CD Skripsi
Klasifikasi Sampah Plastik Menggunakan Pencitraan Hiperspektral
Plastic waste is a global problem in everyday life that still requires further processing to overcome it. The plastic waste sorting system in Indonesia is still done manually using experience personnel to separate plastic wastes based on type. This research aims to classify plastic waste according to type using hyperspectral imaging method. There were three types of plastic waste samples used, High-Density Polyethylene (HDPE), Polyethylene Terephthalate (PET), and Polypropylene (PP) with 20 samples for each type. The hyperspectral imaging system has advantages for classifying plastic waste because the wavelength spectrum used is from 400 nm – 1000 nm. Image processing was done using Matlab to obtain reflectance intensity for each samples. Principal Component Analysis (PCA) and K-Mean Clustering are used to classify each type of plastic waste. The PCA method uses two variables to analyze hyperspectral images, namely PC1 and PC2. The results showed significant differences in the reflectance intensity of hyperspectral images, especially at the wavelength of 870 nm, which increased in the three plastic waste samples. HDPE plastic has the highest reflectance intensity because it contains colored and thick samples. The PCA results have a cumulative percentage of 98,71%. Meanwhile, the k-mean clustering of three plastic wastes obtained an accuracy of 70.29%. Based on the pattern formed in the scatter plot of the two analyses, where there is a grouping of data for each type plastic waste samples, it is shown that hyperspectral imaging has potential to classify the plastic wastes.
Keywords : Hyperspectral Imaging, Plastic Waste, K-Mean Clustering, Principal Component Analysis (PCA), Reflectance Intensity
Tidak tersedia versi lain