Important shape features for Romanian medicinal herb identification based on leaf image

Abstract

Herbal medicines can be used in treating health conditions under qualified medical observation. Various advantages are perceived by many consumers as being associated with using herbal medicines compared to conventional pharmaceutical products, such as the reduced risk of side effects, effectiveness with chronic conditions, lower cost and large availability, although robust evidence supporting such perceptions is most often not available. When the plant collection is done by persons without appropriate training, the risks of misidentification may be serious, even fatal. There are launched several applications for plant identification from photographs and a large number of papers dedicated to plant recognition using leaves images. There are mobile applications now which can be used for plant recognition and we want to create one that can be able to recognize the Romanian medicinal plants, allowing the user a safe collection. We present in this paper the current methods for plant identification using leaf image, as well as the parameters which are used for the classification. We continue the study with the feasibility of Romanian medicinal plant recognition from leaf images and perform a feature selection procedure on the parameters proposed in literature.