CD Skripsi
Karakterisasi Sistem Pencitraan Multispektral Menggunakan Filter Optik Pada Tandan Buah Segar Kelapa Sawit
ABSTRACT
Sorting and grading of oil palm fresh fruit bunches (FFBs) is essential in producing quality Crude Palm Oil (CPO). Multispectral imaging systems are a fast, non-destructive alternative to sorting because they use fewer wavelengths. This research aims to classify the maturity of oil palm FFBs using a multispectral imaging system. The multispectral imaging system consists of a filter wheel that has 8 wavelengths namely 520, 680, 710, 740, 770, 800, 830 and 880 nm. The oil palms used consist of dura and tenera oil palms with 3 maturity levels, namely unripe, ripe, and overripe. The maturity level of oil palm FFB is known based on the hardness value of oil palm. The data acquisition and data processing used a Python-based computer program. The results showed that the reflectance intensity of each type and maturity of oil palm has a significant difference in pattern. Ripe palms have a higher reflectance intensity compared to unripe and overripe palms. The average hardness level of oil palm FFB shows that the level of ripeness is at the highest point and overripe is at the lowest point. PCA (Principal Component Analysis) analysis on multispectral image data is used to facilitate data visualization. The initial data consisting of 8 wavelength variables were converted into 2 variables, namely PC1 and PC2, which represented 90.41% of all analyzed image data grouped in a score plot. The score plot on the research data shows the grouping results of each maturity level of oil palm FFB.
Keywords: Fresh fruit bunches (FFBs), multispectral imaging, oil palm, PCA (Principal Component Analysis), Python, sorting and grading.
Tidak tersedia versi lain