Leaf Venation Enhancing for Texture Feature Extraction in a Plant Classification Task

Abstract

In a computer science approach, the plant classification task focuses on the extraction of many leaf attributes, such as texture or veins, which are closely related and commonly analyzed together. Thereby, this study proposes a method to enhance the venation patterns over the leaf area, in order to improve the texture feature extraction in windows areas in the plant species identification. Regarding the experimentation, two types of texture features are contrasted, and it is performed over an own dataset with high-resolution image of 10 plant species. The obtained results demonstrate that the veins enhancing process improve the species classification task significantly for a texture descriptor based on the analysis of relation between neighboring pixels.