PENDUGAAN TINGKAT KUALITAS FISIK BIJI KACANG TANAH (Arachis hypogaea L.) MENGGUNAKAN PENGOLAHAN CITRA DIGITAL DAN METODE JARINGAN SARAF TIRUAN

• Main Authors: , FAJARINA SETIANING B, , Dr. Atris Suyantohadi, STP, M.T
• Format: Theses and Dissertations NonPeerReviewed
• Published: [Yogyakarta] : Universitas Gadjah Mada 2013
• Subjects: ETD
• Online Access: https://repository.ugm.ac.id/126724/; http://etd.ugm.ac.id/index.php?mod=penelitian_detail&sub=PenelitianDetail&act=view&typ=html&buku_id=66954
• Institution: Universitas Gadjah Mada

Peanut is one of the important commodities due to its high economic value, especially as foodstuff. Before processed, peanut should have a good physical quality seed with appropriate quality standards that has been set. Peanut�s quality classification is still conducted manually by comparing the proportion of its type based on visual observation. This method has a weakness that the result will be qualitative and subjective. Thus the stage classification can be inconsistent. Therefore, it is necessary to design a classification system to determine peanut�s physical quality which is quantitative and objective at the operator level, one of which is using image processing method and artificial neural networks. This study aims to conduct a series of experiments on samples of peanut seeds of which their quality level will later be classified based on Indonesian National Standard (SNI) 01-3921-1995. The stage of peanut image processing begins with a black box which is equipped with a webcam Genius I Slim 2020AF. Peanut�s image consist of normal, wrinkled, cracked, damaged, and other color type. Image processing is performed to analyze the color and texture of the image that will be used as the physical parameters of quality classification. Color and texture parameters used are red, green, blue, entropy, contrast, homogenity, and energy. Architecture of artificial neural network using feedforward backpropagation algorithm. Next there should be an identification of sample types using artificial neural network with 240 training samples and 60 testing samples. Physical quality classification based on the proportion of each type of peanut is conducted using 204 trial validation samples of peanut seeds. Image processing stage shows that it�s only color parameter that is used to determine peanut�s physical quality classification. The result of the study on trial validation samples displayed in the GUI (Graphical User Interface) shows that samples data classified into physical quality III with 91,67% accuracy rate with the proportion of peanut�s normal type 53,43%, wrinkled type 21,57%, cracked type 15,69%, damaged type 5,39%, and other color type 3,92%.