PERBANDINGAN EKSTRAKSI TEKSTUR CITRA UNTUK PEMILIHAN BENIH KEDELAI DENGAN METODE STATISTIK ORDE I DAN STATISTIK ORDE II

Abstract

The problem in determining the selection of soybean seeds for replanting, especially in East Nusa Tenggara is still an important issue. The thing that affects the quality of soybean seeds is found broken seeds, dull seeds, dirty seeds, and broken seeds due to the process of drying and shelling. Determination of soy bean quality is usually done manually by visual observation. The manual system takes a long time and produces products with inconsistent quality due to visual limitations, fatigue, and different perceptions of each observer. This research was conducted using comparison of image texture extraction with statistical methods of order I (color moment) and order II statistics (GLCM) for soy bean selection. Order I statistics (color moment) show the probability of the appearance of the value of the gray degree of pixels in an image, while the order II statistics (GLCM) show the probability of a neighborhood relationship between two pixels that form a cohesion matrix from the image data. This research is expected to help the classification process in determining soybean seeds. The k-Nearest Neighbor (k-NN) algorithm used in previous studies to classify the image objects to be examined. The results of this study were successfully conducted using k-Nearest Neighbor (k-NN) with euclidean distance and k = 1 with the results of color moment extracts getting the highest accuracy of 88% and the results of GLCM feature extraction for homogeneity characteristics of 75.5%, correlations of 78.67% , contrast is 65.75% and energy is 63.83% with an average accuracy of 70.93%.