IDENTIFIKASI CITRA DAUN TANAMAN JERUK DENGAN LOCAL BINARY PATTERN DAN MOMENT INVARIANT

Abstract

Citrus species identification can use from citrus leaf image. The research purposes build the identificate citrus species based on leaf texture and shape by using Local Binary Pattern as texture feature and Moment Invariant as shape feature, and Euclidean Distance as image distance measurement. The research data use citrus leaf image consist of Citrus aurantifolia, Citrus sinesis, Citrus hystrix, Citrus limon, Citrus maxima, Citrus amblycarpa, and Citrus microcarpa. Based on experiment tests, we can conclude that (1) 100% accuracy for citrus leaf from a smartphone, (2) 100% accuracy for citrus leaf from a smartphone with red background, (3) 85,71% accuracy for citrus leaf from a smartphone with green background, (4) 100% accuracy for citrus leaf from a smartphone with blue background, (5) 85,71% accuracy for citrus leaf from a smartphone with black background, (6) 85,71% for images from the internet.