Leaf Classification using Local Binary Pattern and Histogram of Oriented Gradients

Abstract

Categorization of plant species is a significant process in studying the diversity of different plant species in order to utilize it as medical treatment and to keep track of invasive plant species to maintain the balance of the environment. However, plants have extremely complex structure and diverse with millions of species around the world which makes the classification process extremely tedious. This paper introduces a method which utilizes the combination of Local Binary Pattern and Histogram Oriented Gradient as feature extractor for leaf classification which increases the accuracy during classification. Support Vector Machine was used as classifier of the leaf features. Two well-known datasets, Swedish Leaf Dataset and Flavia Dataset, were used to carry out the experimental studies. Our proposed method performed the best when compared to three other methods. © 2019 IEEE.