Analisa Kehandalan Ekstraksi Fitur Ordo Kedua terhadap Ordo Pertama untuk Identifikasi Ciri Berbasis Tekstur Warna

Abstract

The feature extraction method is done by first converting RGB images into grayscale images. Based on the grayscale image, two different processing methods are carried out, namely first order feature extraction and second order feature extraction. The extraction features of the first order used 5 characteristic parameters, namely Mean, Variance, Skewness, Kurtosis, and Entropy, while the extraction features of the second order used 6 characteristic parameters, namely Angular Second Moment, Contrast, Correlation, Standard Deviation, Inverse Difference Moment and Homogenity. The 11 characteristic parameters will then be classified using the LVQ artificial neural network method to find the final weight used as the reference weight for characteristics identification based on color texture. In this research, 30 image samples were used (15 image samples in category A and 15 image samples in category B) which were divided into 16 image samples for training and 14 image samples for testing. The results of the research analysis show that the second order feature extraction method is more reliable than the first order feature extraction method.