Rancang Bangun Hyperspectral Imaging System Hemat Biaya Berbasis LED

Azry, Muhammad Ashobta (2021) Rancang Bangun Hyperspectral Imaging System Hemat Biaya Berbasis LED. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 02311640000107-Undergraduate_Thesis.pdf] Text
02311640000107-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (1MB) | Request a copy

Abstract

Secara komersial, kamera hyperspectral imaging dijual dengan harga mencapai USD 1000 bahkan lebih. Padahal di era ini, hyperspectral imaging menjadi metode yang populer untuk aplikasi dalam berbagai bidang, termasuk agrikultur. Indonesia sendiri memiliki jeruk sebagai salah satu produk agrikultur dengan produksi mencapai 2 juta ton. Dalam Tugas Akhir ini dirancang hyperspectral imaging system dengan biaya rendah dengan aplikasi pada buah jeruk. Sistem yang dirancang terdiri atas LED putih, sebuah kamera smartphone, dan rekonstruksi spektral menggunakan CNN HRNet. Hasil citra memiliki panjang gelombang 400-700 nm dengan resolusi 10 nm. Ukuran citra keluaran sebesar 482x512 dengan rata-rata waktu rekonstruksi 1-2 menit per citra. Rata-rata RMSE spektral dari citra keluaran berkisar 0,0131. Hasil sampel jeruk dapat terklasifikasi dengan mudah pada panjang gelombang 540 nm di mana sampel menunjukkan penurunan intensitas dari hari ke-0 sampai ke-6.
=====================================================================================================
Commercially, the price of hyperspectral imaging camera is around USD 1000 and even more. Hyperspectral imaging is a popular method of imaging in many fields, including in agriculture. In Indonesia, one of its main agricultural products is orange, produced 2 million ton of oranges every year. This research tried to pursue a cheaper way to design hyperspectral imaging system, using orange as sample object. The final design consists of simple white LED lighting, a smartphone camera, and image reconstruction with CNN HRNet. The band spectral ranges from 400 nm to 700 nm with 10 nm resolution. The size of output image is 482x512, with total reconstruction time between 1-2 minutes each image. The average of RMSE is 0.0131. Orange samples show that on 540 nm wavelength, we can easily classify the maturity of orange samples in which samples shown a lowering light reflectance intensity from day 0 until day 6.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: hyperspectral imaging, jeruk, oranges, LED
Subjects: T Technology > T Technology (General)
Divisions: Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User: Muhammad Ashobta Azry
Date Deposited: 08 Sep 2021 12:22
Last Modified: 08 Sep 2021 12:22
URI: http://repository.its.ac.id/id/eprint/91835

Actions (login required)

View Item View Item