Ventral side of a Leaf Image: Another Alternative for Leaf Image Classification

The plants have been used by human beings since ages for satisfying his general needs for food, shelter and medicinal values. Many of the plant species are on the verge of extinction, therefore, in order to preserve them for future, naming convention has a role to play. Many of the plant species are yet unknown and due to climatic changes and human factors, the plant species are dying out. Therefore, it is a dire necessity to know these unknown species in a better and a faster way. Leaf is an ornament of the plant and has a crucial role in classifying the plant on its own basis. The objective of this study is to find an alternative for the dorsal leaf image classification. As of now, only the dorsal side of the leaf image is considered for this purpose. This study proposes to utilize the texture features available on the ventral side of the leaf image for classification purpose using Gabor based texture features. The discrimination of the leaf images have been performed through classification algorithms: K-nearest neighbor, J48, Classification and Regression Tree and Random Forest. The results show better classification accuracy rates for ventral side of the leaf image over the dorsal side.