Plant Identification by Image Processing of Leaf Veins

This study focuses on building a portable device capable of plant identification by image processing of leaf veins using Raspberry pi. The devise that the study will develop can help professionals in the field of Botany and Biology. It can be used by Pathologist, plant breeders and Doctors that specializes on medicinal plants. This study will concentrate on identifying the plant using its leaf veins. Its concentration is on simple leaf type. Twenty different kinds of plant leaves will be tested with 10 trials per leaf. Each leaf image will have its own labeled folder in the database that is created automatically after registration of leaf image. Series of algorithms were used for the leaf recognition method. Scale-Invariant Feature Transform (SIFT) extraction will be used to describe and detect the local features of the recognized leaf vein image. With the use of Support Vectors Machines (SVM), the matrix produced by the SIFT will be used to classify the correct plant to be shown on the Liquid Crystal Display (LCD) screen as the output containing the plant name, description and image. Initial results showed accuracy rate of 84.29% while the error rate was 15.71%.