Texture Histogram Features for Tea Leaf Identification using Visible Digital Camera
Alma and Bambang Heru Iswanto

Complex Systems Research Group, Department of Physics, FMIPA, Universitas Negeri Jakarta, Jl. Rawamangun Muka, Jakarta 13220, Indonesia

Abstract

This paper presents our study on the statistical texture histogram features to identify fresh tea leaves using a visible digital camera. For this purpose, the tea leaves were shooted every three days using the camera with 8 different orientations, a multiple of 45 degrees. Features of the images were extracted using the method to collect the feature dataset. Therefore, Principal Component Analysis (PCA), LBGU-EM clustering method, and Fisher’s Linear Discriminant Analysis (LDA) were applied to analyze the tea leaves based on the dataset. Experimental results using 320 image samples of four different categories show that the proposed method generated the image features that can significantly distinguish the fresh tea leaves categories.

Keywords: feature selection, histogram, principal component analysis, classification, tea

Topic: Computer Science