Ripeness Level Classification of Oil Palm Fresh Fruit Bunch Using Laser Induced Fluorescence Imaging

Abstract

Tingkat Kematangan Tandan Buah Segar (TBS) kelapa Sawit merupakan faktor penentu kualitas crude palm oil (CPO) yang dihasilkan pabrik kelapa sawit. Metode penyortiran TBS setelah panen atau sebelum memasuki proses perebusan pada umumnya dilakukan secara manual mengandalkan penglihatan dan pengalaman. Metode ini rentan kesalahan dan bersifat subyektif. Metode pencitraan berkembang sangat cepat karena kemajuan dalam bidang komputer dan teknik pengolahan citra, khususnya untuk sistem sortasi dan grading. Penelitian ini mengunakan metode pencitraan fluoresensi yang diinduksi laser untuk mengakses dan mengklasifikasi tingkat kematangan TBS kelapa sawit. Hubungan antara tingkat keabuan dan tingkat kekerasan buah TBS dianalisa. Sampel terdiri dari 27 TBS kelapa sawit varietas Tenera. Tingkat kematangan dikategorikan oleh pemanen berpengalaman menjadi mentah, matang, dan lewat matang. Tiga bagian TBS yaitu pangkal, tengah, dan ujung disinari laser dioda 640 nm mengenai 5 buah pada tiap bagian. Kemudian citra direkam mengunakan kamera CMOS monokrom. Selanjutnya 15 buah tersebut diuji tingkat kekerasan mengunakan penetrometer. Klasifikasi tingkat kematangan dilakukan mengunakan K-mean clustering. Hasil penelitian memperlihatkan bahwa metode pencitraan fluoresensi yang diinduksi laser potensial digunakan dalam mengklasifikasi tingkat kematangan TBS. Tingkat kekerasan buah berkorelasi positif terhadap tingkat keabuan citra TBS. K-mean clustering memperlihatkan tiga kelompok tingkat kematangan yang terdiri dari 0, 1 dan 2. Ripeness levels of oil palm fresh fruit bunches (FFB) are the main factor to determine the quality of crude palm oil (CPO) produced by Oil Palm Mill. Sorting oil palm FFB after harvest or before entering the boiling process is generally done manually which relies on human vision and experience. Imaging methods has developed vastly due to advances in computer and image processing techniques. This study used a laser-induced fluorescence imaging to access and classify the ripeness levels of oil palm FFB of Tenera variety. The relationship between gray value and the level of firmness of FFB fruit was analyzed. The samples consisted of 27 oil palm FFB categorized  by experienced harvester as unripe, ripe, and overripe. Laser light was shone on equatorial part of each FFB such that 5 fruitlets were covered by laser light, then the image of the front part was acquire using a monochrome CMOS camera. The step was repeated for basil and apical parts in sequent. All 15 fruitlets were testing for the firmness level using a penetrometer. Ripeness level classification was done using K-mean clustering. The results showed that the laser-induced fluorescence imaging method are potential to be used to determine the ripeness levels of FFB. The fruit firmness is positively correlated with the gray value of the image of FFB. K-mean clustering shows three ripeness centroid of 0, 1 and 2 . Keyword: Fluorecence Imaging, Oil Palm, Fresh Fruit Bunches, Firmness, Laser Induced Fluorecence