An ensemble classifier based leaf recognition approach for plant species classification using leaf texture, morphology and shape

Plant recognition is a main problem for biologists, environmentalists and chemists. Human experts of these fields perform plant recognition manually, which requires more time and is less efficient. Plant recognition systems are used to classify plants into appropriate taxonomies. Such information is useful for botanists, industrialists, food engineers and physicians. Botanists use morphological features of the leaves to identify them. These features are used in terms of automation in identifying the plants. Leaf images of different plants have different characteristics which help in the classification of these species. The proposed approach identifies plant species in three distinct phases: (1) preprocessing. (2) feature extraction and (3) classification. Leaf features like texture, morphology and Zernike moment are extracted and treated as input vector to the four different clasifiers. The best accuracy achieved by a combination of textural and morphological features is 87 percent.(author)