Identifikasi Varietas Berdasarkan Warna dan Tekstur Permukaan Beras Menggunakan Pengolahan Citra Digital dan Jaringan Syaraf Tiruan
Adnan, Adnan; Balai Pengkajian Teknologi Pertanian Papua Jl. Pembangunan-Pertanian, Jati-jati, Merauke, Papua
Suhartini, Suhartini; Balai Besar Penelitian Tanaman Padi Jl. Raya 9 Sukamandi, Subang 41256
Kusbiantoro, Bram; Balai Besar Penelitian Tanaman Padi Jl. Raya 9 Sukamandi, Subang 41256
MetadataShow full item record
Research on milled rice color and texture using digital image processing technique is becoming important, due to its potential use as a basic method for rice derived variety identification. The research was aimed to identify milled rice of varieties Basmati, Inpari 1, and Sintanur using color and texture analysis based on digital image processing. A combination of color and texture analysis was used as input parameter and then analyzed in the next step using the artificial neural network (ANN) to determine input parameter that has the highest accuracy level. The experiment was conducted at Merauke Experiment Station, Papua Institute for Agriculture Assessment Technology from May to July 2011. The materials used were milled rice of varieties Basmati 370, Inpari 1, and Sintanur that were previously grown at Sukamandi Experiment Station of the Indonesian Center for Rice Research (ICRR). All samples were qualified as grade 1 based on SNI 6128:2008. Fifty image samples were taken from each variety to get a total of 150 images to be analyzed for their colors and textures using the digital image processing. The color and texture data were analyzed using the analysis of variance (ANOVA) and further tested using the Duncan’s Multiple Range Test (DMRT) to obtain the median, while the data spread was analyzed using the boxplot method. The combination of color and texture as input parameters were analyzed using the ANN. One hundred and five rice data were used for training and 45 data were for testing. The results showed that the digital image processing and ANN recognized three output parameters in rice varieties of Basmati, Inpari 1, and Sintanur. Texture analysis with five input parameters were considered the best factor to be used in the ANN model with 100% accuracy.