CD Tesis

Rancang Bangun Sistem Pencitraan Multispektral Berbasis Led Sebagai Sumber Cahaya Untuk Klasifikasi Tingkat Kematangan Tandan Buah Segar Kelapa Sawit

---

Design of an LED-Based
Multispectral Imaging System as a Light Source for Classification of Ripeness
Levels of Oil Palm Fresh Fruit Bunches, supervised by Minarni and Rahmondia
Nanda Setiadi.
The ripeness level of oil palm fresh fruit bunches (FFB) is an important
factor in determining the quality of palm oil. Quality and high-quality palm FFB
has a low Free Fatty Acid (FFA) content so that the oil content produced is quite
high. Ripe and good quality oil palm FFB has an FFA level of 20%. The process of harvesting and sorting also influences the level of
ripeness. Currently, the determination of the ripeness level of oil palm FFB is
done manually so it still has weaknesses, namely it is still subjective and slow.
This research aims to classify the ripeness level non-destructively and quickly
using multispectral imaging methods.
In this study, a multispectral imaging system (MSI) was built based on
LEDs as a light source to help classify the ripeness level of oil palm FFB in the
sorting process. This MSI system consists of a camera as a detector and an LED
as a light source. The use of LEDs apart from being a light source also serves to
replace the bandwidth filter as a wavelength filtering device because it has a
discrete wavelength. The wavelengths used are 8, 680 nm, 700 nm, 750 nm, 780
nm, 810 nm, 850 nm, 880 nm, and 900 nm, each of which has 4 wavelengths, so
that the circuit that is built uses 32 LEDs. The sample used was oil palm FFB with
2 ripeness levels which were determined by an experienced foreman during
harvest. The process of image acquisition and data processing and data analysis
uses the Python programming language.
The recorded multispectral images are processed to obtain the relative
reflectance intensity values for each wavelength. The relative reflectance intensity
of multispectral images has increased in the visible light range (680 nm) and
near-infrared (700 nm-900 nm) where the ripe category is higher than the raw
category, this is due to the interaction of matter at the molecular level in the
infrared spectral region. The process of validating the level of ripeness is carried
out using the destructive method by measuring the levels of oil and Free Fatty
Acids (ALB). Ripeness level classification is carried out using Principal
Component Analysis (PCA) and k-mean clustering, the PCA and k-mean
clustering methods can classify oil palm FFB into 2 ripeness levels with
percentages of 88% and 69% respectively, which can be used to represent the
number of images analyzed for each ripeness level.
Keywords: Multispectral Imaging, LED, Relative Reflectance Intensity, Oil and
Free Fatty Acid Content

---

Ketersediaan

Informasi Detail

Judul Seri

-

No. Panggil

191024788

Penerbit

Pekanbaru : Universitas Riau – Pascasarjana – Tesis Fisika., 2023

Deskripsi Fisik

-

Bahasa

Indonesia

ISBN/ISSN

-

Klasifikasi

191024788

Tipe Isi

-

Tipe Media

-

Tipe Pembawa

-

Edisi

-

Subjek

FISIKA

Info Detail Spesifik

-

Pernyataan Tanggungjawab

FATAH

Versi lain/terkait

Lampiran Berkas

Komentar

---

Anda harus masuk sebelum memberikan komentar