Extraction of leaves from herbarium images

Detection of a leaf from the image that contains the leaf, branches, and other background material is challenging. The existing approaches for automated leaf extraction do not provide satisfactory results when the end-users provide a plant image. For such an application to be feasible, the automated leaf extraction algorithm should handle leaves of various colors, shapes, sizes, and locations within the image. This paper presents an algorithm for automating the process of extracting the possible target leaves from herbarium plant images. Our results indicate high level of accuracy for the proposed algorithm.