A unique morpho‐feature extraction algorithm for medicinal plant identification

Abstract

An image is a set of numbers arranged in matrix form. The image feature extraction algorithm converts the input image into different numerical forms to extract the useful information from the input image and the selection of appropriate feature extraction algorithm is crucial for medicinal plant identification. In medicinal plants, the leaves are an available important resource of morphological features. Botanists use these morphological features of leaf images for medicinal plant identification. The existing leaf‐based medicinal plant identification strategies include shape, colour and texture features. In these methods, environmental factors directly influence the features and hence, the impact can be observed in the accuracy of the result. To overcome these limitations, we have proposed a unique morpho‐feature extraction algorithm (UMFEA) for accurate identification of medicinal plants. The UMFEA includes three sub‐algorithms for shape, apex, base, and vein features extraction. The proposed UMFEA is tested over Flavia, Swedish, Leaf and our databases. The performance comparison of UMFEA is done on different databases and the results obtained were remarkably good.