fruit bunches (FFB) during harvesting is important to ensure that they are harvested during the optimum stage for maximum oil production. Currently, harvesting oil palm FFB of wrong ripeness stage has caused almost half of the total palm oil mills in Malaysia had oil extraction rate (OER) less than 20%. This project presents an oil palm fruit ripeness detection kit that detects the fruit ripeness prior harvesting and provides a reliable harvesting decision. The project is progressed in five stages, which are acquisition of oil palm FFB sample images, development of image processing algorithm, development of feature extraction method, development of artificial neural network (ANN) ripeness classifier, and finally hardware implementation. The embedded kit utilizes the National Instruments sbRIO-9632XT controller board, an internet protocol camera, and a LCD touch panel display module. The system works in the following way: (1) acquire oil palm FFB image; (2) perform fruit image segmentation; (3) extract hue values from processed image; (4) classify the fruit ripeness using multilayer perceptron (MLP) neural network; and (5) display the fruit ripeness and harvesting decision on the LCD panel. Investigation on the best feature extraction method and MLP neural network model were carried out. The ripeness classification accuracy achieved for the final design is 80.88% and its harvesting decision accuracy is 86.76%. This oil palm fruit ripeness detection kit aims to replace the conventional human grading of oil palm FFB as it creates an objective standard in classifying the FFB ripeness and provides a reliable harvesting decision to the harvester.
Contributor(s):
Heng Peng Zhi - Author
Primary Item Type:
Final Year Project
Identifiers:
Accession Number : 875005137
Language:
English
Subject Keywords:
Ripeness; (FFB); oil
First presented to the public:
6/1/2013
Original Publication Date:
11/14/2019
Previously Published By:
Universiti Sains Malaysia
Place Of Publication:
School of Electrical & Electronic Engineering
Citation:
Extents:
Number of Pages - 122
License Grantor / Date Granted:
  / ( View License )
Date Deposited
2019-12-02 16:31:26.573
Submitter:
Mohd Jasnizam Mohd Salleh

All Versions

Thumbnail Name Version Created Date
Oil palm fruit ripeness detection kit for harvesting decision1 2019-12-02 16:31:26.573